fix: Filter SIESTA pseudopotentials by system atom types

[Copilot]

SIESTA input generation failed with AssertionError when systems contained a subset of elements from fp_pp_files (e.g., pure C systems when both C.psf and H.psf were provided).

Changes

• dpgen/generator/run.py: Filter fp_pp_files to match system atom types before passing to make_siesta_input, following PWSCF pattern:

  pps = []
  for iii in sys_data["atom_names"]:
      pps.append(fp_pp_files[jdata["type_map"].index(iii)])
  ret = make_siesta_input(sys_data, pps, fp_params)

• dpgen/tools/relabel.py: Apply same filtering in make_siesta helper and fix parameter passing

• tests/generator/test_make_fp.py: Add TestMakeFPSIESTASubsetElements to validate subset element handling

The assertion assert ntypes == len(pps) in _make_siesta_02_species now receives correctly filtered pseudopotentials matching the system's atom types.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>[BUG] Un-necessary assertion ntypes < len(pps) when making siesta input files</issue_title>
<issue_description>### Bug summary

Hi!
I'm running DPGEN for systems containing element 'C' and 'H' with SIESTA as the DFT engine.
In this task, 'sys_configs' includes pure 'C' systems and systems with both 'C' and 'H'. The 'fp_pp_files = ["C.psf", "H.psf"]'. However, an unexpected error occurs:

INFO:dpgen:-------------------------iter.000015 task 06--------------------------
INFO:dpgen:system 015 candidate :      1 in     11   9.09 %
INFO:dpgen:system 015 failed    :     10 in     11  90.91 %
INFO:dpgen:system 015 accurate  :      0 in     11   0.00 %
INFO:dpgen:system 015 accurate_ratio:   0.0000    thresholds: 1.0000 and 1.0000   eff. task min and max   -1   20   number of fp tasks:      1
Traceback (most recent call last):
  File "/home/ubuntu/miniconda3/envs/dpgen/bin/dpgen", line 7, in <module>
    sys.exit(main())
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/main.py", line 255, in main
    args.func(args)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/run.py", line 5721, in gen_run
    run_iter(args.PARAM, args.MACHINE)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/run.py", line 5070, in run_iter
    make_fp(ii, jdata, mdata)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/run.py", line 4011, in make_fp
    make_fp_calculation(iter_index, jdata, mdata)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/run.py", line 4034, in make_fp_calculation
    make_fp_siesta(iter_index, jdata)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/run.py", line 3723, in make_fp_siesta
    ret = make_siesta_input(sys_data, fp_pp_files, fp_params)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/lib/siesta.py", line 137, in make_siesta_input
    ret += _make_siesta_02_species(sys_data, fp_pp_files)
  File "/home/ubuntu/miniconda3/envs/dpgen/lib/python3.10/site-packages/dpgen/generator/lib/siesta.py", line 50, in _make_siesta_02_species
    assert ntypes == len(pps)
AssertionError

The root reason seems to be a unnecessary assert check in dpgen/generator/lib/siesta.py:_make_siesta_02_species. The 'ntypes' is just necessary to be a subset of 'pps'.

DP-GEN Version

Version: 0.13.2.dev62+g037d6b819

Platform, Python Version, Remote Platform, etc

No response

Input Files, Running Commands, Error Log, etc.

'param.json'

