hess option does not work in minimize function

[CatNofishing]

issue

i traied use hess= {2-point’, ‘3-point’, ‘cs’, HessianUpdateStrategy} according to doc,but it failed.

Reproducing code example:

import numpy as np
from scipy.optimize import minimize,show_options,SR1
import sys
def rosen(x):
    """The Rosenbrock function"""
    return sum(100.0*(x[1:]-x[:-1]**2.0)**2.0 + (1-x[:-1])**2.0)

def rosen_der(x):
    xm = x[1:-1]
    xm_m1 = x[:-2]
    xm_p1 = x[2:]
    der = np.zeros_like(x)
    der[1:-1] = 200*(xm-xm_m1**2) - 400*(xm_p1 - xm**2)*xm - 2*(1-xm)
    der[0] = -400*x[0]*(x[1]-x[0]**2) - 2*(1-x[0])
    der[-1] = 200*(x[-1]-x[-2]**2)
    return der


x0 = np.array([1.3, 0.7, 0.8, 1.9, 1.2])
minimize(rosen, x0, method='Newton-CG',
               options={'disp': False},jac=rosen_der,hess='cs')

Error message:

Scipy/Numpy/Python version information:

1.5.4 1.19.2 sys.version_info(major=3, minor=8, micro=5, releaselevel='final', serial=0)

[tupui]

Thanks @CatNofishing. I think this is just a phrasing issue as it also says:

Available quasi-Newton methods implementing this interface are:

BFGS;
SR1.

Meaning Newton-CG is not compatible here. There could be a validation here as well.

Is this correct @andyfaff?

[CatNofishing]

@tupui Sorry, I still do not understand the description in the document, I suggest to provide a few examples of the use of the hess option parameter。 I set hess=‘2-point’/‘3-point’/’cs’ to all the solvers that provide the hess option, all of them do not work。

[tupui]

@tupui Sorry, I still do not understand the description in the document, I suggest to provide a few examples of the use of the hess option parameter。 I set hess=‘2-point’/‘3-point’/’cs’ to all the solvers that provide the hess option, all of them do not work。

All the trust methods (except exact as the doc states) are working fine. It's true that 'newton-cg', 'dogleg' do not work.

Ex.

minimize(rosen, x0, method='trust-krylov',
         jac=rosen_der, hess='cs')

Any improvement to the doc is always welcome if you want to help.

[tupui]

@andyfaff this goes on the pile for the general rework of the interface. We should really plan this for SciPy 2.

[Vipbone]

I am a new contributor and want to share my opinion

minimize function needs to somehow get to know what frequencies and response are. Need to modify minimize function so that it takes those two parameters in, and then add an argument in your call to minimize to include those arguments.