Clean up unused methods in region.py and preprocess.py

[njzjz]

source/tests/pt/model/test_region.py::TestLegacyRegion::test_inter_to_inter
  /opt/hostedtoolcache/Python/3.11.9/x64/lib/python3.11/site-packages/torch/utils/_device.py:78: UserWarning: Using torch.cross without specifying the dim arg is deprecated.
  Please either pass the dim explicitly or simply use torch.linalg.cross.
  The default value of dim will change to agree with that of linalg.cross in a future release. (Triggered internally at ../aten/src/ATen/native/Cross.cpp:62.)
    return func(*args, **kwargs)

[njzjz]

Affected:

deepmd-kit/deepmd/pt/utils/region.py

Lines 73 to 77 in fd2daeb

	c_yz = torch.cross(cell[1], cell[2])
	_h2yz = volume / torch.linalg.norm(c_yz)
	c_zx = torch.cross(cell[2], cell[0])
	_h2zx = volume / torch.linalg.norm(c_zx)
	c_xy = torch.cross(cell[0], cell[1])

deepmd-kit/deepmd/pt/utils/preprocess.py

Lines 28 to 32 in fd2daeb

	c_yz = torch.cross(boxt[1], boxt[2])
	self._h2yz = self.volume / torch.linalg.norm(c_yz)
	c_zx = torch.cross(boxt[2], boxt[0])
	self._h2zx = self.volume / torch.linalg.norm(c_zx)
	c_xy = torch.cross(boxt[0], boxt[1])

[wanghan-iapcm]

all functions in preprocess except compute_smooth_weight can be removed.

[njzjz]

@wanghan-iapcm @iProzd could we find and remove all dead codes at once?

[wanghan-iapcm]

May not necessarily do it at once ? or is there tools for automatically removing the dead codes?

[njzjz]

May not necessarily do it at once ? or is there tools for automatically removing the dead codes?

Indeed we can find them using Codecov, assuming all used functions are tested.

[njzjz]

I see preprocess.py has a 93% coverage, though

[wanghan-iapcm]

good idea. I am wondering how much codes that are not dead are not covered by the UTs.

[njzjz]

I just found _to_face_distance in region.py is also not used at all, but left there for an unclear reason.

[njzjz]

Since both code blocks are not used, I unassigned myself and moved from mustfix to enhance. cc @wanghan-iapcm @iProzd

[njzjz]

I've used deadcode to find out all used functions. Could you take a look? @wanghan-iapcm @iProzd

deepmd/dpmodel/output_def.py:416:0: DC02 Function `check_deriv` is never used
deepmd/pt/infer/deep_eval.py:595:0: DC02 Function `eval_model` is never used
deepmd/pt/model/model/transform_output.py:111:0: DC02 Function `get_atom_axis` is never used
deepmd/pt/model/network/init.py:425:0: DC02 Function `xavier_normal_` is never used
deepmd/pt/utils/nlist.py:175:0: DC02 Function `build_directional_neighbor_list` is never used
deepmd/pt/utils/preprocess.py:241:0: DC02 Function `make_env_mat` is never used
deepmd/pt/utils/region.py:71:0: DC02 Function `_to_face_distance` is never used
deepmd/tf/nvnmd/utils/network.py:31:0: DC02 Function `matmul2_qq` is never used
deepmd/tf/nvnmd/utils/network.py:45:0: DC02 Function `matmul3_qq` is never used
deepmd/tf/train/trainer.py:76:0: DC02 Function `_is_subdir` is never used
deepmd/tf/utils/graph.py:157:0: DC02 Function `get_embedding_net_nodes` is never used
deepmd/tf/utils/graph.py:285:0: DC02 Function `get_embedding_net_variables` is never used
deepmd/tf/utils/graph.py:337:0: DC02 Function `get_fitting_net_nodes` is never used
deepmd/tf/utils/graph.py:375:0: DC02 Function `get_fitting_net_variables` is never used
deepmd/utils/econf_embd.py:207:0: DC02 Function `make_econf_embedding` is never used
deepmd/utils/econf_embd.py:217:0: DC02 Function `print_econf_embedding` is never used