Class BFGSOptimizer

java.lang.Object

org.episteme.core.mathematics.optimization.BFGSOptimizer

public class BFGSOptimizer
extends Object

BFGS (Broyden-Fletcher-Goldfarb-Shanno) quasi-Newton optimization.

Efficient unconstrained optimization without computing Hessian directly. Uses rank-2 updates to approximate inverse Hessian.

*

Reference:
Broyden, C. G., Fletcher, R., Goldfarb, D., invalid input: '&' Shanno, D. F. (1970). The BFGS Algorithm. Journal of the Institute of Mathematics and Its Applications.

Since:

1.0

Author:

Silvere Martin-Michiellot, Gemini AI (Google DeepMind)

Constructor Summary

Constructors

Constructor	Description
BFGSOptimizer()

Method Summary

Modifier and Type	Method	Description
static Real[]	minimize(Function<Real[],Real> f, Function<Real[],Real[]> gradient, Real[] initialGuess, double tolerance, int maxIterations)	Minimizes a multivariate function using BFGS.

Methods inherited from class Object

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

Constructor Details

BFGSOptimizer

public BFGSOptimizer()

Method Details

minimize

public static Real[] minimize(Function<Real[],Real> f,
 Function<Real[],Real[]> gradient,
 Real[] initialGuess,
 double tolerance,
 int maxIterations)

Minimizes a multivariate function using BFGS.

Parameters:

f - Objective function (Real[] -> Real)

gradient - Gradient of f (Real[] -> Real[])

initialGuess - Starting point

tolerance - Convergence tolerance (gradient norm)

maxIterations - Maximum iterations

Returns:

Approximate minimum point