We need to replace all instances of np.asfarray(x) with np.asarray(x, dtype=float) -- a very easy fix. The vim/sed command :%s/asfarray(\(.*\))/asarray(\1, dtype=float)/g will do it automatically.
test_poincare is failing, which compares an array against golden answer data stored in a .npy file. The arrays are almost identical for the first 22,000 elements, but start disagreeing in the last 8,000. No idea what's going on there.
For now, installing Pyrokinetics will only grab NumPy versions up to 1.26, as some of our dependencies require numpy == 1.x (pyloidal is one of them, but I can fix that very easily). If somebody can figure out what's causing test_poincare to fail, we'll be ready to upgrade.
NumPy 2.0 was released a couple of days ago :tada:
You can test compatibility by installing into a fresh virtual environment as so:
We're mostly compatible, with just two issues:
np.asfarray(x)
withnp.asarray(x, dtype=float)
-- a very easy fix. The vim/sed command:%s/asfarray(\(.*\))/asarray(\1, dtype=float)/g
will do it automatically.test_poincare
is failing, which compares an array against golden answer data stored in a.npy
file. The arrays are almost identical for the first 22,000 elements, but start disagreeing in the last 8,000. No idea what's going on there.For now, installing Pyrokinetics will only grab NumPy versions up to 1.26, as some of our dependencies require
numpy == 1.x
(pyloidal
is one of them, but I can fix that very easily). If somebody can figure out what's causingtest_poincare
to fail, we'll be ready to upgrade.