CODEWITHSUNDEEP
pythonnumpyoptimizationscipy
ap21
ap21

Reputation: 2614

Scipy optimisation: Get the function to print out its iterations

I am using scipy.optimize.minimize()to minimise a certain function. I want to compare the performance of different methods, BFGSand L-BFGS-B, and for that, I would like the function to print out its values and error margins as it is optimising.

The L-BFGS-B does this automatically in fact, and it looks like the following:

At X0         0 variables are exactly at the bounds

At iterate    0    f=  7.73701D+04    |proj g|=  1.61422D+03

At iterate    1    f=  4.33415D+04    |proj g|=  1.16289D+03

At iterate    2    f=  9.97661D+03    |proj g|=  5.04925D+02

At iterate    3    f=  4.10666D+03    |proj g|=  3.04707D+02

....

At iterate  194    f=  3.34407D+00    |proj g|=  3.55117D-04

At iterate  195    f=  3.34407D+00    |proj g|=  3.36692D-04

At iterate  196    f=  3.34407D+00    |proj g|=  9.58307D-04

Tit   = total number of iterations
Tnf   = total number of function evaluations
Tnint = total number of segments explored during Cauchy searches
Skip  = number of BFGS updates skipped
Nact  = number of active bounds at final generalized Cauchy point
Projg = norm of the final projected gradient
F     = final function value

       * * *

N    Tit     Tnf  Tnint  Skip  Nact     Projg        F
243    196    205      1     0     0   9.583D-04   3.344D+00
F =   3.34407234824719

Does anyone know how I can do the same for BFGS?

Note: This question is related to a larger question posted here: SciPy optimisation: Newton-CG vs BFGS vs L-BFGS, on the divergence between the behaviour between these two algorithms in a particular optimisation problem. I want to track down where these two algorithms are diverging.

Upvotes: 3

Views: 6196

Answers (1)

ap21
ap21

Reputation: 2614

I found an answer to this here: How to display progress of scipy.optimize function?

The callback option of optimize.minimize() allows us to feed in a method that has access to the variable x_n calculated by optimize.minimize() at time step n. We can use this to print out the data; I chose to write out to an external file as follows:

##Print callback function
def printx(Xi):
    global Nfeval
    global fout
    fout.write('At iterate {0:4d},  f={1: 3.6f} '.format(Nfeval, energy(Xi)) + '\n')
    Nfeval += 1

Nfeval = 1
fout = open('BFGS_steps_NN%d' %NN +'.txt','w')

res = minimize(energy, xyzInit, method='BFGS', jac = energy_der, callback=printx, options={'disp': True})
fout.close()

It works perfectly!

Upvotes: 2

Related Questions