Open BauerMarco opened 2 months ago
If you rebase this against the current master
branch, the pipeline should (hopefully) run.
Done
Now the clang-formatter should be happy too
I still don't fully understand the purpose of this PR.
What do you mean by "not accessible" from the adcc
module?
In your project, you could just import from sub-modules directly, e.g.,
from adcc.workflow import validate_state_parameters
rather than exposing, e.g., validate_state_parameters
via adcc/__init__.py
.
Am I missing something?
This is true.
The important part in this PR is the cpp part, where I introduced the new function set_lt_scalar, which is something I cannot hack together with a few lines of code outside of adcc.
Since I needed the modules for the external package, I thought they should be exposed through adcc/__init__
, but if you don't agree, I can reduce it to the change in libadcc.
I've addressed your points in the new commit and also extended the initialization from a single scalar to a 1d array.
I've searched through everything, which is imported to the python side from libadcc, but could't find anything to initialize libtensor arrays of arbitrary dimension and length from the python side. If there is anything, I've missed it.
I rewrote pull request #153 as a separate repository called polaritonic_adcc on my home github, where I noticed that some functionalities I needed were not accessible from the adcc module. Before I simply copy, paste and reference a lot of the adcc code, it would be much easier if this pull request got accepted, in case somebody else also wants to build a similar project.
I also had to add an additional function to the libadcc site, which lets you set a libtensor array with one entry from the python side. I didn't really see an other option, but if you guys have a good idea for a workaround, I could remove it from this pull request.