Search Results

Search found 25 results on 1 pages for 'joshreuben'.

Page 1/1 | 1 

  • Workflow 4.5 is Awesome, cant wait for 5.0!

    - by JoshReuben
    About 2 years ago I wrote a blog post describing what I would like to see in Workflow vnext: http://geekswithblogs.net/JoshReuben/archive/2010/08/25/workflow-4.0---not-there-yet.aspx At the time WF 4.0 was a little rough around the edges – the State Machine was on codeplex and people were simulating state machines with Flowcharts. Last year I built a near- realtime machine management system using WF 4.0.1 – its managing the internal operations of this device: http://landanano.com/products/commercial   Well WF 4.5 has come a long way – many of my gripes have been addressed: C# expressions - no more VB 'AndAlso' clauses state machine awesomeness - can query current state many designer improvements - Document Outline is so much more succinct than Designer! Separate WCF Service Contract interfaces and ability to generate activities from contract operations ability to rehydrate to updated flow definitions via DynamicUpdateMap and WorkflowIdentity you can read about the new features here: http://msdn.microsoft.com/en-us/library/hh305677(VS.110).aspx   2013 could be the year of Workflow evangelism for .NET, as it comes together as the DSL language. Eg on Azure it could be used to graphically orchestrate between WebRoles, WorkerRoles and AppFabric Queues and the ServiceBus – that would be grand.   Here’s a list of things I’d like to see in Workflow 5.0: Stronger Parallelism support for true multithreaded workflows . A Workflow executes on a single thread – wouldn’t it be great if we had the ability to model TPL DataFlow? Parallel is not really parallel, just allows AsyncCodeActivity.     support for recursion an ExpressionTree activity with an editor design surface a math activity pack return of application level protocol (3.51 WF services) – automatically expose a state machine as a WCF service with bookmark Receive activities generated from OperationContract automatically placed in state transition triggers. A new HTML5 ActivityDesigner control – support with different CSS3  skinnable hooks,  remote connectivity (had to roll my own) A data flow view – crucial to understanding the big picture Ability to refactor a Sequence to custom activity in a separate .xaml file – like Expression Blend does for UserControl state machine global error handling - if all states goto an error state, you quickly get visual spagetti. Now you could nest a state machine, but what if you want an application level protocol whereby each state exposes certain WCF ops. DSL RAD editing - Make the Document Outline into a DSL editor for adding activities  – For WF to really succeed as a higher level of abstraction, It needs to be more productive than raw coding - drag & drop on the designer is currently too slow compared to just typing code. Extensible Wizard API - for pluggable WF editor experience other execution models beyond Sequence, Flowchart & StateMachine: SSIS, Behavior Trees,  Wolfram Model tool – surprise us! improvements to Designer debugging API - SourceLocation is tied to XAML file line number and char position, and ModelService access seems convoluted - why not leverage WPF LogicalTreeHelper / VisualTreeHelper ? Workflow Team , keep on rocking!

    Read the article

  • A Taxonomy of Numerical Methods v1

    - by JoshReuben
    Numerical Analysis – When, What, (but not how) Once you understand the Math & know C++, Numerical Methods are basically blocks of iterative & conditional math code. I found the real trick was seeing the forest for the trees – knowing which method to use for which situation. Its pretty easy to get lost in the details – so I’ve tried to organize these methods in a way that I can quickly look this up. I’ve included links to detailed explanations and to C++ code examples. I’ve tried to classify Numerical methods in the following broad categories: Solving Systems of Linear Equations Solving Non-Linear Equations Iteratively Interpolation Curve Fitting Optimization Numerical Differentiation & Integration Solving ODEs Boundary Problems Solving EigenValue problems Enjoy – I did ! Solving Systems of Linear Equations Overview Solve sets of algebraic equations with x unknowns The set is commonly in matrix form Gauss-Jordan Elimination http://en.wikipedia.org/wiki/Gauss%E2%80%93Jordan_elimination C++: http://www.codekeep.net/snippets/623f1923-e03c-4636-8c92-c9dc7aa0d3c0.aspx Produces solution of the equations & the coefficient matrix Efficient, stable 2 steps: · Forward Elimination – matrix decomposition: reduce set to triangular form (0s below the diagonal) or row echelon form. If degenerate, then there is no solution · Backward Elimination –write the original matrix as the product of ints inverse matrix & its reduced row-echelon matrix à reduce set to row canonical form & use back-substitution to find the solution to the set Elementary ops for matrix decomposition: · Row multiplication · Row switching · Add multiples of rows to other rows Use pivoting to ensure rows are ordered for achieving triangular form LU Decomposition http://en.wikipedia.org/wiki/LU_decomposition C++: http://ganeshtiwaridotcomdotnp.blogspot.co.il/2009/12/c-c-code-lu-decomposition-for-solving.html Represent the matrix as a product of lower & upper triangular matrices A modified version of GJ Elimination Advantage – can easily apply forward & backward elimination to solve triangular matrices Techniques: · Doolittle Method – sets the L matrix diagonal to unity · Crout Method - sets the U matrix diagonal to unity Note: both the L & U matrices share the same unity diagonal & can be stored compactly in the same matrix Gauss-Seidel Iteration http://en.wikipedia.org/wiki/Gauss%E2%80%93Seidel_method C++: http://www.nr.com/forum/showthread.php?t=722 Transform the linear set of equations into a single equation & then use numerical integration (as integration formulas have Sums, it is implemented iteratively). an optimization of Gauss-Jacobi: 1.5 times faster, requires 0.25 iterations to achieve the same tolerance Solving Non-Linear Equations Iteratively find roots of polynomials – there may be 0, 1 or n solutions for an n order polynomial use iterative techniques Iterative methods · used when there are no known analytical techniques · Requires set functions to be continuous & differentiable · Requires an initial seed value – choice is critical to convergence à conduct multiple runs with different starting points & then select best result · Systematic - iterate until diminishing returns, tolerance or max iteration conditions are met · bracketing techniques will always yield convergent solutions, non-bracketing methods may fail to converge Incremental method if a nonlinear function has opposite signs at 2 ends of a small interval x1 & x2, then there is likely to be a solution in their interval – solutions are detected by evaluating a function over interval steps, for a change in sign, adjusting the step size dynamically. Limitations – can miss closely spaced solutions in large intervals, cannot detect degenerate (coinciding) solutions, limited to functions that cross the x-axis, gives false positives for singularities Fixed point method http://en.wikipedia.org/wiki/Fixed-point_iteration C++: http://books.google.co.il/books?id=weYj75E_t6MC&pg=PA79&lpg=PA79&dq=fixed+point+method++c%2B%2B&source=bl&ots=LQ-5P_taoC&sig=lENUUIYBK53tZtTwNfHLy5PEWDk&hl=en&sa=X&ei=wezDUPW1J5DptQaMsIHQCw&redir_esc=y#v=onepage&q=fixed%20point%20method%20%20c%2B%2B&f=false Algebraically rearrange a solution to isolate a variable then apply incremental method Bisection method http://en.wikipedia.org/wiki/Bisection_method C++: http://numericalcomputing.wordpress.com/category/algorithms/ Bracketed - Select an initial interval, keep bisecting it ad midpoint into sub-intervals and then apply incremental method on smaller & smaller intervals – zoom in Adv: unaffected by function gradient à reliable Disadv: slow convergence False Position Method http://en.wikipedia.org/wiki/False_position_method C++: http://www.dreamincode.net/forums/topic/126100-bisection-and-false-position-methods/ Bracketed - Select an initial interval , & use the relative value of function at interval end points to select next sub-intervals (estimate how far between the end points the solution might be & subdivide based on this) Newton-Raphson method http://en.wikipedia.org/wiki/Newton's_method C++: http://www-users.cselabs.umn.edu/classes/Summer-2012/csci1113/index.php?page=./newt3 Also known as Newton's method Convenient, efficient Not bracketed – only a single initial guess is required to start iteration – requires an analytical expression for the first derivative of the function as input. Evaluates the function & its derivative at each step. Can be extended to the Newton MutiRoot method for solving multiple roots Can be easily applied to an of n-coupled set of non-linear equations – conduct a Taylor Series expansion of a function, dropping terms of order n, rewrite as a Jacobian matrix of PDs & convert to simultaneous linear equations !!! Secant Method http://en.wikipedia.org/wiki/Secant_method C++: http://forum.vcoderz.com/showthread.php?p=205230 Unlike N-R, can estimate first derivative from an initial interval (does not require root to be bracketed) instead of inputting it Since derivative is approximated, may converge slower. Is fast in practice as it does not have to evaluate the derivative at each step. Similar implementation to False Positive method Birge-Vieta Method http://mat.iitm.ac.in/home/sryedida/public_html/caimna/transcendental/polynomial%20methods/bv%20method.html C++: http://books.google.co.il/books?id=cL1boM2uyQwC&pg=SA3-PA51&lpg=SA3-PA51&dq=Birge-Vieta+Method+c%2B%2B&source=bl&ots=QZmnDTK3rC&sig=BPNcHHbpR_DKVoZXrLi4nVXD-gg&hl=en&sa=X&ei=R-_DUK2iNIjzsgbE5ID4Dg&redir_esc=y#v=onepage&q=Birge-Vieta%20Method%20c%2B%2B&f=false combines Horner's method of polynomial evaluation (transforming into lesser degree polynomials that are more computationally efficient to process) with Newton-Raphson to provide a computational speed-up Interpolation Overview Construct new data points for as close as possible fit within range of a discrete set of known points (that were obtained via sampling, experimentation) Use Taylor Series Expansion of a function f(x) around a specific value for x Linear Interpolation http://en.wikipedia.org/wiki/Linear_interpolation C++: http://www.hamaluik.com/?p=289 Straight line between 2 points à concatenate interpolants between each pair of data points Bilinear Interpolation http://en.wikipedia.org/wiki/Bilinear_interpolation C++: http://supercomputingblog.com/graphics/coding-bilinear-interpolation/2/ Extension of the linear function for interpolating functions of 2 variables – perform linear interpolation first in 1 direction, then in another. Used in image processing – e.g. texture mapping filter. Uses 4 vertices to interpolate a value within a unit cell. Lagrange Interpolation http://en.wikipedia.org/wiki/Lagrange_polynomial C++: http://www.codecogs.com/code/maths/approximation/interpolation/lagrange.php For polynomials Requires recomputation for all terms for each distinct x value – can only be applied for small number of nodes Numerically unstable Barycentric Interpolation http://epubs.siam.org/doi/pdf/10.1137/S0036144502417715 C++: http://www.gamedev.net/topic/621445-barycentric-coordinates-c-code-check/ Rearrange the terms in the equation of the Legrange interpolation by defining weight functions that are independent of the interpolated value of x Newton Divided Difference Interpolation http://en.wikipedia.org/wiki/Newton_polynomial C++: http://jee-appy.blogspot.co.il/2011/12/newton-divided-difference-interpolation.html Hermite Divided Differences: Interpolation polynomial approximation for a given set of data points in the NR form - divided differences are used to approximately calculate the various differences. For a given set of 3 data points , fit a quadratic interpolant through the data Bracketed functions allow Newton divided differences to be calculated recursively Difference table Cubic Spline Interpolation http://en.wikipedia.org/wiki/Spline_interpolation C++: https://www.marcusbannerman.co.uk/index.php/home/latestarticles/42-articles/96-cubic-spline-class.html Spline is a piecewise polynomial Provides smoothness – for interpolations with significantly varying data Use weighted coefficients to bend the function to be smooth & its 1st & 2nd derivatives are continuous through the edge points in the interval Curve Fitting A generalization of interpolating whereby given data points may contain noise à the curve does not necessarily pass through all the points Least Squares Fit http://en.wikipedia.org/wiki/Least_squares C++: http://www.ccas.ru/mmes/educat/lab04k/02/least-squares.c Residual – difference between observed value & expected value Model function is often chosen as a linear combination of the specified functions Determines: A) The model instance in which the sum of squared residuals has the least value B) param values for which model best fits data Straight Line Fit Linear correlation between independent variable and dependent variable Linear Regression http://en.wikipedia.org/wiki/Linear_regression C++: http://www.oocities.org/david_swaim/cpp/linregc.htm Special case of statistically exact extrapolation Leverage least squares Given a basis function, the sum of the residuals is determined and the corresponding gradient equation is expressed as a set of normal linear equations in matrix form that can be solved (e.g. using LU Decomposition) Can be weighted - Drop the assumption that all errors have the same significance –-> confidence of accuracy is different for each data point. Fit the function closer to points with higher weights Polynomial Fit - use a polynomial basis function Moving Average http://en.wikipedia.org/wiki/Moving_average C++: http://www.codeproject.com/Articles/17860/A-Simple-Moving-Average-Algorithm Used for smoothing (cancel fluctuations to highlight longer-term trends & cycles), time series data analysis, signal processing filters Replace each data point with average of neighbors. Can be simple (SMA), weighted (WMA), exponential (EMA). Lags behind latest data points – extra weight can be given to more recent data points. Weights can decrease arithmetically or exponentially according to distance from point. Parameters: smoothing factor, period, weight basis Optimization Overview Given function with multiple variables, find Min (or max by minimizing –f(x)) Iterative approach Efficient, but not necessarily reliable Conditions: noisy data, constraints, non-linear models Detection via sign of first derivative - Derivative of saddle points will be 0 Local minima Bisection method Similar method for finding a root for a non-linear equation Start with an interval that contains a minimum Golden Search method http://en.wikipedia.org/wiki/Golden_section_search C++: http://www.codecogs.com/code/maths/optimization/golden.php Bisect intervals according to golden ratio 0.618.. Achieves reduction by evaluating a single function instead of 2 Newton-Raphson Method Brent method http://en.wikipedia.org/wiki/Brent's_method C++: http://people.sc.fsu.edu/~jburkardt/cpp_src/brent/brent.cpp Based on quadratic or parabolic interpolation – if the function is smooth & parabolic near to the minimum, then a parabola fitted through any 3 points should approximate the minima – fails when the 3 points are collinear , in which case the denominator is 0 Simplex Method http://en.wikipedia.org/wiki/Simplex_algorithm C++: http://www.codeguru.com/cpp/article.php/c17505/Simplex-Optimization-Algorithm-and-Implemetation-in-C-Programming.htm Find the global minima of any multi-variable function Direct search – no derivatives required At each step it maintains a non-degenerative simplex – a convex hull of n+1 vertices. Obtains the minimum for a function with n variables by evaluating the function at n-1 points, iteratively replacing the point of worst result with the point of best result, shrinking the multidimensional simplex around the best point. Point replacement involves expanding & contracting the simplex near the worst value point to determine a better replacement point Oscillation can be avoided by choosing the 2nd worst result Restart if it gets stuck Parameters: contraction & expansion factors Simulated Annealing http://en.wikipedia.org/wiki/Simulated_annealing C++: http://code.google.com/p/cppsimulatedannealing/ Analogy to heating & cooling metal to strengthen its structure Stochastic method – apply random permutation search for global minima - Avoid entrapment in local minima via hill climbing Heating schedule - Annealing schedule params: temperature, iterations at each temp, temperature delta Cooling schedule – can be linear, step-wise or exponential Differential Evolution http://en.wikipedia.org/wiki/Differential_evolution C++: http://www.amichel.com/de/doc/html/ More advanced stochastic methods analogous to biological processes: Genetic algorithms, evolution strategies Parallel direct search method against multiple discrete or continuous variables Initial population of variable vectors chosen randomly – if weighted difference vector of 2 vectors yields a lower objective function value then it replaces the comparison vector Many params: #parents, #variables, step size, crossover constant etc Convergence is slow – many more function evaluations than simulated annealing Numerical Differentiation Overview 2 approaches to finite difference methods: · A) approximate function via polynomial interpolation then differentiate · B) Taylor series approximation – additionally provides error estimate Finite Difference methods http://en.wikipedia.org/wiki/Finite_difference_method C++: http://www.wpi.edu/Pubs/ETD/Available/etd-051807-164436/unrestricted/EAMPADU.pdf Find differences between high order derivative values - Approximate differential equations by finite differences at evenly spaced data points Based on forward & backward Taylor series expansion of f(x) about x plus or minus multiples of delta h. Forward / backward difference - the sums of the series contains even derivatives and the difference of the series contains odd derivatives – coupled equations that can be solved. Provide an approximation of the derivative within a O(h^2) accuracy There is also central difference & extended central difference which has a O(h^4) accuracy Richardson Extrapolation http://en.wikipedia.org/wiki/Richardson_extrapolation C++: http://mathscoding.blogspot.co.il/2012/02/introduction-richardson-extrapolation.html A sequence acceleration method applied to finite differences Fast convergence, high accuracy O(h^4) Derivatives via Interpolation Cannot apply Finite Difference method to discrete data points at uneven intervals – so need to approximate the derivative of f(x) using the derivative of the interpolant via 3 point Lagrange Interpolation Note: the higher the order of the derivative, the lower the approximation precision Numerical Integration Estimate finite & infinite integrals of functions More accurate procedure than numerical differentiation Use when it is not possible to obtain an integral of a function analytically or when the function is not given, only the data points are Newton Cotes Methods http://en.wikipedia.org/wiki/Newton%E2%80%93Cotes_formulas C++: http://www.siafoo.net/snippet/324 For equally spaced data points Computationally easy – based on local interpolation of n rectangular strip areas that is piecewise fitted to a polynomial to get the sum total area Evaluate the integrand at n+1 evenly spaced points – approximate definite integral by Sum Weights are derived from Lagrange Basis polynomials Leverage Trapezoidal Rule for default 2nd formulas, Simpson 1/3 Rule for substituting 3 point formulas, Simpson 3/8 Rule for 4 point formulas. For 4 point formulas use Bodes Rule. Higher orders obtain more accurate results Trapezoidal Rule uses simple area, Simpsons Rule replaces the integrand f(x) with a quadratic polynomial p(x) that uses the same values as f(x) for its end points, but adds a midpoint Romberg Integration http://en.wikipedia.org/wiki/Romberg's_method C++: http://code.google.com/p/romberg-integration/downloads/detail?name=romberg.cpp&can=2&q= Combines trapezoidal rule with Richardson Extrapolation Evaluates the integrand at equally spaced points The integrand must have continuous derivatives Each R(n,m) extrapolation uses a higher order integrand polynomial replacement rule (zeroth starts with trapezoidal) à a lower triangular matrix set of equation coefficients where the bottom right term has the most accurate approximation. The process continues until the difference between 2 successive diagonal terms becomes sufficiently small. Gaussian Quadrature http://en.wikipedia.org/wiki/Gaussian_quadrature C++: http://www.alglib.net/integration/gaussianquadratures.php Data points are chosen to yield best possible accuracy – requires fewer evaluations Ability to handle singularities, functions that are difficult to evaluate The integrand can include a weighting function determined by a set of orthogonal polynomials. Points & weights are selected so that the integrand yields the exact integral if f(x) is a polynomial of degree <= 2n+1 Techniques (basically different weighting functions): · Gauss-Legendre Integration w(x)=1 · Gauss-Laguerre Integration w(x)=e^-x · Gauss-Hermite Integration w(x)=e^-x^2 · Gauss-Chebyshev Integration w(x)= 1 / Sqrt(1-x^2) Solving ODEs Use when high order differential equations cannot be solved analytically Evaluated under boundary conditions RK for systems – a high order differential equation can always be transformed into a coupled first order system of equations Euler method http://en.wikipedia.org/wiki/Euler_method C++: http://rosettacode.org/wiki/Euler_method First order Runge–Kutta method. Simple recursive method – given an initial value, calculate derivative deltas. Unstable & not very accurate (O(h) error) – not used in practice A first-order method - the local error (truncation error per step) is proportional to the square of the step size, and the global error (error at a given time) is proportional to the step size In evolving solution between data points xn & xn+1, only evaluates derivatives at beginning of interval xn à asymmetric at boundaries Higher order Runge Kutta http://en.wikipedia.org/wiki/Runge%E2%80%93Kutta_methods C++: http://www.dreamincode.net/code/snippet1441.htm 2nd & 4th order RK - Introduces parameterized midpoints for more symmetric solutions à accuracy at higher computational cost Adaptive RK – RK-Fehlberg – estimate the truncation at each integration step & automatically adjust the step size to keep error within prescribed limits. At each step 2 approximations are compared – if in disagreement to a specific accuracy, the step size is reduced Boundary Value Problems Where solution of differential equations are located at 2 different values of the independent variable x à more difficult, because cannot just start at point of initial value – there may not be enough starting conditions available at the end points to produce a unique solution An n-order equation will require n boundary conditions – need to determine the missing n-1 conditions which cause the given conditions at the other boundary to be satisfied Shooting Method http://en.wikipedia.org/wiki/Shooting_method C++: http://ganeshtiwaridotcomdotnp.blogspot.co.il/2009/12/c-c-code-shooting-method-for-solving.html Iteratively guess the missing values for one end & integrate, then inspect the discrepancy with the boundary values of the other end to adjust the estimate Given the starting boundary values u1 & u2 which contain the root u, solve u given the false position method (solving the differential equation as an initial value problem via 4th order RK), then use u to solve the differential equations. Finite Difference Method For linear & non-linear systems Higher order derivatives require more computational steps – some combinations for boundary conditions may not work though Improve the accuracy by increasing the number of mesh points Solving EigenValue Problems An eigenvalue can substitute a matrix when doing matrix multiplication à convert matrix multiplication into a polynomial EigenValue For a given set of equations in matrix form, determine what are the solution eigenvalue & eigenvectors Similar Matrices - have same eigenvalues. Use orthogonal similarity transforms to reduce a matrix to diagonal form from which eigenvalue(s) & eigenvectors can be computed iteratively Jacobi method http://en.wikipedia.org/wiki/Jacobi_method C++: http://people.sc.fsu.edu/~jburkardt/classes/acs2_2008/openmp/jacobi/jacobi.html Robust but Computationally intense – use for small matrices < 10x10 Power Iteration http://en.wikipedia.org/wiki/Power_iteration For any given real symmetric matrix, generate the largest single eigenvalue & its eigenvectors Simplest method – does not compute matrix decomposition à suitable for large, sparse matrices Inverse Iteration Variation of power iteration method – generates the smallest eigenvalue from the inverse matrix Rayleigh Method http://en.wikipedia.org/wiki/Rayleigh's_method_of_dimensional_analysis Variation of power iteration method Rayleigh Quotient Method Variation of inverse iteration method Matrix Tri-diagonalization Method Use householder algorithm to reduce an NxN symmetric matrix to a tridiagonal real symmetric matrix vua N-2 orthogonal transforms     Whats Next Outside of Numerical Methods there are lots of different types of algorithms that I’ve learned over the decades: Data Mining – (I covered this briefly in a previous post: http://geekswithblogs.net/JoshReuben/archive/2007/12/31/ssas-dm-algorithms.aspx ) Search & Sort Routing Problem Solving Logical Theorem Proving Planning Probabilistic Reasoning Machine Learning Solvers (eg MIP) Bioinformatics (Sequence Alignment, Protein Folding) Quant Finance (I read Wilmott’s books – interesting) Sooner or later, I’ll cover the above topics as well.

