Vectors have different size error (Migrated from community.research.microsoft.com)

• Question

• 

  

  0

  Sign in to vote

  johno posted on 02-17-2009 6:15 PM

  Dear Infer.NET users,

  Thanks in advance...

  This error message occured either due to my zero experience in C# or Infer.NET:

  System.ArgumentException: Vectors have different size

  Parameter name: that

  at MicrosoftResearch.Infer.Maths.Vector.CheckCompatible(Vector that, String paramName) in C:\infernetBuilds\05-12-2008_17-31\Runtime\Core\Math\Vector.cs:line 926

  after calling Variable.InnerProduct. However I think the vectors (vbeta and vx) are both of length 2.

  I'm trying to build a simple linear regression model (code below).

  Any assistance would be greatly appreciated.

  Thanks, John

  public     void Run() {

  // Get the arrays of ages and prices from mitsub.txt

  double[ ages; double[ prices; int n;

  LoadData(@"data/mitsub.txt", out ages, out prices, out n);

  double A_tau = 0.01;

  double B_tau = 0.01; double sigma2_beta = 1.0E+8;

  Vector vmu_vbeta = new Vector(new double[ { 0.0, 0.0 });

  PositiveDefiniteMatrix mSigma_vbeta =

  PositiveDefiniteMatrix.IdentityScaledBy(2,1/sigma2_beta);

   

  // Create priors for tau, vbeta

  Variable<double> tau = Variable.GammaFromShapeAndScale(A_tau,1/B_tau).Named("tau");

  Variable<Vector> vbeta = Variable.VectorGaussianFromMeanAndPrecision(vmu_vbeta, mSigma_vbeta).Named("vbeta");

   

  Range agesRange = new Range(n).Named("n");

  Vector[ xdata = new Vector;

  for (int i = 0; i < n; i++) xdata = new Vector(1, ages);

  VariableArray<Vector> vx = Variable.Observed(xdata,agesRange).Named("vx");

   

  VariableArray<double> vmu = Variable.Array<double>(agesRange);

  using(Variable.ForEach(agesRange))  {

  vmu[agesRange] = Variable.InnerProduct(vbeta,vx[agesRange]);

  }

   

  VariableArray<double> vy = Variable.Array<double>(agesRange);

  vy[agesRange] = Variable.GaussianFromMeanAndPrecision(vmu[agesRange],tau).Named("vy");

  vy.ObservedValue = prices;

  // Create an inference engine for VMP

  InferenceEngine engine = new InferenceEngine(new VariationalMessagePassing());

  Console.WriteLine("Dist over tau=\n" + engine.Infer<Gamma>(tau));

  Console.WriteLine("Dist over vbeta=\n" + engine.Infer<VectorGaussian>(vbeta));

  }

   

  Friday, June 3, 2011 4:44 PM