Ipopt test doesn't pass--too many arguments to nlp.solve. about pypower HOT 8 OPEN

I have push up an old version that I think I must have used:

https://github.com/rwl/pyipopt

However, it seems that PyIPOPT has received a considerable amount of attention since I last looked at it:

https://github.com/xuy/pyipopt

I will leave this issue open until PYPOWER has been updated to use that branch. Ultimately, platform specific eggs should be available here:

https://pypi.python.org/pypi/pyipopt

from pypower.

Running the latest pyipopt with Ipopt 3.11.1 and removing the 2nd argument (m) now results in a segmentation fault in Ipopt running the t_opf_ipopt.py test. It's not clear where the issue is (pypower, pyipopt, ipopt)...lots of moving parts. I will try your uploaded pyipopt to see if that changes things (obviously you got it working at one point).

from pypower.

Unfortunately, I still get errors using rwl/pyipopt (in fact it segfaults when creating the problem) which would suggest some api change in Ipopt that I have versus the version that pyipopt works with.

from pypower.

I just remembered that I wrote this guide for setting up external solvers:

https://github.com/rwl/PYPOWER/blob/master/doc/solvers.rst

It looks like I was using Ipopt 3.9.3.

from pypower.

I tried this from scratch at home using the Ipopt 3.11.5 and pyipopt from the current github master and was able to run the optimization part successfully by simply just removing the (m) argument and the modifying the output arguments (zg is moved). So clearly the segfaults are something on my end.

from pypower.

I don't know if any user still have this problem.. but if so, i found a solution for this problem: change the output of eval_jac_g and eval_h functions when flag=True to: return ((Js.row).astype(int), (Js.col).astype(int)) and return ((Hs.row).astype(int), (Hs.col).astype(int)), respectively. Apparentely without this the solver returns a segmentation fault.

from pypower.

Hi kennyvinente, how did you solve the original problem:
x, zl, zu, obj, status, zg = nlp.solve(x0, m, userdata) (TypeError: function takes at most 2 arguments (3 given))
Did you also remove the m? I tried removing the m but then zg is an integer and not an array. This results in my script crashing here:
muSf[il] = 2 * info['lmbda'][2 * nb + arange(nl2)] * branch[il, RATE_A] / baseMVA (TypeError: 'int' object has no attribute 'getitem')
I am using ipopt version 3.12.7

from pypower.

Hello @daetzm .. some changes that i performed was:
x, zl, zu, zg, obj, status, = nlp.solve(x0, userdata) (there is a change in the outputs, probably because of the ipopt version). Also, i'm using ipopt version 3.12.12 and python3.6
And most important.. check the return of all functions, specially the eval_g and eval_h functions..check if they are running correctly.. for example, try with a 3bus system (easy to analise).

the seg fault problem is because this functions are not returning the results appropriately, so i did those changes and now its fine..

here is an image with ipopt running an opf problem

from pypower.