    Read the article

  • C# Neural Networks with Encog

    - by JoshReuben
    Neural Networks ·       I recently read a book Introduction to Neural Networks for C# , by Jeff Heaton. http://www.amazon.com/Introduction-Neural-Networks-C-2nd/dp/1604390093/ref=sr_1_2?ie=UTF8&s=books&qid=1296821004&sr=8-2-spell. Not the 1st ANN book I've perused, but a nice revision.   ·       Artificial Neural Networks (ANNs) are a mechanism of machine learning – see http://en.wikipedia.org/wiki/Artificial_neural_network , http://en.wikipedia.org/wiki/Category:Machine_learning ·       Problems Not Suited to a Neural Network Solution- Programs that are easily written out as flowcharts consisting of well-defined steps, program logic that is unlikely to change, problems in which you must know exactly how the solution was derived. ·       Problems Suited to a Neural Network – pattern recognition, classification, series prediction, and data mining. Pattern recognition - network attempts to determine if the input data matches a pattern that it has been trained to recognize. Classification - take input samples and classify them into fuzzy groups. ·       As far as machine learning approaches go, I thing SVMs are superior (see http://en.wikipedia.org/wiki/Support_vector_machine ) - a neural network has certain disadvantages in comparison: an ANN can be overtrained, different training sets can produce non-deterministic weights and it is not possible to discern the underlying decision function of an ANN from its weight matrix – they are black box. ·       In this post, I'm not going to go into internals (believe me I know them). An autoassociative network (e.g. a Hopfield network) will echo back a pattern if it is recognized. ·       Under the hood, there is very little maths. In a nutshell - Some simple matrix operations occur during training: the input array is processed (normalized into bipolar values of 1, -1) - transposed from input column vector into a row vector, these are subject to matrix multiplication and then subtraction of the identity matrix to get a contribution matrix. The dot product is taken against the weight matrix to yield a boolean match result. For backpropogation training, a derivative function is required. In learning, hill climbing mechanisms such as Genetic Algorithms and Simulated Annealing are used to escape local minima. For unsupervised training, such as found in Self Organizing Maps used for OCR, Hebbs rule is applied. ·       The purpose of this post is not to mire you in technical and conceptual details, but to show you how to leverage neural networks via an abstraction API - Encog   Encog ·       Encog is a neural network API ·       Links to Encog: http://www.encog.org , http://www.heatonresearch.com/encog, http://www.heatonresearch.com/forum ·       Encog requires .Net 3.5 or higher – there is also a Silverlight version. Third-Party Libraries – log4net and nunit. ·       Encog supports feedforward, recurrent, self-organizing maps, radial basis function and Hopfield neural networks. ·       Encog neural networks, and related data, can be stored in .EG XML files. ·       Encog Workbench allows you to edit, train and visualize neural networks. The Encog Workbench can generate code. Synapses and layers ·       the primary building blocks - Almost every neural network will have, at a minimum, an input and output layer. In some cases, the same layer will function as both input and output layer. ·       To adapt a problem to a neural network, you must determine how to feed the problem into the input layer of a neural network, and receive the solution through the output layer of a neural network. ·       The Input Layer - For each input neuron, one double value is stored. An array is passed as input to a layer. Encog uses the interface INeuralData to hold these arrays. The class BasicNeuralData implements the INeuralData interface. Once the neural network processes the input, an INeuralData based class will be returned from the neural network's output layer. ·       convert a double array into an INeuralData object : INeuralData data = new BasicNeuralData(= new double[10]); ·       the Output Layer- The neural network outputs an array of doubles, wraped in a class based on the INeuralData interface. ·        The real power of a neural network comes from its pattern recognition capabilities. The neural network should be able to produce the desired output even if the input has been slightly distorted. ·       Hidden Layers– optional. between the input and output layers. very much a “black box”. If the structure of the hidden layer is too simple it may not learn the problem. If the structure is too complex, it will learn the problem but will be very slow to train and execute. Some neural networks have no hidden layers. The input layer may be directly connected to the output layer. Further, some neural networks have only a single layer. A single layer neural network has the single layer self-connected. ·       connections, called synapses, contain individual weight matrixes. These values are changed as the neural network learns. Constructing a Neural Network ·       the XOR operator is a frequent “first example” -the “Hello World” application for neural networks. ·       The XOR Operator- only returns true when both inputs differ. 0 XOR 0 = 0 1 XOR 0 = 1 0 XOR 1 = 1 1 XOR 1 = 0 ·       Structuring a Neural Network for XOR  - two inputs to the XOR operator and one output. ·       input: 0.0,0.0 1.0,0.0 0.0,1.0 1.0,1.0 ·       Expected output: 0.0 1.0 1.0 0.0 ·       A Perceptron - a simple feedforward neural network to learn the XOR operator. ·       Because the XOR operator has two inputs and one output, the neural network will follow suit. Additionally, the neural network will have a single hidden layer, with two neurons to help process the data. The choice for 2 neurons in the hidden layer is arbitrary, and often comes down to trial and error. ·       Neuron Diagram for the XOR Network ·       ·       The Encog workbench displays neural networks on a layer-by-layer basis. ·       Encog Layer Diagram for the XOR Network:   ·       Create a BasicNetwork - Three layers are added to this network. the FinalizeStructure method must be called to inform the network that no more layers are to be added. The call to Reset randomizes the weights in the connections between these layers. var network = new BasicNetwork(); network.AddLayer(new BasicLayer(2)); network.AddLayer(new BasicLayer(2)); network.AddLayer(new BasicLayer(1)); network.Structure.FinalizeStructure(); network.Reset(); ·       Neural networks frequently start with a random weight matrix. This provides a starting point for the training methods. These random values will be tested and refined into an acceptable solution. However, sometimes the initial random values are too far off. Sometimes it may be necessary to reset the weights again, if training is ineffective. These weights make up the long-term memory of the neural network. Additionally, some layers have threshold values that also contribute to the long-term memory of the neural network. Some neural networks also contain context layers, which give the neural network a short-term memory as well. The neural network learns by modifying these weight and threshold values. ·       Now that the neural network has been created, it must be trained. Training a Neural Network ·       construct a INeuralDataSet object - contains the input array and the expected output array (of corresponding range). Even though there is only one output value, we must still use a two-dimensional array to represent the output. public static double[][] XOR_INPUT ={ new double[2] { 0.0, 0.0 }, new double[2] { 1.0, 0.0 }, new double[2] { 0.0, 1.0 }, new double[2] { 1.0, 1.0 } };   public static double[][] XOR_IDEAL = { new double[1] { 0.0 }, new double[1] { 1.0 }, new double[1] { 1.0 }, new double[1] { 0.0 } };   INeuralDataSet trainingSet = new BasicNeuralDataSet(XOR_INPUT, XOR_IDEAL); ·       Training is the process where the neural network's weights are adjusted to better produce the expected output. Training will continue for many iterations, until the error rate of the network is below an acceptable level. Encog supports many different types of training. Resilient Propagation (RPROP) - general-purpose training algorithm. All training classes implement the ITrain interface. The RPROP algorithm is implemented by the ResilientPropagation class. Training the neural network involves calling the Iteration method on the ITrain class until the error is below a specific value. The code loops through as many iterations, or epochs, as it takes to get the error rate for the neural network to be below 1%. Once the neural network has been trained, it is ready for use. ITrain train = new ResilientPropagation(network, trainingSet);   for (int epoch=0; epoch < 10000; epoch++) { train.Iteration(); Debug.Print("Epoch #" + epoch + " Error:" + train.Error); if (train.Error > 0.01) break; } Executing a Neural Network ·       Call the Compute method on the BasicNetwork class. Console.WriteLine("Neural Network Results:"); foreach (INeuralDataPair pair in trainingSet) { INeuralData output = network.Compute(pair.Input); Console.WriteLine(pair.Input[0] + "," + pair.Input[1] + ", actual=" + output[0] + ",ideal=" + pair.Ideal[0]); } ·       The Compute method accepts an INeuralData class and also returns a INeuralData object. Neural Network Results: 0.0,0.0, actual=0.002782538818034049,ideal=0.0 1.0,0.0, actual=0.9903741937121177,ideal=1.0 0.0,1.0, actual=0.9836807956566187,ideal=1.0 1.0,1.0, actual=0.0011646072586172778,ideal=0.0 ·       the network has not been trained to give the exact results. This is normal. Because the network was trained to 1% error, each of the results will also be within generally 1% of the expected value.

    Read the article

  • NET Math Libraries

    - by JoshReuben
    NET Mathematical Libraries   .NET Builder for Matlab The MathWorks Inc. - http://www.mathworks.com/products/netbuilder/ MATLAB Builder NE generates MATLAB based .NET and COM components royalty-free deployment creates the components by encrypting MATLAB functions and generating either a .NET or COM wrapper around them. .NET/Link for Mathematica www.wolfram.com a product that 2-way integrates Mathematica and Microsoft's .NET platform call .NET from Mathematica - use arbitrary .NET types directly from the Mathematica language. use and control the Mathematica kernel from a .NET program. turns Mathematica into a scripting shell to leverage the computational services of Mathematica. write custom front ends for Mathematica or use Mathematica as a computational engine for another program comes with full source code. Leverages MathLink - a Wolfram Research's protocol for sending data and commands back and forth between Mathematica and other programs. .NET/Link abstracts the low-level details of the MathLink C API. Extreme Optimization http://www.extremeoptimization.com/ a collection of general-purpose mathematical and statistical classes built for the.NET framework. It combines a math library, a vector and matrix library, and a statistics library in one package. download the trial of version 4.0 to try it out. Multi-core ready - Full support for Task Parallel Library features including cancellation. Broad base of algorithms covering a wide range of numerical techniques, including: linear algebra (BLAS and LAPACK routines), numerical analysis (integration and differentiation), equation solvers. Mathematics leverages parallelism using .NET 4.0's Task Parallel Library. Basic math: Complex numbers, 'special functions' like Gamma and Bessel functions, numerical differentiation. Solving equations: Solve equations in one variable, or solve systems of linear or nonlinear equations. Curve fitting: Linear and nonlinear curve fitting, cubic splines, polynomials, orthogonal polynomials. Optimization: find the minimum or maximum of a function in one or more variables, linear programming and mixed integer programming. Numerical integration: Compute integrals over finite or infinite intervals, over 2D and higher dimensional regions. Integrate systems of ordinary differential equations (ODE's). Fast Fourier Transforms: 1D and 2D FFT's using managed or fast native code (32 and 64 bit) BigInteger, BigRational, and BigFloat: Perform operations with arbitrary precision. Vector and Matrix Library Real and complex vectors and matrices. Single and double precision for elements. Structured matrix types: including triangular, symmetrical and band matrices. Sparse matrices. Matrix factorizations: LU decomposition, QR decomposition, singular value decomposition, Cholesky decomposition, eigenvalue decomposition. Portability and performance: Calculations can be done in 100% managed code, or in hand-optimized processor-specific native code (32 and 64 bit). Statistics Data manipulation: Sort and filter data, process missing values, remove outliers, etc. Supports .NET data binding. Statistical Models: Simple, multiple, nonlinear, logistic, Poisson regression. Generalized Linear Models. One and two-way ANOVA. Hypothesis Tests: 12 14 hypothesis tests, including the z-test, t-test, F-test, runs test, and more advanced tests, such as the Anderson-Darling test for normality, one and two-sample Kolmogorov-Smirnov test, and Levene's test for homogeneity of variances. Multivariate Statistics: K-means cluster analysis, hierarchical cluster analysis, principal component analysis (PCA), multivariate probability distributions. Statistical Distributions: 25 29 continuous and discrete statistical distributions, including uniform, Poisson, normal, lognormal, Weibull and Gumbel (extreme value) distributions. Random numbers: Random variates from any distribution, 4 high-quality random number generators, low discrepancy sequences, shufflers. New in version 4.0 (November, 2010) Support for .NET Framework Version 4.0 and Visual Studio 2010 TPL Parallellized – multicore ready sparse linear program solver - can solve problems with more than 1 million variables. Mixed integer linear programming using a branch and bound algorithm. special functions: hypergeometric, Riemann zeta, elliptic integrals, Frensel functions, Dawson's integral. Full set of window functions for FFT's. Product  Price Update subscription Single Developer License $999  $399  Team License (3 developers) $1999  $799  Department License (8 developers) $3999  $1599  Site License (Unlimited developers in one physical location) $7999  $3199    NMath http://www.centerspace.net .NET math and statistics libraries matrix and vector classes random number generators Fast Fourier Transforms (FFTs) numerical integration linear programming linear regression curve and surface fitting optimization hypothesis tests analysis of variance (ANOVA) probability distributions principal component analysis cluster analysis built on the Intel Math Kernel Library (MKL), which contains highly-optimized, extensively-threaded versions of BLAS (Basic Linear Algebra Subroutines) and LAPACK (Linear Algebra PACKage). Product  Price Update subscription Single Developer License $1295 $388 Team License (5 developers) $5180 $1554   DotNumerics http://www.dotnumerics.com/NumericalLibraries/Default.aspx free DotNumerics is a website dedicated to numerical computing for .NET that includes a C# Numerical Library for .NET containing algorithms for Linear Algebra, Differential Equations and Optimization problems. The Linear Algebra library includes CSLapack, CSBlas and CSEispack, ports from Fortran to C# of LAPACK, BLAS and EISPACK, respectively. Linear Algebra (CSLapack, CSBlas and CSEispack). Systems of linear equations, eigenvalue problems, least-squares solutions of linear systems and singular value problems. Differential Equations. Initial-value problem for nonstiff and stiff ordinary differential equations ODEs (explicit Runge-Kutta, implicit Runge-Kutta, Gear's BDF and Adams-Moulton). Optimization. Unconstrained and bounded constrained optimization of multivariate functions (L-BFGS-B, Truncated Newton and Simplex methods).   Math.NET Numerics http://numerics.mathdotnet.com/ free an open source numerical library - includes special functions, linear algebra, probability models, random numbers, interpolation, integral transforms. A merger of dnAnalytics with Math.NET Iridium in addition to a purely managed implementation will also support native hardware optimization. constants & special functions complex type support real and complex, dense and sparse linear algebra (with LU, QR, eigenvalues, ... decompositions) non-uniform probability distributions, multivariate distributions, sample generation alternative uniform random number generators descriptive statistics, including order statistics various interpolation methods, including barycentric approaches and splines numerical function integration (quadrature) routines integral transforms, like fourier transform (FFT) with arbitrary lengths support, and hartley spectral-space aware sequence manipulation (signal processing) combinatorics, polynomials, quaternions, basic number theory. parallelized where appropriate, to leverage multi-core and multi-processor systems fully managed or (if available) using native libraries (Intel MKL, ACMS, CUDA, FFTW) provides a native facade for F# developers

