Package ffx.numerics.optimization

Class LineSearch

java.lang.Object

ffx.numerics.optimization.LineSearch

public class LineSearch
extends Object

This class implements an algorithm for uni-dimensional line search using parabolic extrapolation and cubic interpolation with both function and gradient values.

The algorithm works as follows:

1. Initialize with the current point, function value, gradient, and search direction
2. Check if the search direction makes a reasonable angle with the negative gradient
3. Set an initial step size based on previous function decrease
4. Perform parabolic extrapolation to find a better step size
5. If needed, perform cubic interpolation to refine the step size
6. Return the best point found along the search direction

This implementation is a translation of FORTRAN code written by Jay Ponder (search.f).

Since:

1.0

Author:

Michael J. Schnieders

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	LineSearch.LineSearchResult	Enum representing the possible outcomes of a line search operation.

Method Summary

Modifier and Type	Method	Description
double	search(int n, double[] x, double f, double[] g, double[] p, double[] angle, double fMove, LineSearch.LineSearchResult[] info, int[] functionEvaluations, OptimizationInterface optimizationSystem)	Minimize a function along a search direction.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

search

public double search(int n,
 double[] x,
 double f,
 double[] g,
 double[] p,
 double[] angle,
 double fMove,
 LineSearch.LineSearchResult[] info,
 int[] functionEvaluations,
 OptimizationInterface optimizationSystem)

Minimize a function along a search direction.

This method performs a unidimensional line search along the specified search direction using parabolic extrapolation and cubic interpolation with both function and gradient values. The search attempts to find a step size that provides sufficient decrease in the function value while maintaining a reasonable angle between the search direction and negative gradient.

If the search is forced to proceed in an uphill direction (angle > DEFAULT_ANGLEMAX), the method returns after the initial step without performing the search.

The method modifies the input arrays x and g to contain the coordinates and gradient at the best point found along the search direction.

Parameters:

n - Number of variables in the optimization problem.

x - Current variable values (modified to contain the best point found).

f - Current function value at point x.

g - Current gradient values at point x (modified to contain the gradient at the best point).

p - Search direction vector.

angle - Output parameter that will contain the angle between the gradient and search direction.

fMove - Change in function value due to the previous optimization step.

info - Output parameter that will contain the line search result status.

functionEvaluations - Input/output parameter for tracking the number of function evaluations.

optimizationSystem - Implementation of the OptimizationInterface that provides function values and gradients.

Returns:

The final function value at the best point found.

Since:

1.0