Can't run the default DP-GEN example; KeyError: 'elements'

[aFrogAsks]

Good afternoon for everyone!

After installation of DP-GEN I try to run the default example (https://dp-public.oss-cn-beijing.aliyuncs.com/community/dpgen_example.tar.xz) to check if it works normal.

I go to the ../dpgen_example/run directory and then use

dpgen init_bulk param.json machine.json

It gives me the result:

DeepModeling

Version: 0.13.1
Path: ../.local/lib/python3.10/site-packages/dpgen

Dependency

 numpy     1.26.1   ../sw/slc7_x86-64/Python/v3.10.13/lib/python3.10/site-packages/numpy
dpdata     0.2.24   ../.local/lib/python3.10/site-packages/dpdata

pymatgen unknown version or path
monty 2025.3.3 ../.local/lib/python3.10/site-packages/monty
ase 3.24.0 ../.local/lib/python3.10/site-packages/ase
paramiko 3.5.1 ../.local/lib/python3.10/site-packages/paramiko
custodian 2025.5.12 ../.local/lib/python3.10/site-packages/custodian

Reference

Please cite:
Yuzhi Zhang, Haidi Wang, Weijie Chen, Jinzhe Zeng, Linfeng Zhang, Han Wang, and Weinan E,
DP-GEN: A concurrent learning platform for the generation of reliable deep learning
based potential energy models, Computer Physics Communications, 2020, 107206.

Description

Traceback (most recent call last):
File "../.local/bin/dpgen", line 8, in
sys.exit(main())
File "../.local/lib/python3.10/site-packages/dpgen/main.py", line 255, in main
args.func(args)
File "../.local/lib/python3.10/site-packages/dpgen/data/gen.py", line 1463, in gen_init_bulk
out_dir = out_dir_name(jdata)
File "../.local/lib/python3.10/site-packages/dpgen/data/gen.py", line 79, in out_dir_name
elements = jdata["elements"]
KeyError: 'elements'

I have no idea how to fix this.

Could anyone help me, please?

Thank you in advance

[njzjz-bot]

The command is using the concurrent-learning example's run/param.json as an initialization parameter file. Those are different schemas.

dpgen init_bulk requires an initialization parameter file containing keys such as elements, super_cell, and md_nstep; the code accesses jdata["elements"] immediately when constructing the output directory (source). That is why the tutorial's run-stage param.json raises KeyError: 'elements'.

If you are already in the downloaded example's run/ directory and want to start concurrent learning, use:

dpgen run param.json machine.json

If instead you want to generate the initial bulk dataset, run dpgen init_bulk with the parameter file supplied for the init workflow, after updating its structure, elements, INCAR/POTCAR paths, and machine settings. Do not add an arbitrary elements field to the run-stage file; use the correct file for the chosen subcommand.

Coding agent: Codex
Codex version: codex-cli 0.149.0
Model: gpt-5.6-sol
Reasoning effort: xhigh