    Read the article

  • A Guided Tour of Complexity

    - by JoshReuben
    I just re-read Complexity – A Guided Tour by Melanie Mitchell , protégé of Douglas Hofstadter ( author of “Gödel, Escher, Bach”) http://www.amazon.com/Complexity-Guided-Tour-Melanie-Mitchell/dp/0199798109/ref=sr_1_1?ie=UTF8&qid=1339744329&sr=8-1 here are some notes and links:   Evolved from Cybernetics, General Systems Theory, Synergetics some interesting transdisciplinary fields to investigate: Chaos Theory - http://en.wikipedia.org/wiki/Chaos_theory – small differences in initial conditions (such as those due to rounding errors in numerical computation) yield widely diverging outcomes for chaotic systems, rendering long-term prediction impossible. System Dynamics / Cybernetics - http://en.wikipedia.org/wiki/System_Dynamics – study of how feedback changes system behavior Network Theory - http://en.wikipedia.org/wiki/Network_theory – leverage Graph Theory to analyze symmetric  / asymmetric relations between discrete objects Algebraic Topology - http://en.wikipedia.org/wiki/Algebraic_topology – leverage abstract algebra to analyze topological spaces There are limits to deterministic systems & to computation. Chaos Theory definitely applies to training an ANN (artificial neural network) – different weights will emerge depending upon the random selection of the training set. In recursive Non-Linear systems http://en.wikipedia.org/wiki/Nonlinear_system – output is not directly inferable from input. E.g. a Logistic map: Xt+1 = R Xt(1-Xt) Different types of bifurcations, attractor states and oscillations may occur – e.g. a Lorenz Attractor http://en.wikipedia.org/wiki/Lorenz_system Feigenbaum Constants http://en.wikipedia.org/wiki/Feigenbaum_constants express ratios in a bifurcation diagram for a non-linear map – the convergent limit of R (the rate of period-doubling bifurcations) is 4.6692016 Maxwell’s Demon - http://en.wikipedia.org/wiki/Maxwell%27s_demon - the Second Law of Thermodynamics has only a statistical certainty – the universe (and thus information) tends towards entropy. While any computation can theoretically be done without expending energy, with finite memory, the act of erasing memory is permanent and increases entropy. Life & thought is a counter-example to the universe’s tendency towards entropy. Leo Szilard and later Claude Shannon came up with the Information Theory of Entropy - http://en.wikipedia.org/wiki/Entropy_(information_theory) whereby Shannon entropy quantifies the expected value of a message’s information in bits in order to determine channel capacity and leverage Coding Theory (compression analysis). Ludwig Boltzmann came up with Statistical Mechanics - http://en.wikipedia.org/wiki/Statistical_mechanics – whereby our Newtonian perception of continuous reality is a probabilistic and statistical aggregate of many discrete quantum microstates. This is relevant for Quantum Information Theory http://en.wikipedia.org/wiki/Quantum_information and the Physics of Information - http://en.wikipedia.org/wiki/Physical_information. Hilbert’s Problems http://en.wikipedia.org/wiki/Hilbert's_problems pondered whether mathematics is complete, consistent, and decidable (the Decision Problem – http://en.wikipedia.org/wiki/Entscheidungsproblem – is there always an algorithm that can determine whether a statement is true).  Godel’s Incompleteness Theorems http://en.wikipedia.org/wiki/G%C3%B6del's_incompleteness_theorems  proved that mathematics cannot be both complete and consistent (e.g. “This statement is not provable”). Turing through the use of Turing Machines (http://en.wikipedia.org/wiki/Turing_machine symbol processors that can prove mathematical statements) and Universal Turing Machines (http://en.wikipedia.org/wiki/Universal_Turing_machine Turing Machines that can emulate other any Turing Machine via accepting programs as well as data as input symbols) that computation is limited by demonstrating the Halting Problem http://en.wikipedia.org/wiki/Halting_problem (is is not possible to know when a program will complete – you cannot build an infinite loop detector). You may be used to thinking of 1 / 2 / 3 dimensional systems, but Fractal http://en.wikipedia.org/wiki/Fractal systems are defined by self-similarity & have non-integer Hausdorff Dimensions !!!  http://en.wikipedia.org/wiki/List_of_fractals_by_Hausdorff_dimension – the fractal dimension quantifies the number of copies of a self similar object at each level of detail – eg Koch Snowflake - http://en.wikipedia.org/wiki/Koch_snowflake Definitions of complexity: size, Shannon entropy, Algorithmic Information Content (http://en.wikipedia.org/wiki/Algorithmic_information_theory - size of shortest program that can generate a description of an object) Logical depth (amount of info processed), thermodynamic depth (resources required). Complexity is statistical and fractal. John Von Neumann’s other machine was the Self-Reproducing Automaton http://en.wikipedia.org/wiki/Self-replicating_machine  . Cellular Automata http://en.wikipedia.org/wiki/Cellular_automaton are alternative form of Universal Turing machine to traditional Von Neumann machines where grid cells are locally synchronized with their neighbors according to a rule. Conway’s Game of Life http://en.wikipedia.org/wiki/Conway's_Game_of_Life demonstrates various emergent constructs such as “Glider Guns” and “Spaceships”. Cellular Automatons are not practical because logical ops require a large number of cells – wasteful & inefficient. There are no compilers or general program languages available for Cellular Automatons (as far as I am aware). Random Boolean Networks http://en.wikipedia.org/wiki/Boolean_network are extensions of cellular automata where nodes are connected at random (not to spatial neighbors) and each node has its own rule –> they demonstrate the emergence of complex  & self organized behavior. Stephen Wolfram’s (creator of Mathematica, so give him the benefit of the doubt) New Kind of Science http://en.wikipedia.org/wiki/A_New_Kind_of_Science proposes the universe may be a discrete Finite State Automata http://en.wikipedia.org/wiki/Finite-state_machine whereby reality emerges from simple rules. I am 2/3 through this book. It is feasible that the universe is quantum discrete at the plank scale and that it computes itself – Digital Physics: http://en.wikipedia.org/wiki/Digital_physics – a simulated reality? Anyway, all behavior is supposedly derived from simple algorithmic rules & falls into 4 patterns: uniform , nested / cyclical, random (Rule 30 http://en.wikipedia.org/wiki/Rule_30) & mixed (Rule 110 - http://en.wikipedia.org/wiki/Rule_110 localized structures – it is this that is interesting). interaction between colliding propagating signal inputs is then information processing. Wolfram proposes the Principle of Computational Equivalence - http://mathworld.wolfram.com/PrincipleofComputationalEquivalence.html - all processes that are not obviously simple can be viewed as computations of equivalent sophistication. Meaning in information may emerge from analogy & conceptual slippages – see the CopyCat program: http://cognitrn.psych.indiana.edu/rgoldsto/courses/concepts/copycat.pdf Scale Free Networks http://en.wikipedia.org/wiki/Scale-free_network have a distribution governed by a Power Law (http://en.wikipedia.org/wiki/Power_law - much more common than Normal Distribution). They are characterized by hubs (resilience to random deletion of nodes), heterogeneity of degree values, self similarity, & small world structure. They grow via preferential attachment http://en.wikipedia.org/wiki/Preferential_attachment – tipping points triggered by positive feedback loops. 2 theories of cascading system failures in complex systems are Self-Organized Criticality http://en.wikipedia.org/wiki/Self-organized_criticality and Highly Optimized Tolerance http://en.wikipedia.org/wiki/Highly_optimized_tolerance. Computational Mechanics http://en.wikipedia.org/wiki/Computational_mechanics – use of computational methods to study phenomena governed by the principles of mechanics. This book is a great intuition pump, but does not cover the more mathematical subject of Computational Complexity Theory – http://en.wikipedia.org/wiki/Computational_complexity_theory I am currently reading this book on this subject: http://www.amazon.com/Computational-Complexity-Christos-H-Papadimitriou/dp/0201530821/ref=pd_sim_b_1   stay tuned for that review!

