Hi, I'm trying to port a python project to Rust and I'm using your library as a replacement to scipy.optimize. rustimization seems to be working ok but there's something I don't like. In scipy/optimize/lbfgsb.py, we can find this
if jac is None:
def func_and_grad(x):
f = fun(x, *args)
g = _approx_fprime_helper(x, fun, epsilon, args=args, f0=f)
return f, g
which is to say 1) use this if g wasn't provided 2) use the last value of f() when calling g(). Using your library, doing 1) is possible, but it's impossible to do 2). It's quite important if f() is expensive.
I already coded _approx_fprime_helper and I'm pretty sure I can code this myself in your library. My question is more: is this project still alive? Do you accept PR?
Currently I am little bit busy with my work so I am unable to spend time on this repository. But in future i have some plans with it. And yeah i do accept PR.
Hi, I'm trying to port a python project to Rust and I'm using your library as a replacement to scipy.optimize.
rustimization
seems to be working ok but there's something I don't like. In scipy/optimize/lbfgsb.py, we can find thiswhich is to say 1) use this if
g
wasn't provided 2) use the last value off()
when callingg()
. Using your library, doing 1) is possible, but it's impossible to do 2). It's quite important iff()
is expensive.I already coded
_approx_fprime_helper
and I'm pretty sure I can code this myself in your library. My question is more: is this project still alive? Do you accept PR?