net.finmath.analytic.calibration

Interface ParameterTransformation

public interface ParameterTransformation

Interface for parameter transformation. A parameter transformation can be used to implement constrains. Example: The solver/optimizer assumes that its parameters are unbounded, i.e. admissible values are searched in the range of Double.NEGATIVE_INFINITY to Double.POSITIVE_INFINITY. If you like to constrain the first parameter to be > 0, then you might use the parameter transformation where getParameter is just Math:exp and getSolverParameter is Math:log.

Version:

1.0

Author:

Christian Fries

Method Summary

Modifier and Type	Method and Description
RandomVariable[]	getParameter(RandomVariable[] solverParameter) Return the original parameter for the given (unbounded) solver parameter.
RandomVariable[]	getSolverParameter(RandomVariable[] parameter) Return the (unbounded) solver parameter for the given original parameter.

Method Detail

getParameter

RandomVariable[] getParameter(RandomVariable[] solverParameter)

Return the original parameter for the given (unbounded) solver parameter.

Parameters:

solverParameter - The given solver parameter (unbounded).

Returns:

The (possibly bounded) parameter.

getSolverParameter

RandomVariable[] getSolverParameter(RandomVariable[] parameter)

Return the (unbounded) solver parameter for the given original parameter.

Parameters:

parameter - The parameter.

Returns:

The corresponding unbounded solver parameter.