    Read the article

  • Azure WNS to Win8 - Push Notifications for Metro Apps

    - by JoshReuben
    Background The Windows Azure Toolkit for Windows 8 allows you to build a Windows Azure Cloud Service that can send Push Notifications to registered Metro apps via Windows Notification Service (WNS). Some configuration is required - you need to: Register the Metro app for Windows Live Application Management Provide Package SID & Client Secret to WNS Modify the Azure Cloud App cscfg file and the Metro app package.appxmanifest file to contain matching Metro package name, SID and client secret. The Mechanism: These notifications take the form of XAML Tile, Toast, Raw or Badge UI notifications. The core engine is provided via the WNS nuget recipe, which exposes an API for constructing payloads and posting notifications to WNS. An application receives push notifications by requesting a notification channel from WNS, which returns a channel URI that the application then registers with a cloud service. In the cloud service, A WnsAccessTokenProvider authenticates with WNS by providing its credentials, the package SID and secret key, and receives in return an access token that the provider caches and can reuse for multiple notification requests. The cloud service constructs a notification request by filling out a template class that contains the information that will be sent with the notification, including text and image references. Using the channel URI of a registered client, the cloud service can then send a notification whenever it has an update for the user. The package contains the NotificationSendUtils class for submitting notifications. The Windows Azure Toolkit for Windows 8 (WAT) provides the PNWorker sample pair of solutions - The Azure server side contains a WebRole & a WorkerRole. The WebRole allows submission of new push notifications into an Azure Queue which the WorkerRole extracts and processes. Further background resources: http://watwindows8.codeplex.com/ - Windows Azure Toolkit for Windows 8 http://watwindows8.codeplex.com/wikipage?title=Push%20Notification%20Worker%20Sample - WAT WNS sample setup http://watwindows8.codeplex.com/wikipage?title=Using%20the%20Windows%208%20Cloud%20Application%20Services%20Application – using Windows 8 with Cloud Application Services A bit of Configuration Register the Metro apps for Windows Live Application Management From the current app manifest of your metro app Publish tab, copy the Package Display Name and the Publisher From: https://manage.dev.live.com/Build/ Package name: <-- we need to change this Client secret: keep this Package Security Identifier (SID): keep this Verify the app here: https://manage.dev.live.com/Applications/Index - so this step is done "If you wish to send push notifications in your application, provide your Package Security Identifier (SID) and client secret to WNS." Provide Package SID & Client Secret to WNS http://msdn.microsoft.com/en-us/library/windows/apps/hh465407.aspx - How to authenticate with WNS https://appdev.microsoft.com/StorePortals/en-us/Account/Signup/PurchaseSubscription - register app with dashboard - need registration code or register a new account & pay $170 shekels http://msdn.microsoft.com/en-us/library/windows/apps/hh868184.aspx - Registering for a Windows Store developer account http://msdn.microsoft.com/en-us/library/windows/apps/hh868187.aspx - Picking a Microsoft account for the Windows Store The WNS Nuget Recipe The WNS Recipe is a nuget package that provides an API for authenticating against WNS, constructing payloads and posting notifications to WNS. After installing this package, a WnsRecipe assembly is added to project references. To send notifications using WNS, first register the application at the Windows Push Notifications & Live Connect portal to obtain Package Security Identifier (SID) and a secret key that your cloud service uses to authenticate with WNS. An application receives push notifications by requesting a notification channel from WNS, which returns a channel URI that the application then registers with a cloud service. In the cloud service, the WnsAccessTokenProvider authenticates with WNS by providing its credentials, the package SID and secret key, and receives in return an access token that the provider caches and can reuse for multiple notification requests. The cloud service constructs a notification request by filling out a template class that contains the information that will be sent with the notification, including text and image references.Using the channel URI of a registered client, the cloud service can then send a notification whenever it has an update for the user. var provider = new WnsAccessTokenProvider(clientId, clientSecret); var notification = new ToastNotification(provider) {     ToastType = ToastType.ToastText02,     Text = new List<string> { "blah"} }; notification.Send(channelUri); the WNS Recipe is instrumented to write trace information via a trace listener – configuratively or programmatically from Application_Start(): WnsDiagnostics.Enable(); WnsDiagnostics.TraceSource.Listeners.Add(new DiagnosticMonitorTraceListener()); WnsDiagnostics.TraceSource.Switch.Level = SourceLevels.Verbose; The WAT PNWorker Sample The Azure server side contains a WebRole & a WorkerRole. The WebRole allows submission of new push notifications into an Azure Queue which the WorkerRole extracts and processes. Overview of Push Notification Worker Sample The toolkit includes a sample application based on the same solution structure as the one created by theWindows 8 Cloud Application Services project template. The sample demonstrates how to off-load the job of sending Windows Push Notifications using a Windows Azure worker role. You can find the source code in theSamples\PNWorker folder. This folder contains a full version of the sample application showing how to use Windows Push Notifications using ASP.NET Membership as the authentication mechanism. The sample contains two different solution files: WATWindows.Azure.sln: This solution must be opened with Visual Studio 2010 and contains the projects related to the Windows Azure web and worker roles. WATWindows.Client.sln: This solution must be opened with Visual Studio 11 and contains the Windows Metro style application project. Only Visual Studio 2010 supports Windows Azure cloud projects so you currently need to use this edition to launch the server application. This will change in a future release of the Windows Azure tools when support for Visual Studio 11 is enabled. Important: Setting up the PNWorker Sample Before running the PNWorker sample, you need to register the application and configure it: 1. Register the app: To register your application, go to the Windows Live Application Management site for Metro style apps at https://manage.dev.live.com/build and sign in with your Windows Live ID. In the Windows Push Notifications & Live Connect page, enter the following information. Package Display Name PNWorker.Sample Publisher CN=127.0.0.1, O=TESTING ONLY, OU=Windows Azure DevFabric 2. 3. Once you register the application, make a note of the values shown in the portal for Client Secret,Package Name and Package SID. 4. Configure the app - double-click the SetupSample.cmd file located inside the Samples\PNWorker folder to launch a tool that will guide you through the process of configuring the sample. setup runs a PowerShell script that requires running with administration privileges to allow the scripts to execute in your machine. When prompted, enter the Client Secret, Package Name, and Package Security Identifier you obtained previously and wait until the tool finishes configuring your sample. Running the PNWorker Sample To run this sample, you must run both the client and the server application projects. 1. Open Visual Studio 2010 as an administrator. Open the WATWindows.Azure.sln solution. Set the start-up project of the solution as the cloud project. Run the app in the dev fabric to test. 2. Open Visual Studio 11 and open the WATWindows.Client.sln solution. Run the Metro client application. In the client application, click Reopen channel and send to server. à the application opens the channel and registers it with the cloud application, & the Output area shows the channel URI. 3. Refresh the WebRole's Push Notifications page to see the UI list the newly registered client. 4. Send notifications to the client application by clicking the Send Notification button. Setup 3 command files + 1 powershell script: SetupSample.cmd –> SetupWPNS.vbs –> SetupWPNS.cmd –> SetupWPNS.UpdateWPNSCredentialsInServiceConfiguration.ps1 appears to set PackageName – from manifest Client Id package security id (SID) – from registration Client Secret – from registration The following configs are modified: WATWindows\ServiceConfiguration.Cloud.cscfg WATWindows\ServiceConfiguration.Local.cscfg WATWindows.Client\package.appxmanifest WatWindows.Notifications A class library – it references the following WNS DLL: C:\WorkDev\CountdownValue\AzureToolkits\WATWindows8\Samples\PNWorker\packages\WnsRecipe.0.0.3.0\lib\net40\WnsRecipe.dll NotificationJobRequest A DataContract for triggering notifications:     using System.Runtime.Serialization; using Microsoft.Windows.Samples.Notifications;     [DataContract]     [KnownType(typeof(WnsAccessTokenProvider))] public class NotificationJobRequest     {               [DataMember] public bool ProcessAsync { get; set; }          [DataMember] public string Payload { get; set; }         [DataMember] public string ChannelUrl { get; set; }         [DataMember] public NotificationType NotificationType { get; set; }         [DataMember] public IAccessTokenProvider AccessTokenProvider { get; set; }         [DataMember] public NotificationSendOptions NotificationSendOptions{ get; set; }     } Investigated these types: WnsAccessTokenProvider – a DataContract that contains the client Id and client secret NotificationType – an enum that can be: Tile, Toast, badge, Raw IAccessTokenProvider – get or reset the access token NotificationSendOptions – SecondsTTL, NotificationPriority (enum), isCache, isRequestForStatus, Tag   There is also a NotificationJobSerializer class which basically wraps a DataContractSerializer serialization / deserialization of NotificationJobRequest The WNSNotificationJobProcessor class This class wraps the NotificationSendUtils API – it periodically extracts any NotificationJobRequest objects from a CloudQueue and submits them to WNS. The ProcessJobMessageRequest method – this is the punchline: it will deserialize a CloudQueueMessage into a NotificationJobRequest & send pass its contents to NotificationUtils to SendAsynchronously / SendSynchronously, (and then dequeue the message).     public override void ProcessJobMessageRequest(CloudQueueMessage notificationJobMessageRequest)         { Trace.WriteLine("Processing a new Notification Job Request", "Information"); NotificationJobRequest pushNotificationJob =                 NotificationJobSerializer.Deserialize(notificationJobMessageRequest.AsString); if (pushNotificationJob != null)             { if (pushNotificationJob.ProcessAsync)                 { Trace.WriteLine("Sending the notification asynchronously", "Information"); NotificationSendUtils.SendAsynchronously( new Uri(pushNotificationJob.ChannelUrl),                         pushNotificationJob.AccessTokenProvider,                         pushNotificationJob.Payload,                         result => this.ProcessSendResult(pushNotificationJob, result),                         result => this.ProcessSendResultError(pushNotificationJob, result),                         pushNotificationJob.NotificationType,                         pushNotificationJob.NotificationSendOptions);                 } else                 { Trace.WriteLine("Sending the notification synchronously", "Information"); NotificationSendResult result = NotificationSendUtils.Send( new Uri(pushNotificationJob.ChannelUrl),                         pushNotificationJob.AccessTokenProvider,                         pushNotificationJob.Payload,                         pushNotificationJob.NotificationType,                         pushNotificationJob.NotificationSendOptions); this.ProcessSendResult(pushNotificationJob, result);                 }             } else             { Trace.WriteLine("Could not deserialize the notification job", "Error");             } this.queue.DeleteMessage(notificationJobMessageRequest);         } Investigation of NotificationSendUtils class - This is the engine – it exposes Send and a SendAsyncronously overloads that take the following params from the NotificationJobRequest: Channel Uri AccessTokenProvider Payload NotificationType NotificationSendOptions WebRole WebRole is a large MVC project – it references WatWindows.Notifications as well as the following WNS DLL: \AzureToolkits\WATWindows8\Samples\PNWorker\packages\WnsRecipe.0.0.3.0\lib\net40\NotificationsExtensions.dll Controllers\PushNotificationController.cs Notification related namespaces:     using Notifications;     using NotificationsExtensions;     using NotificationsExtensions.BadgeContent;     using NotificationsExtensions.RawContent;     using NotificationsExtensions.TileContent;     using NotificationsExtensions.ToastContent;     using Windows.Samples.Notifications; TokenProvider – initialized from the Azure RoleEnvironment:   IAccessTokenProvider tokenProvider = new WnsAccessTokenProvider(         RoleEnvironment.GetConfigurationSettingValue("WNSPackageSID"),         RoleEnvironment.GetConfigurationSettingValue("WNSClientSecret")); SendNotification method – calls QueuePushMessage method to create and serialize a NotificationJobRequest and enqueue it in a CloudQueue [HttpPost]         public ActionResult SendNotification(             [ModelBinder(typeof(NotificationTemplateModelBinder))] INotificationContent notification,             string channelUrl,             NotificationPriority priority = NotificationPriority.Normal)         {             var payload = notification.GetContent();             var options = new NotificationSendOptions()             {                 Priority = priority             };             var notificationType =                 notification is IBadgeNotificationContent ? NotificationType.Badge :                 notification is IRawNotificationContent ? NotificationType.Raw :                 notification is ITileNotificationContent ? NotificationType.Tile :                 NotificationType.Toast;             this.QueuePushMessage(payload, channelUrl, notificationType, options);             object response = new             {                 Status = "Queued for delivery to WNS"             };             return this.Json(response);         } GetSendTemplate method: Create the cshtml partial rendering based on the notification type     [HttpPost]         public ActionResult GetSendTemplate(NotificationTemplateViewModel templateOptions)         {             PartialViewResult result = null;             switch (templateOptions.NotificationType)             {                 case "Badge":                     templateOptions.BadgeGlyphValueContent = Enum.GetNames(typeof( GlyphValue));                     ViewBag.ViewData = templateOptions;                     result = PartialView("_" + templateOptions.NotificationTemplateType);                     break;                 case "Raw":                     ViewBag.ViewData = templateOptions;                     result = PartialView("_Raw");                     break;                 case "Toast":                     templateOptions.TileImages = this.blobClient.GetAllBlobsInContainer(ConfigReader.GetConfigValue("TileImagesContainer")).OrderBy(i => i.FileName).ToList();                     templateOptions.ToastAudioContent = Enum.GetNames(typeof( ToastAudioContent));                     templateOptions.Priorities = Enum.GetNames(typeof( NotificationPriority));                     ViewBag.ViewData = templateOptions;                     result = PartialView("_" + templateOptions.NotificationTemplateType);                     break;                 case "Tile":                     templateOptions.TileImages = this.blobClient.GetAllBlobsInContainer(ConfigReader.GetConfigValue("TileImagesContainer")).OrderBy(i => i.FileName).ToList();                     ViewBag.ViewData = templateOptions;                     result = PartialView("_" + templateOptions.NotificationTemplateType);                     break;             }             return result;         } Investigated these types: ToastAudioContent – an enum of different Win8 sound effects for toast notifications GlyphValue – an enum of different Win8 icons for badge notifications · Infrastructure\NotificationTemplateModelBinder.cs WNS Namespace references     using NotificationsExtensions.BadgeContent;     using NotificationsExtensions.RawContent;     using NotificationsExtensions.TileContent;     using NotificationsExtensions.ToastContent; Various NotificationFactory derived types can server as bindable models in MVC for creating INotificationContent types. Default values are also set for IWideTileNotificationContent & IToastNotificationContent. Type factoryType = null;             switch (notificationType)             {                 case "Badge":                     factoryType = typeof(BadgeContentFactory);                     break;                 case "Tile":                     factoryType = typeof(TileContentFactory);                     break;                 case "Toast":                     factoryType = typeof(ToastContentFactory);                     break;                 case "Raw":                     factoryType = typeof(RawContentFactory);                     break;             } Investigated these types: BadgeContentFactory – CreateBadgeGlyph, CreateBadgeNumeric (???) TileContentFactory – many notification content creation methods , apparently one for every tile layout type ToastContentFactory – many notification content creation methods , apparently one for every toast layout type RawContentFactory – passing strings WorkerRole WNS Namespace references using Notifications; using Notifications.WNS; using Windows.Samples.Notifications; OnStart() Method – on Worker Role startup, initialize the NotificationJobSerializer, the CloudQueue, and the WNSNotificationJobProcessor _notificationJobSerializer = new NotificationJobSerializer(); _cloudQueueClient = this.account.CreateCloudQueueClient(); _pushNotificationRequestsQueue = _cloudQueueClient.GetQueueReference(ConfigReader.GetConfigValue("RequestQueueName")); _processor = new WNSNotificationJobProcessor(_notificationJobSerializer, _pushNotificationRequestsQueue); Run() Method – poll the Azure Queue for NotificationJobRequest messages & process them:   while (true)             { Trace.WriteLine("Checking for Messages", "Information"); try                 { Parallel.ForEach( this.pushNotificationRequestsQueue.GetMessages(this.batchSize), this.processor.ProcessJobMessageRequest);                 } catch (Exception e)                 { Trace.WriteLine(e.ToString(), "Error");                 } Trace.WriteLine(string.Format("Sleeping for {0} seconds", this.pollIntervalMiliseconds / 1000)); Thread.Sleep(this.pollIntervalMiliseconds);                                            } How I learned to appreciate Win8 There is really only one application architecture for Windows 8 apps: Metro client side and Azure backend – and that is a good thing. With WNS, tier integration is so automated that you don’t even have to leverage a HTTP push API such as SignalR. This is a pretty powerful development paradigm, and has changed the way I look at Windows 8 for RAD business apps. When I originally looked at Win8 and the WinRT API, my first opinion on Win8 dev was as follows – GOOD:WinRT, WRL, C++/CX, WinJS, XAML (& ease of Direct3D integration); BAD: low projected market penetration,.NET lobotomized (Only 8% of .NET 4.5 classes can be used in Win8 non-desktop apps - http://bit.ly/HRuJr7); UGLY:Metro pascal tiles! Perhaps my 80s teenage years gave me a punk reactionary sense of revulsion towards the Partridge Family 70s style that Metro UX seems to have appropriated: On second thought though, it simplifies UI dev to a single paradigm (although UX guys will need to change career) – you will not find an easier app dev environment. Speculation: If LightSwitch is going to support HTML5 client app generation, then its a safe guess to say that vnext will support Win8 Metro XAML - a much easier port from Silverlight XAML. Given the VS2012 LightSwitch integration as a thumbs up from the powers that be at MS, and given that Win8 C#/XAML Metro apps tend towards a streamlined 'golden straight-jacket' cookie cutter app dev style with an Azure back-end supporting Win8 push notifications... --> its easy to extrapolate than LightSwitch vnext could well be the Win8 Metro XAML to Azure RAD tool of choice! The hook is already there - :) Why else have the space next to the HTML Client box? This high level of application development abstraction will facilitate rapid app cookie-cutter architecture-infrastructure frameworks for wrapping any app. This will allow me to avoid too much XAML code-monkeying around & focus on my area of interest: Technical Computing.

    Read the article

  • HPC Server Dynamic Job Scheduling: when jobs spawn jobs

    - by JoshReuben
    HPC Job Types HPC has 3 types of jobs http://technet.microsoft.com/en-us/library/cc972750(v=ws.10).aspx · Task Flow – vanilla sequence · Parametric Sweep – concurrently run multiple instances of the same program, each with a different work unit input · MPI – message passing between master & slave tasks But when you try go outside the box – job tasks that spawn jobs, blocking the parent task – you run the risk of resource starvation, deadlocks, and recursive, non-converging or exponential blow-up. The solution to this is to write some performance monitoring and job scheduling code. You can do this in 2 ways: manually control scheduling - allocate/ de-allocate resources, change job priorities, pause & resume tasks , restrict long running tasks to specific compute clusters Semi-automatically - set threshold params for scheduling. How – Control Job Scheduling In order to manage the tasks and resources that are associated with a job, you will need to access the ISchedulerJob interface - http://msdn.microsoft.com/en-us/library/microsoft.hpc.scheduler.ischedulerjob_members(v=vs.85).aspx This really allows you to control how a job is run – you can access & tweak the following features: max / min resource values whether job resources can grow / shrink, and whether jobs can be pre-empted, whether the job is exclusive per node the creator process id & the job pool timestamp of job creation & completion job priority, hold time & run time limit Re-queue count Job progress Max/ min Number of cores, nodes, sockets, RAM Dynamic task list – can add / cancel jobs on the fly Job counters When – poll perf counters Tweaking the job scheduler should be done on the basis of resource utilization according to PerfMon counters – HPC exposes 2 Perf objects: Compute Clusters, Compute Nodes http://technet.microsoft.com/en-us/library/cc720058(v=ws.10).aspx You can monitor running jobs according to dynamic thresholds – use your own discretion: Percentage processor time Number of running jobs Number of running tasks Total number of processors Number of processors in use Number of processors idle Number of serial tasks Number of parallel tasks Design Your algorithms correctly Finally , don’t assume you have unlimited compute resources in your cluster – design your algorithms with the following factors in mind: · Branching factor - http://en.wikipedia.org/wiki/Branching_factor - dynamically optimize the number of children per node · cutoffs to prevent explosions - http://en.wikipedia.org/wiki/Limit_of_a_sequence - not all functions converge after n attempts. You also need a threshold of good enough, diminishing returns · heuristic shortcuts - http://en.wikipedia.org/wiki/Heuristic - sometimes an exhaustive search is impractical and short cuts are suitable · Pruning http://en.wikipedia.org/wiki/Pruning_(algorithm) – remove / de-prioritize unnecessary tree branches · avoid local minima / maxima - http://en.wikipedia.org/wiki/Local_minima - sometimes an algorithm cant converge because it gets stuck in a local saddle – try simulated annealing, hill climbing or genetic algorithms to get out of these ruts   watch out for rounding errors – http://en.wikipedia.org/wiki/Round-off_error - multiple iterations can in parallel can quickly amplify & blow up your algo ! Use an epsilon, avoid floating point errors,  truncations, approximations Happy Coding !

    Read the article

  • Unification of TPL TaskScheduler and RX IScheduler

    - by JoshReuben
    using System; using System.Collections.Generic; using System.Reactive.Concurrency; using System.Security; using System.Threading; using System.Threading.Tasks; using System.Windows.Threading; namespace TPLRXSchedulerIntegration { public class MyScheduler :TaskScheduler, IScheduler     { private readonly Dispatcher _dispatcher; private readonly DispatcherScheduler _rxDispatcherScheduler; //private readonly TaskScheduler _tplDispatcherScheduler; private readonly SynchronizationContext _synchronizationContext; public MyScheduler(Dispatcher dispatcher)         {             _dispatcher = dispatcher;             _rxDispatcherScheduler = new DispatcherScheduler(dispatcher); //_tplDispatcherScheduler = FromCurrentSynchronizationContext();             _synchronizationContext = SynchronizationContext.Current;         }         #region RX public DateTimeOffset Now         { get { return _rxDispatcherScheduler.Now; }         } public IDisposable Schedule<TState>(TState state, DateTimeOffset dueTime, Func<IScheduler, TState, IDisposable> action)         { return _rxDispatcherScheduler.Schedule(state, dueTime, action);         } public IDisposable Schedule<TState>(TState state, TimeSpan dueTime, Func<IScheduler, TState, IDisposable> action)         { return _rxDispatcherScheduler.Schedule(state, dueTime, action);         } public IDisposable Schedule<TState>(TState state, Func<IScheduler, TState, IDisposable> action)         { return _rxDispatcherScheduler.Schedule(state, action);         }         #endregion         #region TPL /// Simply posts the tasks to be executed on the associated SynchronizationContext         [SecurityCritical] protected override void QueueTask(Task task)         {             _dispatcher.BeginInvoke((Action)(() => TryExecuteTask(task))); //TryExecuteTaskInline(task,false); //task.Start(_tplDispatcherScheduler); //m_synchronizationContext.Post(s_postCallback, (object)task);         } /// The task will be executed inline only if the call happens within the associated SynchronizationContext         [SecurityCritical] protected override bool TryExecuteTaskInline(Task task, bool taskWasPreviouslyQueued)         { if (SynchronizationContext.Current != _synchronizationContext)             { SynchronizationContext.SetSynchronizationContext(_synchronizationContext);             } return TryExecuteTask(task);         } // not implemented         [SecurityCritical] protected override IEnumerable<Task> GetScheduledTasks()         { return null;         } /// Implementes the MaximumConcurrencyLevel property for this scheduler class. /// By default it returns 1, because a <see cref="T:System.Threading.SynchronizationContext"/> based /// scheduler only supports execution on a single thread. public override Int32 MaximumConcurrencyLevel         { get             { return 1;             }         } //// preallocated SendOrPostCallback delegate //private static SendOrPostCallback s_postCallback = new SendOrPostCallback(PostCallback); //// this is where the actual task invocation occures //private static void PostCallback(object obj) //{ //    Task task = (Task) obj; //    // calling ExecuteEntry with double execute check enabled because a user implemented SynchronizationContext could be buggy //    task.ExecuteEntry(true); //}         #endregion     } }     What Design Pattern did I use here?