{
     "type_map": ["C","H"],
     "mass_map": [12,1],
     "init_data_prefix": "../init/deepmd_data",
     "init_data_sys": [
  "C110H0", "C112H0", "C123H3", "C126H0", "C128H0",
  "C55H3", "C62H0", "C64H0", "C71H3", "C82H0",
  "C84H0", "C87H3", "C111H3", "C122H0", "C124H0",
  "C127H3", "C54H0", "C56H0", "C63H3", "C70H0",
  "C72H0", "C83H3", "C86H0", "C88H0"
],
     "sys_configs_prefix": "/home/ubuntu/run/CNT_single_1560/init",
     "sys_configs": [
  ["POSCARS_OUT/1000k/6_6/5775/POSCAR"],
  ["POSCARS_OUT/1000k/6_6/DV/POSCAR"],
  ["POSCARS_OUT/1000k/6_6/MVH/POSCAR"],
  ["POSCARS_OUT/1000k/6_6/P/POSCAR"],
  ["POSCARS_OUT/1000k/7_1/5775/POSCAR"],
  ["POSCARS_OUT/1000k/7_1/DV/POSCAR"],
  ["POSCARS_OUT/1000k/7_1/MVH/POSCAR"],
  ["POSCARS_OUT/1000k/7_1/P/POSCAR"],
  ["POSCARS_OUT/1000k/9_0/5775/POSCAR"],
  ["POSCARS_OUT/1000k/9_0/DV/POSCAR"],
  ["POSCARS_OUT/1000k/9_0/MVH/POSCAR"],
  ["POSCARS_OUT/1000k/9_0/P/POSCAR"],
  ["POSCARS_OUT/300k/6_6/5775/POSCAR"],
  ["POSCARS_OUT/300k/6_6/DV/POSCAR"],
  ["POSCARS_OUT/300k/6_6/MVH/POSCAR"],
  ["POSCARS_OUT/300k/6_6/P/POSCAR"],
  ["POSCARS_OUT/300k/7_1/5775/POSCAR"],
  ["POSCARS_OUT/300k/7_1/DV/POSCAR"],
  ["POSCARS_OUT/300k/7_1/MVH/POSCAR"],
  ["POSCARS_OUT/300k/7_1/P/POSCAR"],
  ["POSCARS_OUT/300k/9_0/5775/POSCAR"],
  ["POSCARS_OUT/300k/9_0/DV/POSCAR"],
  ["POSCARS_OUT/300k/9_0/MVH/POSCAR"],
  ["POSCARS_OUT/300k/9_0/P/POSCAR"],
  ["POSCARS_OUT/50k/6_6/5775/POSCAR"],
  ["POSCARS_OUT/50k/6_6/DV/POSCAR"],
  ["POSCARS_OUT/50k/6_6/MVH/POSCAR"],
  ["POSCARS_OUT/50k/6_6/P/POSCAR"],
  ["POSCARS_OUT/50k/7_1/5775/POSCAR"],
  ["POSCARS_OUT/50k/7_1/DV/POSCAR"],
  ["POSCARS_OUT/50k/7_1/MVH/POSCAR"],
  ["POSCARS_OUT/50k/7_1/P/POSCAR"],
  ["POSCARS_OUT/50k/9_0/5775/POSCAR"],
  ["POSCARS_OUT/50k/9_0/DV/POSCAR"],
  ["POSCARS_OUT/50k/9_0/MVH/POSCAR"],
  ["POSCARS_OUT/50k/9_0/P/POSCAR"],
  ["POSCARS_OUT/700k/6_6/5775/POSCAR"],
  ["POSCARS_OUT/700k/6_6/DV/POSCAR"],
  ["POSCARS_OUT/700k/6_6/MVH/POSCAR"],
  ["POSCARS_OUT/700k/6_6/P/POSCAR"],
  ["POSCARS_OUT/700k/7_1/5775/POSCAR"],
  ["POSCARS_OUT/700k/7_1/DV/POSCAR"],
  ["POSCARS_OUT/700k/7_1/MVH/POSCAR"],
  ["POSCARS_OUT/700k/7_1/P/POSCAR"],
  ["POSCARS_OUT/700k/9_0/5775/POSCAR"],
  ["POSCARS_OUT/700k/9_0/DV/POSCAR"],
  ["POSCARS_OUT/700k/9_0/MVH/POSCAR"],
  ["POSC...

</details>



<!-- START COPILOT CODING AGENT SUFFIX -->

- Fixes deepmodeling/dpgen#1859

<!-- START COPILOT CODING AGENT TIPS -->
---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more [Copilot coding agent tips](https://gh.io/copilot-coding-agent-tips) in the docs.

[njzjz-bot]

Verdict: Changes requested. Filtering SIESTA pseudopotentials by the system elements is correct in the main make_fp_siesta path, but the relabel initial-data path still has a deterministic UnboundLocalError. The new test does not cover that branch, and pre-commit is failing.

Note: The Codex quota is about to reset, so I am using the remaining tokens to review all open PRs in this repository.

Coding agent: Codex
Codex version: codex-cli 0.144.6
Model: gpt-5.6-sol
Reasoning effort: xhigh

[njzjz-bot]

[Blocking] When fp_style == "siesta", the preceding mutually exclusive PWSCF branch does not run, so the local variable fp_params is unbound. The SIESTA path in create_init_tasks() therefore deterministically raises UnboundLocalError. Read user_fp_params/fp_params inside the SIESTA branch and add a relabel init-task regression test. For example:

Suggested change

	make_siesta(".", fp_params, fp_pp_path, fp_pp_files, type_map)
	elif fp_style == "siesta":
	if "user_fp_params" in fp_jdata:
	fp_params = fp_jdata["user_fp_params"]
	else:
	fp_params = fp_jdata["fp_params"]
	make_siesta(".", fp_params, fp_pp_path, fp_pp_files, type_map)