lib5c.util.optimization.array_newton() is redundant with latest scipy versions

Thomas Gilgenast reporter

changed milestone to 0.6.0

2019-06-21T19:11:05+00:00

Thomas Gilgenast reporter

an important question for clients (who might need to update client code) is whether or not the signatures of lib5c.util.optimization.array_newton() and scipy.optimize.newton() are the same

lib5c has:

def array_newton(func, x0, fprime=None, args=(), tol=1.48e-8, maxiter=50,
                 fprime2=None, failure_idx_flag=None):

and returns root if failure_idx_flag is False or (root, failures, zero_der) if it is True

while scipy has:

def newton(func, x0, fprime=None, args=(), tol=1.48e-8, maxiter=50,
           fprime2=None, x1=None, rtol=0.0,
           full_output=False, disp=True):

and returns root if full_output is False or (root, converged, zero_der) if it is True

based on this it sounds like an “upgrade guide“ is to

replace from lib5c.util.optimization import array_newton with from scipy.optimize import newton
rename the function calls from array_newton() to newton()
rename the kwarg from failure_idx_flag to full_output
if full output was expected, capture it as (root, converged, zero_der) and then write failures = ~converged

‌