    Read the article

  • Azure Grid Computing - Worker Roles as HPC Compute Nodes

    - by JoshReuben
    Overview ·        With HPC 2008 R2 SP1 You can add Azure worker roles as compute nodes in a local Windows HPC Server cluster. ·        The subscription for Windows Azure like any other Azure Service - charged for the time that the role instances are available, as well as for the compute and storage services that are used on the nodes. ·        Win-Win ? - Azure charges the computer hour cost (according to vm size) amortized over a month – so you save on purchasing compute node hardware. Microsoft wins because you need to purchase HPC to have a local head node for managing this compute cluster grid distributed in the cloud. ·        Blob storage is used to hold input & output files of each job. I can see how Parametric Sweep HPC jobs can be supported (where the same job is run multiple times on each node against different input units), but not MPI.NET (where different HPC Job instances function as coordinated agents and conduct master-slave inter-process communication), unless Azure is somehow tunneling MPI communication through inter-WorkerRole Azure Queues. ·        this is not the end of the story for Azure Grid Computing. If MS requires you to purchase a local HPC license (and administrate it), what's to stop a 3rd party from doing this and encapsulating exposing HPC WCF Broker Service to you for managing compute nodes? If MS doesn’t  provide head node as a service, someone else will! Process ·        requires creation of a worker node template that specifies a connection to an existing subscription for Windows Azure + an availability policy for the worker nodes. ·        After worker nodes are added to the cluster, you can start them, which provisions the Windows Azure role instances, and then bring them online to run HPC cluster jobs. ·        A Windows Azure worker role instance runs a HPC compatible Azure guest operating system which runs on the VMs that host your service. The guest operating system is updated monthly. You can choose to upgrade the guest OS for your service automatically each time an update is released - All role instances defined by your service will run on the guest operating system version that you specify. see Windows Azure Guest OS Releases and SDK Compatibility Matrix (http://go.microsoft.com/fwlink/?LinkId=190549). ·        use the hpcpack command to upload file packages and install files to run on the worker nodes. see hpcpack (http://go.microsoft.com/fwlink/?LinkID=205514). Requirements ·        assuming you have an azure subscription account and the HPC head node installed and configured. ·        Install HPC Pack 2008 R2 SP 1 -  see Microsoft HPC Pack 2008 R2 Service Pack 1 Release Notes (http://go.microsoft.com/fwlink/?LinkID=202812). ·        Configure the head node to connect to the Internet - connectivity is provided by the connection of the head node to the enterprise network. You may need to configure a proxy client on the head node. Any cluster network topology (1-5) is supported). ·        Configure the firewall - allow outbound TCP traffic on the following ports: 80,       443, 5901, 5902, 7998, 7999 ·        Note: HPC Server  uses Admin Mode (Elevated Privileges) in Windows Azure to give the service administrator of the subscription the necessary privileges to initialize HPC cluster services on the worker nodes. ·        Obtain a Windows Azure subscription certificate - the Windows Azure subscription must be configured with a public subscription (API) certificate -a valid X.509 certificate with a key size of at least 2048 bits. Generate a self-sign certificate & upload a .cer file to the Windows Azure Portal Account page > Manage my API Certificates link. see Using the Windows Azure Service Management API (http://go.microsoft.com/fwlink/?LinkId=205526). ·        import the certificate with an associated private key on the HPC cluster head node - into the trusted root store of the local computer account. Obtain Windows Azure Connection Information for HPC Server ·        required for each worker node template ·        copy from azure portal - Get from: navigation pane > Hosted Services > Storage Accounts & CDN ·        Subscription ID - a 32-char hex string in the form xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx. In Properties pane. ·        Subscription certificate thumbprint - a 40-char hex string (you need to remove spaces). In Management Certificates > Properties pane. ·        Service name - the value of <ServiceName> configured in the public URL of the service (http://<ServiceName>.cloudapp.net). In Hosted Services > Properties pane. ·        Blob Storage account name - the value of <StorageAccountName> configured in the public URL of the account (http://<StorageAccountName>.blob.core.windows.net). In Storage Accounts > Properties pane. Import the Azure Subscription Certificate on the HPC Head Node ·        enable the services for Windows HPC Server  to authenticate properly with the Windows Azure subscription. ·        use the Certificates MMC snap-in to import the certificate to the Trusted Root Certification Authorities store of the local computer account. The certificate must be in PFX format (.pfx or .p12 file) with a private key that is protected by a password. ·        see Certificates (http://go.microsoft.com/fwlink/?LinkId=163918). ·        To open the certificates snapin: Run > mmc. File > Add/Remove Snap-in > certificates > Computer account > Local Computer ·        To import the certificate via wizard - Certificates > Trusted Root Certification Authorities > Certificates > All Tasks > Import ·        After the certificate is imported, it appears in the details pane in the Certificates snap-in. You can open the certificate to check its status. Configure a Proxy Client on the HPC Head Node ·        the following Windows HPC Server services must be able to communicate over the Internet (through the firewall) with the services for Windows Azure: HPCManagement, HPCScheduler, HPCBrokerWorker. ·        Create a Windows Azure Worker Node Template ·        Edit HPC node templates in HPC Node Template Editor. ·        Specify: 1) Windows Azure subscription connection info (unique service name) for adding a set of worker nodes to the cluster + 2)worker node availability policy – rules for deploying / removing worker role instances in Windows Azure o   HPC Cluster Manager > Configuration > Navigation Pane > Node Templates > Actions pane > New à Create Node Template Wizard or Edit à Node Template Editor o   Choose Node Template Type page - Windows Azure worker node template o   Specify Template Name page – template name & description o   Provide Connection Information page – Azure Subscription ID (text) & Subscription certificate (browse) o   Provide Service Information page - Azure service name + blob storage account name (optionally click Retrieve Connection Information to get list of available from azure – possible LRT). o   Configure Azure Availability Policy page - how Windows Azure worker nodes start / stop (online / offline the worker role instance -  add / remove) – manual / automatic o   for automatic - In the Configure Windows Azure Worker Availability Policy dialog -select days and hours for worker nodes to start / stop. ·        To validate the Windows Azure connection information, on the template's Connection Information tab > Validate connection information. ·        You can upload a file package to the storage account that is specified in the template - eg upload application or service files that will run on the worker nodes. see hpcpack (http://go.microsoft.com/fwlink/?LinkID=205514). Add Azure Worker Nodes to the HPC Cluster ·        Use the Add Node Wizard – specify: 1) the worker node template, 2) The number of worker nodes   (within the quota of role instances in the azure subscription), and 3)           The VM size of the worker nodes : ExtraSmall, Small, Medium, Large, or ExtraLarge.  ·        to add worker nodes of different sizes, must run the Add Node Wizard separately for each size. ·        All worker nodes that are added to the cluster by using a specific worker node template define a set of worker nodes that will be deployed and managed together in Windows Azure when you start the nodes. This includes worker nodes that you add later by using the worker node template and, if you choose, worker nodes of different sizes. You cannot start, stop, or delete individual worker nodes. ·        To add Windows Azure worker nodes o   In HPC Cluster Manager: Node Management > Actions pane > Add Node à Add Node Wizard o   Select Deployment Method page - Add Azure Worker nodes o   Specify New Nodes page - select a worker node template, specify the number and size of the worker nodes ·        After you add worker nodes to the cluster, they are in the Not-Deployed state, and they have a health state of Unapproved. Before you can use the worker nodes to run jobs, you must start them and then bring them online. ·        Worker nodes are numbered consecutively in a naming series that begins with the root name AzureCN – this is non-configurable. Deploying Windows Azure Worker Nodes ·        To deploy the role instances in Windows Azure - start the worker nodes added to the HPC cluster and bring the nodes online so that they are available to run cluster jobs. This can be configured in the HPC Azure Worker Node Template – Azure Availability Policy -  to be automatic or manual. ·        The Start, Stop, and Delete actions take place on the set of worker nodes that are configured by a specific worker node template. You cannot perform one of these actions on a single worker node in a set. You also cannot perform a single action on two sets of worker nodes (specified by two different worker node templates). ·        ·          Starting a set of worker nodes deploys a set of worker role instances in Windows Azure, which can take some time to complete, depending on the number of worker nodes and the performance of Windows Azure. ·        To start worker nodes manually and bring them online o   In HPC Node Management > Navigation Pane > Nodes > List / Heat Map view - select one or more worker nodes. o   Actions pane > Start – in the Start Azure Worker Nodes dialog, select a node template. o   the state of the worker nodes changes from Not Deployed to track the provisioning progress – worker node Details Pane > Provisioning Log tab. o   If there were errors during the provisioning of one or more worker nodes, the state of those nodes is set to Unknown and the node health is set to Unapproved. To determine the reason for the failure, review the provisioning logs for the nodes. o   After a worker node starts successfully, the node state changes to Offline. To bring the nodes online, select the nodes that are in the Offline state > Bring Online. ·        Troubleshooting o   check node template. o   use telnet to test connectivity: telnet <ServiceName>.cloudapp.net 7999 o   check node status - Deployment status information appears in the service account information in the Windows Azure Portal - HPC queries this -  see  node status information for any failed nodes in HPC Node Management. ·        When role instances are deployed, file packages that were previously uploaded to the storage account using the hpcpack command are automatically installed. You can also upload file packages to storage after the worker nodes are started, and then manually install them on the worker nodes. see hpcpack (http://go.microsoft.com/fwlink/?LinkID=205514). ·        to remove a set of role instances in Windows Azure - stop the nodes by using HPC Cluster Manager (apply the Stop action). This deletes the role instances from the service and changes the state of the worker nodes in the HPC cluster to Not Deployed. ·        Each time that you start a set of worker nodes, two proxy role instances (size Small) are configured in Windows Azure to facilitate communication between HPC Cluster Manager and the worker nodes. The proxy role instances are not listed in HPC Cluster Manager after the worker nodes are added. However, the instances appear in the Windows Azure Portal. The proxy role instances incur charges in Windows Azure along with the worker node instances, and they count toward the quota of role instances in the subscription.

    Read the article

  • ASPNET WebAPI REST Guidance

    - by JoshReuben
    ASP.NET Web API is an ideal platform for building RESTful applications on the .NET Framework. While I may be more partial to NodeJS these days, there is no denying that WebAPI is a well engineered framework. What follows is my investigation of how to leverage WebAPI to construct a RESTful frontend API.   The Advantages of REST Methodology over SOAP Simpler API for CRUD ops Standardize Development methodology - consistent and intuitive Standards based à client interop Wide industry adoption, Ease of use à easy to add new devs Avoid service method signature blowout Smaller payloads than SOAP Stateless à no session data means multi-tenant scalability Cache-ability Testability   General RESTful API Design Overview · utilize HTTP Protocol - Usage of HTTP methods for CRUD, standard HTTP response codes, common HTTP headers and Mime Types · Resources are mapped to URLs, actions are mapped to verbs and the rest goes in the headers. · keep the API semantic, resource-centric – A RESTful, resource-oriented service exposes a URI for every piece of data the client might want to operate on. A REST-RPC Hybrid exposes a URI for every operation the client might perform: one URI to fetch a piece of data, a different URI to delete that same data. utilize Uri to specify CRUD op, version, language, output format: http://api.MyApp.com/{ver}/{lang}/{resource_type}/{resource_id}.{output_format}?{key&filters} · entity CRUD operations are matched to HTTP methods: · Create - POST / PUT · Read – GET - cacheable · Update – PUT · Delete - DELETE · Use Uris to represent a hierarchies - Resources in RESTful URLs are often chained · Statelessness allows for idempotency – apply an op multiple times without changing the result. POST is non-idempotent, the rest are idempotent (if DELETE flags records instead of deleting them). · Cache indication - Leverage HTTP headers to label cacheable content and indicate the permitted duration of cache · PUT vs POST - The client uses PUT when it determines which URI (Id key) the new resource should have. The client uses POST when the server determines they key. PUT takes a second param – the id. POST creates a new resource. The server assigns the URI for the new object and returns this URI as part of the response message. Note: The PUT method replaces the entire entity. That is, the client is expected to send a complete representation of the updated product. If you want to support partial updates, the PATCH method is preferred DELETE deletes a resource at a specified URI – typically takes an id param · Leverage Common HTTP Response Codes in response headers 200 OK: Success 201 Created - Used on POST request when creating a new resource. 304 Not Modified: no new data to return. 400 Bad Request: Invalid Request. 401 Unauthorized: Authentication. 403 Forbidden: Authorization 404 Not Found – entity does not exist. 406 Not Acceptable – bad params. 409 Conflict - For POST / PUT requests if the resource already exists. 500 Internal Server Error 503 Service Unavailable · Leverage uncommon HTTP Verbs to reduce payload sizes HEAD - retrieves just the resource meta-information. OPTIONS returns the actions supported for the specified resource. PATCH - partial modification of a resource. · When using PUT, POST or PATCH, send the data as a document in the body of the request. Don't use query parameters to alter state. · Utilize Headers for content negotiation, caching, authorization, throttling o Content Negotiation – choose representation (e.g. JSON or XML and version), language & compression. Signal via RequestHeader.Accept & ResponseHeader.Content-Type Accept: application/json;version=1.0 Accept-Language: en-US Accept-Charset: UTF-8 Accept-Encoding: gzip o Caching - ResponseHeader: Expires (absolute expiry time) or Cache-Control (relative expiry time) o Authorization - basic HTTP authentication uses the RequestHeader.Authorization to specify a base64 encoded string "username:password". can be used in combination with SSL/TLS (HTTPS) and leverage OAuth2 3rd party token-claims authorization. Authorization: Basic sQJlaTp5ZWFslylnaNZ= o Rate Limiting - Not currently part of HTTP so specify non-standard headers prefixed with X- in the ResponseHeader. X-RateLimit-Limit: 10000 X-RateLimit-Remaining: 9990 · HATEOAS Methodology - Hypermedia As The Engine Of Application State – leverage API as a state machine where resources are states and the transitions between states are links between resources and are included in their representation (hypermedia) – get API metadata signatures from the response Link header - in a truly REST based architecture any URL, except the initial URL, can be changed, even to other servers, without worrying about the client. · error responses - Do not just send back a 200 OK with every response. Response should consist of HTTP error status code (JQuery has automated support for this), A human readable message , A Link to a meaningful state transition , & the original data payload that was problematic. · the URIs will typically map to a server-side controller and a method name specified by the type of request method. Stuff all your calls into just four methods is not as crazy as it sounds. · Scoping - Path variables look like you’re traversing a hierarchy, and query variables look like you’re passing arguments into an algorithm · Mapping URIs to Controllers - have one controller for each resource is not a rule – can consolidate - route requests to the appropriate controller and action method · Keep URls Consistent - Sometimes it’s tempting to just shorten our URIs. not recommend this as this can cause confusion · Join Naming – for m-m entity relations there may be multiple hierarchy traversal paths · Routing – useful level of indirection for versioning, server backend mocking in development ASPNET WebAPI Considerations ASPNET WebAPI implements a lot (but not all) RESTful API design considerations as part of its infrastructure and via its coding convention. Overview When developing an API there are basically three main steps: 1. Plan out your URIs 2. Setup return values and response codes for your URIs 3. Implement a framework for your API.   Design · Leverage Models MVC folder · Repositories – support IoC for tests, abstraction · Create DTO classes – a level of indirection decouples & allows swap out · Self links can be generated using the UrlHelper · Use IQueryable to support projections across the wire · Models can support restful navigation properties – ICollection<T> · async mechanism for long running ops - return a response with a ticket – the client can then poll or be pushed the final result later. · Design for testability - Test using HttpClient , JQuery ( $.getJSON , $.each) , fiddler, browser debug. Leverage IDependencyResolver – IoC wrapper for mocking · Easy debugging - IE F12 developer tools: Network tab, Request Headers tab     Routing · HTTP request method is matched to the method name. (This rule applies only to GET, POST, PUT, and DELETE requests.) · {id}, if present, is matched to a method parameter named id. · Query parameters are matched to parameter names when possible · Done in config via Routes.MapHttpRoute – similar to MVC routing · Can alternatively: o decorate controller action methods with HttpDelete, HttpGet, HttpHead,HttpOptions, HttpPatch, HttpPost, or HttpPut., + the ActionAttribute o use AcceptVerbsAttribute to support other HTTP verbs: e.g. PATCH, HEAD o use NonActionAttribute to prevent a method from getting invoked as an action · route table Uris can support placeholders (via curly braces{}) – these can support default values and constraints, and optional values · The framework selects the first route in the route table that matches the URI. Response customization · Response code: By default, the Web API framework sets the response status code to 200 (OK). But according to the HTTP/1.1 protocol, when a POST request results in the creation of a resource, the server should reply with status 201 (Created). Non Get methods should return HttpResponseMessage · Location: When the server creates a resource, it should include the URI of the new resource in the Location header of the response. public HttpResponseMessage PostProduct(Product item) {     item = repository.Add(item);     var response = Request.CreateResponse<Product>(HttpStatusCode.Created, item);     string uri = Url.Link("DefaultApi", new { id = item.Id });     response.Headers.Location = new Uri(uri);     return response; } Validation · Decorate Models / DTOs with System.ComponentModel.DataAnnotations properties RequiredAttribute, RangeAttribute. · Check payloads using ModelState.IsValid · Under posting – leave out values in JSON payload à JSON formatter assigns a default value. Use with RequiredAttribute · Over-posting - if model has RO properties à use DTO instead of model · Can hook into pipeline by deriving from ActionFilterAttribute & overriding OnActionExecuting Config · Done in App_Start folder > WebApiConfig.cs – static Register method: HttpConfiguration param: The HttpConfiguration object contains the following members. Member Description DependencyResolver Enables dependency injection for controllers. Filters Action filters – e.g. exception filters. Formatters Media-type formatters. by default contains JsonFormatter, XmlFormatter IncludeErrorDetailPolicy Specifies whether the server should include error details, such as exception messages and stack traces, in HTTP response messages. Initializer A function that performs final initialization of the HttpConfiguration. MessageHandlers HTTP message handlers - plug into pipeline ParameterBindingRules A collection of rules for binding parameters on controller actions. Properties A generic property bag. Routes The collection of routes. Services The collection of services. · Configure JsonFormatter for circular references to support links: PreserveReferencesHandling.Objects Documentation generation · create a help page for a web API, by using the ApiExplorer class. · The ApiExplorer class provides descriptive information about the APIs exposed by a web API as an ApiDescription collection · create the help page as an MVC view public ILookup<string, ApiDescription> GetApis()         {             return _explorer.ApiDescriptions.ToLookup(                 api => api.ActionDescriptor.ControllerDescriptor.ControllerName); · provide documentation for your APIs by implementing the IDocumentationProvider interface. Documentation strings can come from any source that you like – e.g. extract XML comments or define custom attributes to apply to the controller [ApiDoc("Gets a product by ID.")] [ApiParameterDoc("id", "The ID of the product.")] public HttpResponseMessage Get(int id) · GlobalConfiguration.Configuration.Services – add the documentation Provider · To hide an API from the ApiExplorer, add the ApiExplorerSettingsAttribute Plugging into the Message Handler pipeline · Plug into request / response pipeline – derive from DelegatingHandler and override theSendAsync method – e.g. for logging error codes, adding a custom response header · Can be applied globally or to a specific route Exception Handling · Throw HttpResponseException on method failures – specify HttpStatusCode enum value – examine this enum, as its values map well to typical op problems · Exception filters – derive from ExceptionFilterAttribute & override OnException. Apply on Controller or action methods, or add to global HttpConfiguration.Filters collection · HttpError object provides a consistent way to return error information in the HttpResponseException response body. · For model validation, you can pass the model state to CreateErrorResponse, to include the validation errors in the response public HttpResponseMessage PostProduct(Product item) {     if (!ModelState.IsValid)     {         return Request.CreateErrorResponse(HttpStatusCode.BadRequest, ModelState); Cookie Management · Cookie header in request and Set-Cookie headers in a response - Collection of CookieState objects · Specify Expiry, max-age resp.Headers.AddCookies(new CookieHeaderValue[] { cookie }); Internet Media Types, formatters and serialization · Defaults to application/json · Request Accept header and response Content-Type header · determines how Web API serializes and deserializes the HTTP message body. There is built-in support for XML, JSON, and form-urlencoded data · customizable formatters can be inserted into the pipeline · POCO serialization is opt out via JsonIgnoreAttribute, or use DataMemberAttribute for optin · JSON serializer leverages NewtonSoft Json.NET · loosely structured JSON objects are serialzed as JObject which derives from Dynamic · to handle circular references in json: json.SerializerSettings.PreserveReferencesHandling =    PreserveReferencesHandling.All à {"$ref":"1"}. · To preserve object references in XML [DataContract(IsReference=true)] · Content negotiation Accept: Which media types are acceptable for the response, such as “application/json,” “application/xml,” or a custom media type such as "application/vnd.example+xml" Accept-Charset: Which character sets are acceptable, such as UTF-8 or ISO 8859-1. Accept-Encoding: Which content encodings are acceptable, such as gzip. Accept-Language: The preferred natural language, such as “en-us”. o Web API uses the Accept and Accept-Charset headers. (At this time, there is no built-in support for Accept-Encoding or Accept-Language.) · Controller methods can take JSON representations of DTOs as params – auto-deserialization · Typical JQuery GET request: function find() {     var id = $('#prodId').val();     $.getJSON("api/products/" + id,         function (data) {             var str = data.Name + ': $' + data.Price;             $('#product').text(str);         })     .fail(         function (jqXHR, textStatus, err) {             $('#product').text('Error: ' + err);         }); }            · Typical GET response: HTTP/1.1 200 OK Server: ASP.NET Development Server/10.0.0.0 Date: Mon, 18 Jun 2012 04:30:33 GMT X-AspNet-Version: 4.0.30319 Cache-Control: no-cache Pragma: no-cache Expires: -1 Content-Type: application/json; charset=utf-8 Content-Length: 175 Connection: Close [{"Id":1,"Name":"TomatoSoup","Price":1.39,"ActualCost":0.99},{"Id":2,"Name":"Hammer", "Price":16.99,"ActualCost":10.00},{"Id":3,"Name":"Yo yo","Price":6.99,"ActualCost": 2.05}] True OData support · Leverage Query Options $filter, $orderby, $top and $skip to shape the results of controller actions annotated with the [Queryable]attribute. [Queryable]  public IQueryable<Supplier> GetSuppliers()  · Query: ~/Suppliers?$filter=Name eq ‘Microsoft’ · Applies the following selection filter on the server: GetSuppliers().Where(s => s.Name == “Microsoft”)  · Will pass the result to the formatter. · true support for the OData format is still limited - no support for creates, updates, deletes, $metadata and code generation etc · vnext: ability to configure how EditLinks, SelfLinks and Ids are generated Self Hosting no dependency on ASPNET or IIS: using (var server = new HttpSelfHostServer(config)) {     server.OpenAsync().Wait(); Tracing · tracability tools, metrics – e.g. send to nagios · use your choice of tracing/logging library, whether that is ETW,NLog, log4net, or simply System.Diagnostics.Trace. · To collect traces, implement the ITraceWriter interface public class SimpleTracer : ITraceWriter {     public void Trace(HttpRequestMessage request, string category, TraceLevel level,         Action<TraceRecord> traceAction)     {         TraceRecord rec = new TraceRecord(request, category, level);         traceAction(rec);         WriteTrace(rec); · register the service with config · programmatically trace – has helper extension methods: Configuration.Services.GetTraceWriter().Info( · Performance tracing - pipeline writes traces at the beginning and end of an operation - TraceRecord class includes aTimeStamp property, Kind property set to TraceKind.Begin / End Security · Roles class methods: RoleExists, AddUserToRole · WebSecurity class methods: UserExists, .CreateUserAndAccount · Request.IsAuthenticated · Leverage HTTP 401 (Unauthorized) response · [AuthorizeAttribute(Roles="Administrator")] – can be applied to Controller or its action methods · See section in WebApi document on "Claim-based-security for ASP.NET Web APIs using DotNetOpenAuth" – adapt this to STS.--> Web API Host exposes secured Web APIs which can only be accessed by presenting a valid token issued by the trusted issuer. http://zamd.net/2012/05/04/claim-based-security-for-asp-net-web-apis-using-dotnetopenauth/ · Use MVC membership provider infrastructure and add a DelegatingHandler child class to the WebAPI pipeline - http://stackoverflow.com/questions/11535075/asp-net-mvc-4-web-api-authentication-with-membership-provider - this will perform the login actions · Then use AuthorizeAttribute on controllers and methods for role mapping- http://sixgun.wordpress.com/2012/02/29/asp-net-web-api-basic-authentication/ · Alternate option here is to rely on MVC App : http://forums.asp.net/t/1831767.aspx/1

