public MonitorFunction(LambdaSolve m, CGRunner.LikelihoodFunction lf, string filename)
{
// end static class LikelihoodFunction
// = 0
this.model = m;
this.lf = lf;
this.filename = filename;
}
/// <summary>Set up a LambdaSolve problem for solution by a Minimizer.</summary>
/// <param name="prob">The problem to solve</param>
/// <param name="filename">Used (with extension) to save intermediate results.</param>
/// <param name="tol">Tolerance of errors (passed to CG)</param>
/// <param name="sigmaSquareds">
/// The prior sigma<sup>2</sup> for each feature: this doubled will be
/// used to divide the lambda<sup>2</sup> values as the
/// prior penalty. This array must have size the number of features.
/// If it is null, no regularization will be performed.
/// </param>
public CGRunner(LambdaSolve prob, string filename, double tol, double[] sigmaSquareds)
{
this.prob = prob;
this.filename = filename;
this.tol = tol;
this.useGaussianPrior = sigmaSquareds != null;
this.sigmaSquareds = sigmaSquareds;
this.priorSigmaS = -1.0;
}
/// <summary>Set up a LambdaSolve problem for solution by a Minimizer.</summary>
/// <param name="prob">The problem to solve</param>
/// <param name="filename">Used (with extension) to save intermediate results.</param>
/// <param name="tol">Tolerance of errors (passed to CG)</param>
/// <param name="priorSigmaS">
/// The prior sigma<sup>2</sup>: this doubled will be
/// used to divide the lambda<sup>2</sup> values as the
/// prior penalty. A value of 0.0
/// or Double.POSITIVE_INFINITY
/// indicates to not use regularization.
/// </param>
public CGRunner(LambdaSolve prob, string filename, double tol, double priorSigmaS)
{
// = null;
this.prob = prob;
this.filename = filename;
this.tol = tol;
this.useGaussianPrior = priorSigmaS != 0.0 && priorSigmaS != double.PositiveInfinity;
this.priorSigmaS = priorSigmaS;
this.sigmaSquareds = null;
}
public LikelihoodFunction(LambdaSolve m, double tol, bool useGaussianPrior, double sigmaSquared, double[] sigmaSquareds)
{
model = m;
this.tol = tol;
this.useGaussianPrior = useGaussianPrior;
if (useGaussianPrior)
{
// keep separate prior on each parameter for flexibility
this.sigmaSquareds = new double[model.lambda.Length];
if (sigmaSquareds != null)
{
System.Array.Copy(sigmaSquareds, 0, this.sigmaSquareds, 0, sigmaSquareds.Length);
}
else
{
Arrays.Fill(this.sigmaSquareds, sigmaSquared);
}
}
else
{
this.sigmaSquareds = null;
}
}
/// <summary>
/// Set up a LambdaSolve problem for solution by a Minimizer,
/// specifying a value for sigma<sup>2</sup>.
/// </summary>
/// <param name="prob">The problem to solve</param>
/// <param name="filename">Used (with extension) to save intermediate results.</param>
/// <param name="priorSigmaS">
/// The prior sigma<sup>2</sup>: this doubled will be
/// used to divide the lambda<sup>2</sup> values as the
/// prior penalty in the likelihood. A value of 0.0
/// or Double.POSITIVE_INFINITY
/// indicates to not use regularization.
/// </param>
public CGRunner(LambdaSolve prob, string filename, double priorSigmaS)
: this(prob, filename, DefaultTolerance, priorSigmaS)
{
}
/// <summary>Set up a LambdaSolve problem for solution by a Minimizer.</summary>
/// <remarks>
/// Set up a LambdaSolve problem for solution by a Minimizer.
/// Uses a Gaussian prior with a sigma<sup>2</sup> of 0.5.
/// </remarks>
/// <param name="prob">The problem to solve</param>
/// <param name="filename">Used (with extension) to save intermediate results.</param>
public CGRunner(LambdaSolve prob, string filename)
: this(prob, filename, DefaultSigmasquared)
{
}
