mindspore.scipy.optimize.line_search

mindspore.scipy.optimize.line_search(f, xk, pk, jac=None, gfk=None, old_fval=None, old_old_fval=None, c1=0.0001, c2=0.9, maxiter=20)[source]

Inexact line search that satisfies strong Wolfe conditions.

Algorithm 3.5 from Wright and Nocedal, 'Numerical Optimization', 1999, pg. 59-61

Note

line_search is not supported on Windows platform yet.

Parameters

f (function) – function of the form f(x) where x is a flat Tensor and returns a real scalar. The function should be composed of operations with vjp defined.
xk (Tensor) – initial guess.
pk (Tensor) – direction to search in. Assumes the direction is a descent direction.
jac (function) – the gradient function at x where x is a flat Tensor and returns a Tensor. The function can be None if you want to use automatic credits.
gfk (Tensor) – initial value of value_and_gradient as position. Default: None .
old_fval (Tensor) – The same as gfk. Default: None .
old_old_fval (Tensor) – unused argument, only for scipy API compliance. Default: None .
c1 (float) – Wolfe criteria constant, see ref. Default: 1e-4 .
c2 (float) – The same as c1. Default: 0.9 .
maxiter (int) – maximum number of iterations to search. Default: 20 .

Returns

LineSearchResults, results of line search results.

Supported Platforms:: GPU CPU

Examples

>>> import numpy as onp
>>> from mindspore.scipy.optimize import line_search
>>> from mindspore import Tensor
>>> x0 = Tensor(onp.ones(2).astype(onp.float32))
>>> p0 = Tensor(onp.array([-1, -1]).astype(onp.float32))
>>> def func(x):
...     return x[0] ** 2 - x[1] ** 3
>>> res = line_search(func, x0, p0)
>>> print(res.a_k)
1.0