    Read the article

  • Mathematica Programming Language&ndash;An Introduction

    - by JoshReuben
    The Mathematica http://www.wolfram.com/mathematica/ programming model consists of a kernel computation engine (or grid of such engines) and a front-end of notebook instances that communicate with the kernel throughout a session. The programming model of Mathematica is incredibly rich & powerful – besides numeric calculations, it supports symbols (eg Pi, I, E) and control flow logic.   obviously I could use this as a simple calculator: 5 * 10 --> 50 but this language is much more than that!   for example, I could use control flow logic & setup a simple infinite loop: x=1; While [x>0, x=x,x+1] Different brackets have different purposes: square brackets for function arguments:  Cos[x] round brackets for grouping: (1+2)*3 curly brackets for lists: {1,2,3,4} The power of Mathematica (as opposed to say Matlab) is that it gives exact symbolic answers instead of a rounded numeric approximation (unless you request it):   Mathematica lets you define scoped variables (symbols): a=1; b=2; c=a+b --> 5 these variables can contain symbolic values – you can think of these as partially computed functions:   use Clear[x] or Remove[x] to zero or dereference a variable.   To compute a numerical approximation to n significant digits (default n=6), use N[x,n] or the //N prefix: Pi //N -->3.14159 N[Pi,50] --> 3.1415926535897932384626433832795028841971693993751 The kernel uses % to reference the lastcalculation result, %% the 2nd last, %%% the 3rd last etc –> clearer statements: eg instead of: Sqrt[Pi+Sqrt[Sqrt[Pi+Sqrt[Pi]]] do: Sqrt[Pi]; Sqrt[Pi+%]; Sqrt[Pi+%] The help system supports wildcards, so I can search for functions like so: ?Inv* Mathematica supports some very powerful programming constructs and a rich function library that allow you to do things that you would have to write allot of code for in a language like C++.   the Factor function – factorization: Factor[x^3 – 6*x^2 +11x – 6] --> (-3+x) (-2+x) (-1+x)   the Solve function – find the roots of an equation: Solve[x^3 – 2x + 1 == 0] -->   the Expand function – express (1+x)^10 in polynomial form: Expand[(1+x)^10] --> 1+10x+45x^2+120x^3+210x^4+252x^5+210x^6+120x^7+45x^8+10x^9+x^10 the Prime function – what is the 1000th prime? Prime[1000] -->7919 Mathematica also has some powerful graphics capabilities:   the Plot function – plot the graph of y=Sin x in a single period: Plot[Sin[x], {x,0,2*Pi}] you can also plot 3D surfaces of functions using Plot3D function

    Read the article

  • Extreme Optimization Numerical Libraries for .NET – Part 1 of n

    - by JoshReuben
    While many of my colleagues are fascinated in constructing the ultimate ViewModel or ServiceBus, I feel that this kind of plumbing code is re-invented far too many times – at some point in the near future, it will be out of the box standard infra. How many times have you been to a customer site and built a different variation of the same kind of code frameworks? How many times can you abstract Prism or reliable and discoverable WCF communication? As the bar is raised for whats bundled with the framework and more tasks become declarative, automated and configurable, Information Systems will expose a higher level of abstraction, forcing software engineers to focus on more advanced computer science and algorithmic tasks. I've spent the better half of the past decade building skills in .NET and expanding my mathematical horizons by working through the Schaums guides. In this series I am going to examine how these skillsets come together in the implementation provided by ExtremeOptimization. Download the trial version here: http://www.extremeoptimization.com/downloads.aspx Overview The library implements a set of algorithms for: linear algebra, complex numbers, numerical integration and differentiation, solving equations, optimization, random numbers, regression, ANOVA, statistical distributions, hypothesis tests. EONumLib combines three libraries in one - organized in a consistent namespace hierarchy. Mathematics Library - Extreme.Mathematics namespace Vector and Matrix Library - Extreme.Mathematics.LinearAlgebra namespace Statistics Library - Extreme.Statistics namespace System Requirements -.NET framework 4.0  Mathematics Library The classes are organized into the following namespace hierarchy: Extreme.Mathematics – common data types, exception types, and delegates. Extreme.Mathematics.Calculus - numerical integration and differentiation of functions. Extreme.Mathematics.Curves - points, lines and curves, including polynomials and Chebyshev approximations. curve fitting and interpolation. Extreme.Mathematics.Generic - generic arithmetic & linear algebra. Extreme.Mathematics.EquationSolvers - root finding algorithms. Extreme.Mathematics.LinearAlgebra - vectors , matrices , matrix decompositions, solvers for simultaneous linear equations and least squares. Extreme.Mathematics.Optimization – multi-d function optimization + linear programming. Extreme.Mathematics.SignalProcessing - one and two-dimensional discrete Fourier transforms. Extreme.Mathematics.SpecialFunctions

    Read the article

  • Extreme Optimization – Curves (Function Mapping) Part 1

    - by JoshReuben
    Overview ·        a curve is a functional map relationship between two factors (i.e. a function - However, the word function is a reserved word). ·        You can use the EO API to create common types of functions, find zeroes and calculate derivatives - currently supports constants, lines, quadratic curves, polynomials and Chebyshev approximations. ·        A function basis is a set of functions that can be combined to form a particular class of functions.   The Curve class ·        the abstract base class from which all other curve classes are derived – it provides the following methods: ·        ValueAt(Double) - evaluates the curve at a specific point. ·        SlopeAt(Double) - evaluates the derivative ·        Integral(Double, Double) - evaluates the definite integral over a specified interval. ·        TangentAt(Double) - returns a Line curve that is the tangent to the curve at a specific point. ·        FindRoots() - attempts to find all the roots or zeroes of the curve. ·        A particular type of curve is defined by a Parameters property, of type ParameterCollection   The GeneralCurve class ·        defines a curve whose value and, optionally, derivative and integrals, are calculated using arbitrary methods. A general curve has no parameters. ·        Constructor params:  RealFunction delegates – 1 for the function, and optionally another 2 for the derivative and integral ·        If no derivative  or integral function is supplied, they are calculated via the NumericalDifferentiation  and AdaptiveIntegrator classes in the Extreme.Mathematics.Calculus namespace. // the function is 1/(1+x^2) private double f(double x) {     return 1 / (1 + x*x); }   // Its derivative is -2x/(1+x^2)^2 private double df(double x) {     double y = 1 + x*x;     return -2*x* / (y*y); }   // The integral of f is Arctan(x), which is available from the Math class. var c1 = new GeneralCurve (new RealFunction(f), new RealFunction(df), new RealFunction(System.Math.Atan)); // Find the tangent to this curve at x=1 (the Line class is derived from Curve) Line l1 = c1.TangentAt(1);

    Read the article

  • Real World Nuget

    - by JoshReuben
    Why Nuget A higher level of granularity for managing references When you have solutions of many projects that depend on solutions of many projects etc à escape from Solution Hell. Links · Using A GUI (Package Explorer) to build packages - http://docs.nuget.org/docs/creating-packages/using-a-gui-to-build-packages · Creating a Nuspec File - http://msdn.microsoft.com/en-us/vs2010trainingcourse_aspnetmvcnuget_topic2.aspx · consuming a Nuget Package - http://msdn.microsoft.com/en-us/vs2010trainingcourse_aspnetmvcnuget_topic3 · Nuspec reference - http://docs.nuget.org/docs/reference/nuspec-reference · updating packages - http://nuget.codeplex.com/wikipage?title=Updating%20All%20Packages · versioning - http://docs.nuget.org/docs/reference/versioning POC Folder Structure POC Setup Steps · Install package explorer · Source o Create a source solution – configure output directory for projects (Project > Properties > Build > Output Path) · Package o Add assemblies to package from output directory (D&D)- add net folder o File > Export – save .nuspec files and lib contents <?xml version="1.0" encoding="utf-16"?> <package > <metadata> <id>MyPackage</id> <version>1.0.0.3</version> <title /> <authors>josh-r</authors> <owners /> <requireLicenseAcceptance>false</requireLicenseAcceptance> <description>My package description.</description> <summary /> </metadata> </package> o File > Save – saves .nupkg file · Create Target Solution o In Tools > Options: Configure package source & Add package Select projects: Output from package manager (powershell console) ------- Installing...MyPackage 1.0.0 ------- Added file 'NugetSource.AssemblyA.dll' to folder 'MyPackage.1.0.0\lib'. Added file 'NugetSource.AssemblyA.pdb' to folder 'MyPackage.1.0.0\lib'. Added file 'NugetSource.AssemblyB.dll' to folder 'MyPackage.1.0.0\lib'. Added file 'NugetSource.AssemblyB.pdb' to folder 'MyPackage.1.0.0\lib'. Added file 'MyPackage.1.0.0.nupkg' to folder 'MyPackage.1.0.0'. Successfully installed 'MyPackage 1.0.0'. Added reference 'NugetSource.AssemblyA' to project 'AssemblyX' Added reference 'NugetSource.AssemblyB' to project 'AssemblyX' Added file 'packages.config'. Added file 'packages.config' to project 'AssemblyX' Added file 'repositories.config'. Successfully added 'MyPackage 1.0.0' to AssemblyX. ============================== o Packages folder created at solution level o Packages.config file generated in each project: <?xml version="1.0" encoding="utf-8"?> <packages>   <package id="MyPackage" version="1.0.0" targetFramework="net40" /> </packages> A local Packages folder is created for package versions installed: Each folder contains the downloaded .nupkg file and its unpacked contents – eg of dlls that the project references Note: this folder is not checked in UpdatePackages o Configure Package Manager to automatically check for updates o Browse packages - It automatically picked up the updates Update Procedure · Modify source · Change source version in assembly info · Build source · Open last package in package explorer · Increment package version number and re-add assemblies · Save package with new version number and export its definition · In target solution – Tools > Manage Nuget Packages – click on All to trigger refresh , then click on recent packages to see updates · If problematic, delete packages folder Versioning uninstall-package mypackage install-package mypackage –version 1.0.0.3 uninstall-package mypackage install-package mypackage –version 1.0.0.4 Dependencies · <?xml version="1.0" encoding="utf-16"?> <package xmlns="http://schemas.microsoft.com/packaging/2012/06/nuspec.xsd"> <metadata> <id>MyDependentPackage</id> <version>1.0.0</version> <title /> <authors>josh-r</authors> <owners /> <requireLicenseAcceptance>false</requireLicenseAcceptance> <description>My package description.</description> <dependencies> <group targetFramework=".NETFramework4.0"> <dependency id="MyPackage" version="1.0.0.4" /> </group> </dependencies> </metadata> </package> Using NuGet without committing packages to source control http://docs.nuget.org/docs/workflows/using-nuget-without-committing-packages Right click on the Solution node in Solution Explorer and select Enable NuGet Package Restore. — Recall that packages folder is not part of solution If you get downloading package ‘Nuget.build’ failed, config proxy to support certificate for https://nuget.org/api/v2/ & allow unrestricted access to packages.nuget.org To test connectivity: get-package –listavailable To test Nuget Package Restore – delete packages folder and open vs as admin. In nugget msbuild: <Import Project="$(SolutionDir)\.nuget\nuget.targets" /> TFSBuild Integration Modify Nuget.Targets file <RestorePackages Condition="  '$(RestorePackages)' == '' "> True </RestorePackages> … <PackageSource Include="\\IL-CV-004-W7D\Packages" /> Add System Environment variable EnableNuGetPackageRestore=true & restart the “visual studio team foundation build service host” service. Important: Ensure Network Service has access to Packages folder Nugetter TFS Build integration Add Nugetter build process templates to TFS source control For Build Controller - Specify location of custom assemblies Generate .nuspec file from Package Explorer: File > Export Edit the file elements – remove path info from src and target attributes <?xml version="1.0" encoding="utf-16"?> <package xmlns="http://schemas.microsoft.com/packaging/2012/06/nuspec.xsd">     <metadata>         <id>Common</id>         <version>1.0.0</version>         <title />         <authors>josh-r</authors>         <owners />         <requireLicenseAcceptance>false</requireLicenseAcceptance>         <description>My package description.</description>         <dependencies>             <group targetFramework=".NETFramework3.5" />         </dependencies>     </metadata>     <files>         <file src="CommonTypes.dll" target="CommonTypes.dll" />         <file src="CommonTypes.pdb" target="CommonTypes.pdb" /> … Add .nuspec file to solution so that it is available for build: Dev\NovaNuget\Common\NuSpec\common.1.0.0.nuspec Add a Build Process Definition based on the Nugetter build process template: Configure the build process – specify: · .sln to build · Base path (output directory) · Nuget.exe file path · .nuspec file path Copy DLLs to a binary folder 1) Set copy local for an assembly reference to false 2)  MSBuild Copy Task – modify .csproj file: http://msdn.microsoft.com/en-us/library/3e54c37h.aspx <ItemGroup>     <MySourceFiles Include="$(MSBuildProjectDirectory)\..\SourceAssemblies\**\*.*" />   </ItemGroup>     <Target Name="BeforeBuild">     <Copy SourceFiles="@(MySourceFiles)" DestinationFolder="bin\debug\SourceAssemblies" />   </Target> 3) Set Probing assembly search path from app.config - http://msdn.microsoft.com/en-us/library/823z9h8w(v=vs.80).aspx -                 <?xml version="1.0" encoding="utf-8" ?> <configuration>   <runtime>     <assemblyBinding xmlns="urn:schemas-microsoft-com:asm.v1">       <probing privatePath="SourceAssemblies"/>     </assemblyBinding>   </runtime> </configuration> Forcing 'copy local = false' The following generic powershell script was added to the packages install.ps1: param($installPath, $toolsPath, $package, $project) if( $project.Object.Project.Name -ne "CopyPackages") { $asms = $package.AssemblyReferences | %{$_.Name} foreach ($reference in $project.Object.References) { if ($asms -contains $reference.Name + ".dll") { $reference.CopyLocal = $false; } } } An empty project named "CopyPackages" was added to the solution - it references all the packages and is the only one set to CopyLocal="true". No MSBuild knowledge required.

