Skip to content

Minimize a function of a single variable using golden section search

License

Notifications You must be signed in to change notification settings

scijs/minimize-golden-section-1d

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

minimize-golden-section-1d

Build Status npm version Dependency Status js-semistandard-style

Minimize a function of a single variable using golden section search

Introduction

Returns the argument that minimizes a function of a single variable using golden section search, a search algorithm similar to bisection search except that the size of the intervals is chosen so that function evaluations can be reused more effectively. The golden section works fine with discontinuities, asymptotes, or oscillations that would throw off derivative-based methods but is somewhat slower to converge, likely requiring more function evaluations.

Installation

$ npm install minimize-golden-section-1d

Example

Returns the argument that minimizes the function:

var minimize = require('minimize-golden-section-1d');

minimize(Math.cos)
// => 3.1415926622945616

minimize(Math.cos, {lowerBound: 0, upperBound: 1});
// => 1

minimize(Math.cos, {guess: -3});
// => -3.1415926432597825

Usage

require('minimize-golden-section-1d')(f[, options[, status]])

Given function f of one Number-valued variable, computes a local minimum. On successful completion, returns the value of the argument that minimizes f (note that this may either be a local or global minimum in the provided range). If the algorithm fails (i.e. if max iterations exceeded, NaN encountered, or unbounded divergence of the argument to Infinity), returns NaN.

If bounds are provided, this module proceeds immediately with golden section search. If upper, lower or both bounds are not provided, the algorithm will make use of an initial guess (a provided guess or just one of the bounds if that's all it has) and expand the search range until a minimum is bracketed.

Options:
  • tolerance [=1e-8]: Convergence tolerance. Algorithm continues until search interval is smaller than tolerance.
  • lowerBound [=-Infinity]: Lower bound of the search interval
  • upperBound [=Infinity]: Upper bound of the search interval
  • maxIterations [=100]: Maximum number of iterations for either bracketing or search phase.
  • guess [=0]: initial guess for unbounded minimization. Unused unless both upperBound and lowerBound are not provided.
  • initialIncrement [=1]: Initial interval by which to expand the search interval for unbounded minimization. Unused unless both upperBound and lowerBound are not provided.
Status:

If status object is provided, the following outputs are written in place:

  • status.converged (Boolean): true if algorithm succeeded
  • status.iterations (Number): number of iterations
  • status.minimum (Number): minimum value of the function achieved before exiting, whether the algorithm converged or not
  • status.argmin (Number): best guess of argmin achieved before exiting, whether the algorithm converged or not.

License

© 2015 Scijs. MIT License.

Authors

Ricky Reusser

About

Minimize a function of a single variable using golden section search

Resources

License

Stars

Watchers

Forks

Packages

No packages published