ATST-Tools is an ASE transition-state workflow toolkit for ABACUS and
DeePMD-kit. Run calculations from YAML with the atst command, or use the
Python API to integrate the same workflows
into another program.
atst run config.yamlChinese guide: 中文 · GitHub · Gitee. Examples: GitHub · Gitee.
| Workflow | calculation.type |
|---|---|
| NEB / DyNEB and adaptive-image NEB | neb, autoneb |
| Dimer, Sella and CCQN saddle searches | dimer, sella, ccqn |
| Double-ended to single-ended TS search | d2s |
| Structure relaxation | relax |
| Vibrations and thermochemistry | vibration |
| Intrinsic reaction coordinates | irc |
| ASE-driven or ABACUS-native molecular dynamics | md |
| Experimental Direct MaxFlux path search | dmf |
| ABACUS constant-potential single points and serial scans | constant_potential |
Constant-potential fixed-cell relaxation, serial NEB and ASE-driven MD
use the explicit compensated_gate boundary. See the
configuration reference
for setup and supported boundaries.
Local commands prepare paths and ABACUS inputs, inspect trajectories, and
summarize existing results. See the CLI reference.
A transition-state candidate can be screened with atst vibration validate
using saved vibration results; workflow completion alone does not establish a
validated TS. Current support and experimental limits are summarized in the
feature matrix.
Requires Python 3.10 or later:
pip install atst-tools
atst --version| Use | Installation |
|---|---|
| Serial workflows, including serial NEB/AutoNEB | pip install atst-tools |
| DeePMD-kit calculator | pip install "atst-tools[dp]" |
| Image-parallel NEB/AutoNEB | pip install "atst-tools[parallel]" |
The default installation includes abacuslite>=1.2.1,<2 and Sella; it does not
install mpi4py. ABACUS calculations also need an ABACUS executable and the
pseudopotential/orbital files referenced by the configuration. DP calculations
need a compatible model. Image-parallel runs require a matching MPI runtime;
DMF additionally requires cyipopt and IPOPT. See the
MPI setup guide for
image-parallel setup and the examples guide for
backend inputs.
Obtain examples at the tag matching your installed version (shown for 2.3.3):
git clone --branch v2.3.3 https://github.com/QuantumMisaka/atst-tools.git
cd atst-toolsFirst try a local path-preparation command; it needs no ABACUS, DP model or GPU:
cd examples/09_lightweight_cli
atst neb make inputs/init.xyz inputs/final.xyz 3 -o first_chain.traj --method linear
atst neb summary inputs/neb_result.extxyz --n-max 1 --tail 5
cd ../..first_chain.traj contains five initial-guess frames. To inspect a calculation
configuration, enter its example directory so relative input paths resolve:
cd examples/06_relax_H2-Au
atst config validate config.yaml --print-normalized
atst run --dry-run config.yaml
# After configuring the ABACUS runtime and compute resources:
atst run config.yamlThis example uses four ABACUS MPI ranks, the GPU cusolver solver and a
one-step optimization limit. Validation and dry-run do not launch ABACUS;
check the resulting convergence status separately. Follow the
examples guide to inspect results and choose other
methods or DP variants. A complete walkthrough is also available in the
Chinese guide.
| Task | Reference |
|---|---|
| Configure a calculation | Configuration reference · All YAML fields |
| Use CLI commands | CLI reference |
| Configure ABACUS input/output and parallelism | ABACUS wrapper guide |
| Embed workflows or use the JSON process runner | Python API |
| Check support and releases | Feature matrix · 2.3.3 release notes |
| Contribute | Developer handover · Source architecture |
The primary documentation is in English, with a Chinese user guide. The documentation index organizes user, developer and project status material. GitHub is the development source; Gitee is a manually updated mirror. PyPI distributes the Python package; obtain examples from the repository.
If you publish work that uses the transition-state workflows in ATST-Tools, please cite the underlying methods alongside ATST-Tools itself:
- ATST-Tools and the double-to-single (D2S) transition-state workflow: Liu, Z.-Q.; Deng, Z.; Zhao, H.; Wang, H.; Chen, M.; Jiang, H. FT²DP: Large Atomic Model Fine-Tuned Machine Learning Potential for Accelerating Atomistic Simulation of Iron-Based Fischer–Tropsch Synthesis. J. Mater. Inform. 5 (2), 27 (2025). https://doi.org/10.20517/jmi.2024.105
- Sella saddle-point search: Hermes, E. D.; Sargsyan, K.; Najm, H. N.; Zádor, J. Sella, an Open-Source Automation-Friendly Molecular Saddle Point Optimizer. J. Chem. Theory Comput. 18 (11), 6974-6988 (2022). https://doi.org/10.1021/acs.jctc.2c00395
- CCQN (cone-shaped constrained quasi-Newton) transition-state optimization: Wu, Y.; Wang, H. Cone-Shaped Constrained Quasi-Newton Method: Efficient and Robust Single-Ended Transition State Optimization Algorithm. J. Chem. Theory Comput. 21 (18), 9054-9065 (2025). https://doi.org/10.1021/acs.jctc.5c01015
LGPL-v3 License.