    Read the article

  • Excel Solver vs Solver Foundation

    - by JoshReuben
    I recently read a book http://www.amazon.com/Scientific-Engineering-Cookbook-Cookbooks-OReilly/dp/0596008791/ref=sr_1_1?ie=UTF8&s=books&qid=1296593374&sr=8-1 - the Excel Scientific and Engineering Cookbook.     The 2 main tools that this book leveraged were the Data Analysis Pack and Excel Solver. I had previously been aquanted with Microsoft Solver Foundation - this is a full fledged API for solving optimization problems, and went beyond being a mere Excel plugin - it exposed a C# programmatic interface for in process and a web service interface for out of process integration. were they the same? apparently not!   2 different solver frameworks for Excel: http://www.solver.com/index.html http://www.solverfoundation.com/ I contacted both vendors to get their perspectives.   Heres what the Excel Solver guys had to say:   "The Solver Foundation requires you to learn and use a very specific modeling language (OML). The Excel solver allows you to formulate your optimization problems without learning any new language simply by entering the formulas into cells on the Excel spreadsheet, something that nearly everyone is already familiar with doing.   The Excel Solver also allows you to seamlessly upgrade to products that combine Monte Carlo Simulation capabilities (our Risk Solver Premium and Risk Solver Platform products) which allow you to include uncertainty into your models when appropriate.   Our advanced Excel Solver Products also have a number of built in reporting tools for advanced analysis of the your model and it's results"           And Heres what the Microsoft Solver Foundation guys had to say:   "  With the release of Solver Foundation 3.0, Solver Foundation has the same kinds of solvers (plus a few more) than what is found in Excel Solver. I think there are two main differences:   1.      Problems are described differently. In Excel Solver the goals and constraints are specified inside the spreadsheet, in formulas. In Solver Foundation they are described either in .Net code that uses the Solver Foundation Services API, or using the OML modeling language in Excel. 2.      Solver Foundation’s primary strength is on solving large linear, mixed integer, and constraint models. That is, models that contain arbitrary nonlinear functions (such as trig functions, IF(), powers, etc) are handled a bit better by the Excel Solver at this point. "

    Read the article

  • MapReduce in DryadLINQ and PLINQ

    - by JoshReuben
    MapReduce See http://en.wikipedia.org/wiki/Mapreduce The MapReduce pattern aims to handle large-scale computations across a cluster of servers, often involving massive amounts of data. "The computation takes a set of input key/value pairs, and produces a set of output key/value pairs. The developer expresses the computation as two Func delegates: Map and Reduce. Map - takes a single input pair and produces a set of intermediate key/value pairs. The MapReduce function groups results by key and passes them to the Reduce function. Reduce - accepts an intermediate key I and a set of values for that key. It merges together these values to form a possibly smaller set of values. Typically just zero or one output value is produced per Reduce invocation. The intermediate values are supplied to the user's Reduce function via an iterator." the canonical MapReduce example: counting word frequency in a text file.     MapReduce using DryadLINQ see http://research.microsoft.com/en-us/projects/dryadlinq/ and http://connect.microsoft.com/Dryad DryadLINQ provides a simple and straightforward way to implement MapReduce operations. This The implementation has two primary components: A Pair structure, which serves as a data container. A MapReduce method, which counts word frequency and returns the top five words. The Pair Structure - Pair has two properties: Word is a string that holds a word or key. Count is an int that holds the word count. The structure also overrides ToString to simplify printing the results. The following example shows the Pair implementation. public struct Pair { private string word; private int count; public Pair(string w, int c) { word = w; count = c; } public int Count { get { return count; } } public string Word { get { return word; } } public override string ToString() { return word + ":" + count.ToString(); } } The MapReduce function  that gets the results. the input data could be partitioned and distributed across the cluster. 1. Creates a DryadTable<LineRecord> object, inputTable, to represent the lines of input text. For partitioned data, use GetPartitionedTable<T> instead of GetTable<T> and pass the method a metadata file. 2. Applies the SelectMany operator to inputTable to transform the collection of lines into collection of words. The String.Split method converts the line into a collection of words. SelectMany concatenates the collections created by Split into a single IQueryable<string> collection named words, which represents all the words in the file. 3. Performs the Map part of the operation by applying GroupBy to the words object. The GroupBy operation groups elements with the same key, which is defined by the selector delegate. This creates a higher order collection, whose elements are groups. In this case, the delegate is an identity function, so the key is the word itself and the operation creates a groups collection that consists of groups of identical words. 4. Performs the Reduce part of the operation by applying Select to groups. This operation reduces the groups of words from Step 3 to an IQueryable<Pair> collection named counts that represents the unique words in the file and how many instances there are of each word. Each key value in groups represents a unique word, so Select creates one Pair object for each unique word. IGrouping.Count returns the number of items in the group, so each Pair object's Count member is set to the number of instances of the word. 5. Applies OrderByDescending to counts. This operation sorts the input collection in descending order of frequency and creates an ordered collection named ordered. 6. Applies Take to ordered to create an IQueryable<Pair> collection named top, which contains the 100 most common words in the input file, and their frequency. Test then uses the Pair object's ToString implementation to print the top one hundred words, and their frequency.   public static IQueryable<Pair> MapReduce( string directory, string fileName, int k) { DryadDataContext ddc = new DryadDataContext("file://" + directory); DryadTable<LineRecord> inputTable = ddc.GetTable<LineRecord>(fileName); IQueryable<string> words = inputTable.SelectMany(x => x.line.Split(' ')); IQueryable<IGrouping<string, string>> groups = words.GroupBy(x => x); IQueryable<Pair> counts = groups.Select(x => new Pair(x.Key, x.Count())); IQueryable<Pair> ordered = counts.OrderByDescending(x => x.Count); IQueryable<Pair> top = ordered.Take(k);   return top; }   To Test: IQueryable<Pair> results = MapReduce(@"c:\DryadData\input", "TestFile.txt", 100); foreach (Pair words in results) Debug.Print(words.ToString());   Note: DryadLINQ applications can use a more compact way to represent the query: return inputTable         .SelectMany(x => x.line.Split(' '))         .GroupBy(x => x)         .Select(x => new Pair(x.Key, x.Count()))         .OrderByDescending(x => x.Count)         .Take(k);     MapReduce using PLINQ The pattern is relevant even for a single multi-core machine, however. We can write our own PLINQ MapReduce in a few lines. the Map function takes a single input value and returns a set of mapped values àLINQ's SelectMany operator. These are then grouped according to an intermediate key à LINQ GroupBy operator. The Reduce function takes each intermediate key and a set of values for that key, and produces any number of outputs per key à LINQ SelectMany again. We can put all of this together to implement MapReduce in PLINQ that returns a ParallelQuery<T> public static ParallelQuery<TResult> MapReduce<TSource, TMapped, TKey, TResult>( this ParallelQuery<TSource> source, Func<TSource, IEnumerable<TMapped>> map, Func<TMapped, TKey> keySelector, Func<IGrouping<TKey, TMapped>, IEnumerable<TResult>> reduce) { return source .SelectMany(map) .GroupBy(keySelector) .SelectMany(reduce); } the map function takes in an input document and outputs all of the words in that document. The grouping phase groups all of the identical words together, such that the reduce phase can then count the words in each group and output a word/count pair for each grouping: var files = Directory.EnumerateFiles(dirPath, "*.txt").AsParallel(); var counts = files.MapReduce( path => File.ReadLines(path).SelectMany(line => line.Split(delimiters)), word => word, group => new[] { new KeyValuePair<string, int>(group.Key, group.Count()) });

    Read the article

  • Extreme Optimization – Numerical Algorithm Support

    - by JoshReuben
    Function Delegates Many calculations involve the repeated evaluation of one or more user-supplied functions eg Numerical integration. The EO MathLib provides delegate types for common function signatures and the FunctionFactory class can generate new delegates from existing ones. RealFunction delegate - takes one Double parameter – can encapsulate most of the static methods of the System.Math class, as well as the classes in the Extreme.Mathematics.SpecialFunctions namespace: var sin = new RealFunction(Math.Sin); var result = sin(1); BivariateRealFunction delegate - takes two Double parameters: var atan2 = new BivariateRealFunction (Math.Atan2); var result = atan2(1, 2); TrivariateRealFunction delegate – represents a function takes three Double arguments ParameterizedRealFunction delegate - represents a function taking one Integer and one Double argument that returns a real number. The Pow method implements such a function, but the arguments need order re-arrangement: static double Power(int exponent, double x) { return ElementaryFunctions.Pow(x, exponent); } ... var power = new ParameterizedRealFunction(Power); var result = power(6, 3.2); A ComplexFunction delegate - represents a function that takes an Extreme.Mathematics.DoubleComplex argument and also returns a complex number. MultivariateRealFunction delegate - represents a function that takes an Extreme.Mathematics.LinearAlgebra.Vector argument and returns a real number. MultivariateVectorFunction delegate - represents a function that takes a Vector argument and returns a Vector. FastMultivariateVectorFunction delegate - represents a function that takes an input Vector argument and an output Matrix argument – avoiding object construction  The FunctionFactory class RealFromBivariateRealFunction and RealFromParameterizedRealFunction helper methods - transform BivariateRealFunction or a ParameterizedRealFunction into a RealFunction delegate by fixing one of the arguments, and treating this as a new function of a single argument. var tenthPower = FunctionFactory.RealFromParameterizedRealFunction(power, 10); var result = tenthPower(x); Note: There is no direct way to do this programmatically in C# - in F# you have partial value functions where you supply a subset of the arguments (as a travelling closure) that the function expects. When you omit arguments, F# generates a new function that holds onto/remembers the arguments you passed in and "waits" for the other parameters to be supplied. let sumVals x y = x + y     let sumX = sumVals 10     // Note: no 2nd param supplied.     // sumX is a new function generated from partially applied sumVals.     // ie "sumX is a partial application of sumVals." let sum = sumX 20     // Invokes sumX, passing in expected int (parameter y from original)  val sumVals : int -> int -> int val sumX : (int -> int) val sum : int = 30 RealFunctionsToVectorFunction and RealFunctionsToFastVectorFunction helper methods - combines an array of delegates returning a real number or a vector into vector or matrix functions. The resulting vector function returns a vector whose components are the function values of the delegates in the array. var funcVector = FunctionFactory.RealFunctionsToVectorFunction(     new MultivariateRealFunction(myFunc1),     new MultivariateRealFunction(myFunc2));  The IterativeAlgorithm<T> abstract base class Iterative algorithms are common in numerical computing - a method is executed repeatedly until a certain condition is reached, approximating the result of a calculation with increasing accuracy until a certain threshold is reached. If the desired accuracy is achieved, the algorithm is said to converge. This base class is derived by many classes in the Extreme.Mathematics.EquationSolvers and Extreme.Mathematics.Optimization namespaces, as well as the ManagedIterativeAlgorithm class which contains a driver method that manages the iteration process.  The ConvergenceTest abstract base class This class is used to specify algorithm Termination , convergence and results - calculates an estimate for the error, and signals termination of the algorithm when the error is below a specified tolerance. Termination Criteria - specify the success condition as the difference between some quantity and its actual value is within a certain tolerance – 2 ways: absolute error - difference between the result and the actual value. relative error is the difference between the result and the actual value relative to the size of the result. Tolerance property - specify trade-off between accuracy and execution time. The lower the tolerance, the longer it will take for the algorithm to obtain a result within that tolerance. Most algorithms in the EO NumLib have a default value of MachineConstants.SqrtEpsilon - gives slightly less than 8 digits of accuracy. ConvergenceCriterion property - specify under what condition the algorithm is assumed to converge. Using the ConvergenceCriterion enum: WithinAbsoluteTolerance / WithinRelativeTolerance / WithinAnyTolerance / NumberOfIterations Active property - selectively ignore certain convergence tests Error property - returns the estimated error after a run MaxIterations / MaxEvaluations properties - Other Termination Criteria - If the algorithm cannot achieve the desired accuracy, the algorithm still has to end – according to an absolute boundary. Status property - indicates how the algorithm terminated - the AlgorithmStatus enum values:NoResult / Busy / Converged (ended normally - The desired accuracy has been achieved) / IterationLimitExceeded / EvaluationLimitExceeded / RoundOffError / BadFunction / Divergent / ConvergedToFalseSolution. After the iteration terminates, the Status should be inspected to verify that the algorithm terminated normally. Alternatively, you can set the ThrowExceptionOnFailure to true. Result property - returns the result of the algorithm. This property contains the best available estimate, even if the desired accuracy was not obtained. IterationsNeeded / EvaluationsNeeded properties - returns the number of iterations required to obtain the result, number of function evaluations.  Concrete Types of Convergence Test classes SimpleConvergenceTest class - test if a value is close to zero or very small compared to another value. VectorConvergenceTest class - test convergence of vectors. This class has two additional properties. The Norm property specifies which norm is to be used when calculating the size of the vector - the VectorConvergenceNorm enum values: EuclidianNorm / Maximum / SumOfAbsoluteValues. The ErrorMeasure property specifies how the error is to be measured – VectorConvergenceErrorMeasure enum values: Norm / Componentwise ConvergenceTestCollection class - represent a combination of tests. The Quantifier property is a ConvergenceTestQuantifier enum that specifies how the tests in the collection are to be combined: Any / All  The AlgorithmHelper Class inherits from IterativeAlgorithm<T> and exposes two methods for convergence testing. IsValueWithinTolerance<T> method - determines whether a value is close to another value to within an algorithm's requested tolerance. IsIntervalWithinTolerance<T> method - determines whether an interval is within an algorithm's requested tolerance.

    Read the article

  • Extreme Optimization –Mathematical Constants and Basic Functions

    - by JoshReuben
    Machine constants The MachineConstants class - contains constants for floating-point arithmetic because the CLS System.Single and Double floating-point types do not follow the standard conventions and are useless. machine constants for the Double type: machine precision: Epsilon , SqrtEpsilon CubeRootEpsilon largest possible value: MaxDouble , SqrtMaxDouble, LogMaxDouble smallest Double-precision floating point number that is greater than zero: MinDouble , SqrtMinDouble , LogMinDouble A similar set of constants is available for the Single Datatype  Mathematical Constants The Constants class contains static fields for many mathematical constants and common expressions involving small integers – if you are doing thousands of iterations, you wouldn't want to calculate OneOverSqrtTwoPi , Sqrt17 or Log17 !!! Fundamental constants E - The base for the natural logarithm, e (2.718...). EulersConstant - (0.577...). GoldenRatio - (1.618...). Pi - the ratio between the circumference and the diameter of a circle (3.1415...). Expressions involving fundamental constants: TwoPi, PiOverTwo, PiOverFour, LogTwoPi, PiSquared, SqrPi, SqrtTwoPi, OneOverSqrtPi, OneOverSqrtTwoPi Square roots of small integers: Sqrt2, Sqrt3, Sqrt5, Sqrt7, Sqrt17 Logarithms of small integers: Log2, Log3, Log10, Log17, InvLog10  Elementary Functions The IterativeAlgorithm<T> class in the Extreme.Mathematics namespace defines many elementary functions that are missing from System.Math. Hyperbolic Trig Functions: Cosh, Coth, Csch, Sinh, Sech, Tanh Inverse Hyperbolic Trig Functions: Acosh, Acoth, Acsch, Asinh, Asech, Atanh Exponential, Logarithmic and Miscellaneous Functions: ExpMinus1 - The exponential function minus one, ex-1. Hypot - The hypotenuse of a right-angled triangle with specified sides. LambertW - Lambert's W function, the (real) solution W of x=WeW. Log1PlusX - The natural logarithm of 1+x. Pow - A number raised to an integer power.

