Sign in
Microsoft.com
 
Microsoft
 
 
 

locked
initialize a random variableArray with different distributions in different elements RRS feed

 > 

  • Question

  • Question
    Sign in to vote
    0
    Sign in to vote

    When we have a VariableArray, and each element comes from a different prior, how can we implement it?

    If all elements have the same prior (called xPrior for example) the code is like this:

    VariableArray<double> x = Variable.Array<double>(xRange).Named("x");
    x[xRange] = Variable<double>.Random(xPrior).ForEach(xRange);

    But what if the prior is different for different elements?

    Can I initialize priors separately? Not all comming from the same distribution?

    I wrote this code but I got an error since ObservedValue cannot be derived from an array element (but from the array itself).

    VariableArray<double> x = Variable.Array<double>(xRange).Named("x");

VariableArray<double> means = Variable.Observed(new double[] { 0, 5 }, xRange); 
VariableArray<double> precs = Variable.Observed(new double[] { 0, 1 }, xRange);

x[xRange] = Variable<double>.Random(Gaussian.FromMeanAndPrecision(means[xRange].ObservedValue, precs[xRange].ObservedValue)).ForEach(xRange);

    Is there any way to handle this? Or I should just switch to C# arrays to handle this?

    Wednesday, October 15, 2014 4:49 PM

Answers

  • Question
    Sign in to vote
    0
    Sign in to vote 
    var xPrior = Variable.Array<Gaussian>(xRange);
x[xRange] = Variable<double>.Random(xPrior[xRange]);

    • Marked as answer by Capli19 Wednesday, October 15, 2014 9:27 PM
    Wednesday, October 15, 2014 5:47 PM
    Owner

All replies

  • Question
    Sign in to vote
    0
    Sign in to vote 
    var xPrior = Variable.Array<Gaussian>(xRange);
x[xRange] = Variable<double>.Random(xPrior[xRange]);

    • Marked as answer by Capli19 Wednesday, October 15, 2014 9:27 PM
    Wednesday, October 15, 2014 5:47 PM
    Owner
  • Question
    Sign in to vote
    0
    Sign in to vote

    Oh I thought I need to do this - so the same problem with range:

    xPrior[0] = Gaussian.FromMeanAndVariance(0,0);
xPrior[1] = Gaussian.FromMeanAndVariance(5,1);

    But I understood I can initialize xPrior like this:

    VariableArray<Gaussian> xPrior = Variable.Observed(
                 new Gaussian[] { Gaussian.FromMeanAndVariance(0,0), Gaussian.FromMeanAndVariance(5,1)  }, 
                 xRange);
    Then I don't need to work with range for initialization and it works. Thanks.


    • Edited by Capli19 Wednesday, October 15, 2014 9:27 PM
    Wednesday, October 15, 2014 9:06 PM