    Read the article

  • RiverTrail - JavaScript GPPGU Data Parallelism

    - by JoshReuben
    Where is WebCL ? The Khronos WebCL working group is working on a JavaScript binding to the OpenCL standard so that HTML 5 compliant browsers can host GPGPU web apps – e.g. for image processing or physics for WebGL games - http://www.khronos.org/webcl/ . While Nokia & Samsung have some protype WebCL APIs, Intel has one-upped them with a higher level of abstraction: RiverTrail. Intro to RiverTrail Intel Labs JavaScript RiverTrail provides GPU accelerated SIMD data-parallelism in web applications via a familiar JavaScript programming paradigm. It extends JavaScript with simple deterministic data-parallel constructs that are translated at runtime into a low-level hardware abstraction layer. With its high-level JS API, programmers do not have to learn a new language or explicitly manage threads, orchestrate shared data synchronization or scheduling. It has been proposed as a draft specification to ECMA a (known as ECMA strawman). RiverTrail runs in all popular browsers (except I.E. of course). To get started, download a prebuilt version https://github.com/downloads/RiverTrail/RiverTrail/rivertrail-0.17.xpi , install Intel's OpenCL SDK http://www.intel.com/go/opencl and try out the interactive River Trail shell http://rivertrail.github.com/interactive For a video overview, see  http://www.youtube.com/watch?v=jueg6zB5XaM . ParallelArray the ParallelArray type is the central component of this API & is a JS object that contains ordered collections of scalars – i.e. multidimensional uniform arrays. A shape property describes the dimensionality and size– e.g. a 2D RGBA image will have shape [height, width, 4]. ParallelArrays are immutable & fluent – they are manipulated by invoking methods on them which produce new ParallelArray objects. ParallelArray supports several constructors over arrays, functions & even the canvas. // Create an empty Parallel Array var pa = new ParallelArray(); // pa0 = <>   // Create a ParallelArray out of a nested JS array. // Note that the inner arrays are also ParallelArrays var pa = new ParallelArray([ [0,1], [2,3], [4,5] ]); // pa1 = <<0,1>, <2,3>, <4.5>>   // Create a two-dimensional ParallelArray with shape [3, 2] using the comprehension constructor var pa = new ParallelArray([3, 2], function(iv){return iv[0] * iv[1];}); // pa7 = <<0,0>, <0,1>, <0,2>>   // Create a ParallelArray from canvas.  This creates a PA with shape [w, h, 4], var pa = new ParallelArray(canvas); // pa8 = CanvasPixelArray   ParallelArray exposes fluent API functions that take an elemental JS function for data manipulation: map, combine, scan, filter, and scatter that return a new ParallelArray. Other functions are scalar - reduce  returns a scalar value & get returns the value located at a given index. The onus is on the developer to ensure that the elemental function does not defeat data parallelization optimization (avoid global var manipulation, recursion). For reduce & scan, order is not guaranteed - the onus is on the dev to provide an elemental function that is commutative and associative so that scan will be deterministic – E.g. Sum is associative, but Avg is not. map Applies a provided elemental function to each element of the source array and stores the result in the corresponding position in the result array. The map method is shape preserving & index free - can not inspect neighboring values. // Adding one to each element. var source = new ParallelArray([1,2,3,4,5]); var plusOne = source.map(function inc(v) {     return v+1; }); //<2,3,4,5,6> combine Combine is similar to map, except an index is provided. This allows elemental functions to access elements from the source array relative to the one at the current index position. While the map method operates on the outermost dimension only, combine, can choose how deep to traverse - it provides a depth argument to specify the number of dimensions it iterates over. The elemental function of combine accesses the source array & the current index within it - element is computed by calling the get method of the source ParallelArray object with index i as argument. It requires more code but is more expressive. var source = new ParallelArray([1,2,3,4,5]); var plusOne = source.combine(function inc(i) { return this.get(i)+1; }); reduce reduces the elements from an array to a single scalar result – e.g. Sum. // Calculate the sum of the elements var source = new ParallelArray([1,2,3,4,5]); var sum = source.reduce(function plus(a,b) { return a+b; }); scan Like reduce, but stores the intermediate results – return a ParallelArray whose ith elements is the results of using the elemental function to reduce the elements between 0 and I in the original ParallelArray. // do a partial sum var source = new ParallelArray([1,2,3,4,5]); var psum = source.scan(function plus(a,b) { return a+b; }); //<1, 3, 6, 10, 15> scatter a reordering function - specify for a certain source index where it should be stored in the result array. An optional conflict function can prevent an exception if two source values are assigned the same position of the result: var source = new ParallelArray([1,2,3,4,5]); var reorder = source.scatter([4,0,3,1,2]); // <2, 4, 5, 3, 1> // if there is a conflict use the max. use 33 as a default value. var reorder = source.scatter([4,0,3,4,2], 33, function max(a, b) {return a>b?a:b; }); //<2, 33, 5, 3, 4> filter // filter out values that are not even var source = new ParallelArray([1,2,3,4,5]); var even = source.filter(function even(iv) { return (this.get(iv) % 2) == 0; }); // <2,4> Flatten used to collapse the outer dimensions of an array into a single dimension. pa = new ParallelArray([ [1,2], [3,4] ]); // <<1,2>,<3,4>> pa.flatten(); // <1,2,3,4> Partition used to restore the original shape of the array. var pa = new ParallelArray([1,2,3,4]); // <1,2,3,4> pa.partition(2); // <<1,2>,<3,4>> Get return value found at the indices or undefined if no such value exists. var pa = new ParallelArray([0,1,2,3,4], [10,11,12,13,14], [20,21,22,23,24]) pa.get([1,1]); // 11 pa.get([1]); // <10,11,12,13,14>

    Read the article

  • WCF Operations and Multidimensional Arrays

    - by JoshReuben
    You cant pass MultiD arrays accross the wire using WCF - you need to pass jagged arrays. heres 2 extension methods that will allow you to convert prior to serialzation and convert back after deserialization:         public static T[,] ToMultiD<T>(this T[][] jArray)         {             int i = jArray.Count();             int j = jArray.Select(x => x.Count()).Aggregate(0, (current, c) => (current > c) ? current : c);                         var mArray = new T[i, j];             for (int ii = 0; ii < i; ii++)             {                 for (int jj = 0; jj < j; jj++)                 {                     mArray[ii, jj] = jArray[ii][jj];                 }             }             return mArray;         }         public static T[][] ToJagged<T>(this T[,] mArray)         {             var cols = mArray.GetLength(0);             var rows = mArray.GetLength(1);             var jArray = new T[cols][];             for (int i = 0; i < cols; i++)             {                 jArray[i] = new T[rows];                 for (int j = 0; j < rows; j++)                 {                     jArray[i][j] = mArray[i, j];                 }             }             return jArray;         } enjoy!

    Read the article

  • What Color is your Jetpack ?

    - by JoshReuben
    I’m a programmer, Im approaching 40, and I’m fairly decent at my job – I’ll keep doing what I’m doing for as long as they let me!   So what are your career options if you know how to code? A Programmer could be ..   An Algorithm developer Pros Interesting High barriers of entry, potential for startup competitive factor Cons Do you have the skill, qualifications? What are working conditions n this mystery niche ? micro-focus An Academic Pros Low pressure Job security – or is this an illusion ? Cons Low Pay Need a PhD A Software Architect Pros: strategic, rather than tactical Setting technology platform and high level vision You say how it should work, others have to figure out why its not working the way its supposed to ! broad view – you are paid to learn (how do you con people into paying for you to learn ??) Cons: Glorified developer – more often than not! competitive – everyone wants to do it ! loose touch with underlying tech in tough times, first guy to get the axe ! A Software Engineer Pros: interesting, always more to learn fun I can do it Fallback Cons: Nothing new under the sun – been there, done that Dealing with poor requirements, deadlines, other peoples code, overtime C#, XAML, Web - Low barriers of entry –> à race to the bottom A Team leader Pros: Setting code standards and proposing technology choices Cons: Glorified developer – more often than not! Inspecting other peoples code and debugging the problems they cannot fix Dealing with mugbies and prima donas Responsible for QA of others A Project Manager Pros No need for debugging other peoples code Cons Low barrier of entry High pressure Responsible for QA of others Loosing touch with technology A lot of bullshit meetings Have to be an asshole A Product Manager Pros No need for debugging other peoples code Learning new skillset of sales and marketing Cons Travel (I'm a family man) May need to know the bs details of an uninteresting product things I want to work with: AI, algorithms, Numerical Computing, Mathematica, C++ AMP – unfortunately, the work here is few & far between. VS & TFS Extensibility, DSLs (Workflow , Lightswitch), Code Generation – one day, code will write code ! Unity3D, WebGL – fun, fun, fun ! Modern Web – Knockout, SignalR, MVC, Node.Js ??? (tentative – I'll wait until things stabilize as this area is undergoing a pre-Cambrian explosion) Things I don’t want to work with: (but will if I'm asked to !) C# – same old, same old – not learning anything new here Old code – blech ! Environment with code & fix mentality , ad hoc requirements, excessive overtime Pc support, System administration – even after 20 years, people still ask you to do this sometimes ! debugging – my skills are just not there yet Oracle Old tech: VB 6, XSLT, WinForms, Net 3.51 or less Old style Web dev Information Systems: ASP.NET webforms, Reporting services / crystal reports, SQL Server CRUD with manual data layer, XAML MVVM – variations of the same concept, ad nauseaum. Low barriers of entry –> race to the bottom.  Metro – an elegant API coupled to a horrendous UX – I'll wait for market penetration viability before investing further in this.   Conclusion So if you are in a slump, take heart: Programming is a great career choice compared to every other job !

    Read the article

  • C++ Numerical Recipes &ndash; A New Adventure!

    - by JoshReuben
    I am about to embark on a great journey – over the next 6 weeks I plan to read through C++ Numerical Recipes 3rd edition http://amzn.to/YtdpkS I'll be reading this with an eye to C++ AMP, thinking about implementing the suitable subset (non-recursive, additive, commutative) to run on the GPU. APIs supporting HPC, GPGPU or MapReduce are all useful – providing you have the ability to choose the correct algorithm to leverage on them. I really think this is the most fascinating area of programming – a lot more exciting than LOB CRUD !!! When you think about it , everything is a function – we categorize & we extrapolate. As abstractions get higher & less leaky, sooner or later information systems programming will become a non-programmer task – you will be using WYSIWYG designers to build: GUIs MVVM service mapping & virtualization workflows ORM Entity relations In the data source SharePoint / LightSwitch are not there yet, but every iteration gets closer. For information workers, managed code is a race to the bottom. As MS futures are a bit shaky right now, the provider agnostic nature & higher barriers of entry of both C++ & Numerical Analysis seem like a rational choice to me. Its also fascinating – stepping outside the box. This is not the first time I've delved into numerical analysis. 6 months ago I read Numerical methods with Applications, which can be found for free online: http://nm.mathforcollege.com/ 2 years ago I learned the .NET Extreme Optimization library www.extremeoptimization.com – not bad 2.5 years ago I read Schaums Numerical Analysis book http://amzn.to/V5yuLI - not an easy read, as topics jump back & forth across chapters: 3 years ago I read Practical Numerical Methods with C# http://amzn.to/V5yCL9 (which is a toy learning language for this kind of stuff) I also read through AI a Modern Approach 3rd edition END to END http://amzn.to/V5yQSp - this took me a few years but was the most rewarding experience. I'll post progress updates – see you on the other side !

    Read the article

  • XNA Multiplayer Games and Networking

    - by JoshReuben
    ·        XNA communication must by default be lightweight – if you are syncing game state between players from the Game.Update method, you must minimize traffic. That game loop may be firing 60 times a second and player 5 needs to know if his tank has collided with any player 3 and the angle of that gun turret. There are no WCF ServiceContract / DataContract niceties here, but at the same time the XNA networking stack simplifies the details. The payload must be simplistic - just an ordered set of numbers that you would map to meaningful enum values upon deserialization.   Overview ·        XNA allows you to create and join multiplayer game sessions, to manage game state across clients, and to interact with the friends list ·        Dependency on Gamer Services - to receive notifications such as sign-in status changes and game invitations ·        two types of online multiplayer games: system link game sessions (LAN) and LIVE sessions (WAN). ·        Minimum dev requirements: 1 Xbox 360 console + Creators Club membership to test network code - run 1 instance of game on Xbox 360, and 1 on a Windows-based computer   Network Sessions ·        A network session is made up of players in a game + up to 8 arbitrary integer properties describing the session ·        create custom enums – (e.g. GameMode, SkillLevel) as keys in NetworkSessionProperties collection ·        Player state: lobby, in-play   Session Types ·        local session - for split-screen gaming - requires no network traffic. ·        system link session - connects multiple gaming machines over a local subnet. ·        Xbox LIVE multiplayer session - occurs on the Internet. Ranked or unranked   Session Updates ·        NetworkSession class Update method - must be called once per frame. ·        performs the following actions: o   Sends the network packets. o   Changes the session state. o   Raises the managed events for any significant state changes. o   Returns the incoming packet data. ·        synchronize the session à packet-received and state-change events à no threading issues   Session Config ·        Session host - gaming machine that creates the session. XNA handles host migration ·        NetworkSession properties: AllowJoinInProgress , AllowHostMigration ·        NetworkSession groups: AllGamers, LocalGamers, RemoteGamers   Subscribe to NetworkSession events ·        GamerJoined ·        GamerLeft ·        GameStarted ·        GameEnded – use to return to lobby ·        SessionEnded – use to return to title screen   Create a Session session = NetworkSession.Create(         NetworkSessionType.SystemLink,         maximumLocalPlayers,         maximumGamers,         privateGamerSlots,         sessionProperties );   Start a Session if (session.IsHost) {     if (session.IsEveryoneReady)     {        session.StartGame();        foreach (var gamer in SignedInGamer.SignedInGamers)        {             gamer.Presence.PresenceMode =                 GamerPresenceMode.InCombat;   Find a Network Session AvailableNetworkSessionCollection availableSessions = NetworkSession.Find(     NetworkSessionType.SystemLink,       maximumLocalPlayers,     networkSessionProperties); availableSessions.AllowJoinInProgress = true;   Join a Network Session NetworkSession session = NetworkSession.Join(     availableSessions[selectedSessionIndex]);   Sending Network Data var packetWriter = new PacketWriter(); foreach (LocalNetworkGamer gamer in session.LocalGamers) {     // Get the tank associated with this player.     Tank myTank = gamer.Tag as Tank;     // Write the data.     packetWriter.Write(myTank.Position);     packetWriter.Write(myTank.TankRotation);     packetWriter.Write(myTank.TurretRotation);     packetWriter.Write(myTank.IsFiring);     packetWriter.Write(myTank.Health);       // Send it to everyone.     gamer.SendData(packetWriter, SendDataOptions.None);     }   Receiving Network Data foreach (LocalNetworkGamer gamer in session.LocalGamers) {     // Keep reading while packets are available.     while (gamer.IsDataAvailable)     {         NetworkGamer sender;          // Read a single packet.         gamer.ReceiveData(packetReader, out sender);          if (!sender.IsLocal)         {             // Get the tank associated with this packet.             Tank remoteTank = sender.Tag as Tank;              // Read the data and apply it to the tank.             remoteTank.Position = packetReader.ReadVector2();             …   End a Session if (session.AllGamers.Count == 1)         {             session.EndGame();             session.Update();         }   Performance •        Aim to minimize payload, reliable in order messages •        Send Data Options: o   Unreliable, out of order -(SendDataOptions.None) o   Unreliable, in order (SendDataOptions.InOrder) o   Reliable, out of order (SendDataOptions.Reliable) o   Reliable, in order (SendDataOptions.ReliableInOrder) o   Chat data (SendDataOptions.Chat) •        Simulate: NetworkSession.SimulatedLatency , NetworkSession.SimulatedPacketLoss •        Voice support – NetworkGamer properties: HasVoice ,IsTalking , IsMutedByLocalUser

    Read the article

  • Azure Futures - Distributed Computing and Number Crunching

    - by JoshReuben
    "the biggest Azure customers today are the ones using HPC on-premises at the current time" - http://www.zdnet.com/blog/microsoft/windows-azure-futures-turning-the-cloud-into-a-supercomputer/8592?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+zdnet%2Fmicrosoft+%28ZDNet+All+About+Microsoft%29&utm_content=Google+Reader   Orleans Framework for cloud computing - http://research.microsoft.com/en-us/projects/orleans     HPC on Azure - http://www.zdnet.com/blog/microsoft/microsoft-finalizes-its-latest-supercomputing-operating-system-release/7414   Dryad is Microsoft’s competitor to Google MapReduce and Apache Hadoop  - http://www.zdnet.com/blog/microsoft/microsoft-takes-a-step-toward-commercializing-its-dryad-distributed-computing-technologies/8255?tag=mantle_skin;content   SQL Server Analysis Services DataMining in the cloud - http://www.sqlmag.com/article/reporting2/azure-data-mining-in-the-cloud.aspx

    Read the article

  • Workflow Activity Extensions, Activity Packs and Unit Testing Framework

    - by JoshReuben
    http://wf.codeplex.com/ contains a plethora of infrastructure code and new activities for extending Workflow Foundation 4. These are also available as Nuget packages. These include: Activity Extensions Security Activity Pack ADO.NET Activity Pack Azure Activity Pack Activity Unit Testing Framework   view my PowerPoint presentation on these and more here: http://www.slideshare.net/joshuareuben9/workflow-foundation-activity-packs-extensions-and-unit-testing

    Read the article

1