deepmodeling

Guide for adding new devices to Uni-Lab-OS (接入新设备). Uses @device decorator + AST auto-scanning instead of manual YAML. Walks through device category, communication protocol, driver creation with decorators, and graph file setup. Use when the user wants to add/integrate a new device, create a device driver, write a device class, or mentions 接入设备/添加设备/设备驱动/物模型.

Operate Virtual Workbench via REST API — prepare materials, move to heating stations, start heating, move to output, transfer resources. Use when the user mentions virtual workbench, virtual_workbench, 虚拟工作台, heating stations, material processing, or workbench operations.

Query backend workflow list, aggregate all tags, and filter workflows by domain/scenario requirements using tags. Use when the user wants to search workflows, find workflows by tags, list available workflow tags, filter workflows by category/domain/scenario, or mentions 工作流筛选/标签查询/workflow tags/按领域查找工作流.

Guide for adding new resources (materials, bottles, carriers, decks, warehouses) to Uni-Lab-OS (添加新物料/资源). Uses @resource decorator for AST auto-scanning. Covers Bottle, Carrier, Deck, WareHouse definitions. Use when the user wants to add resources, define materials, create a deck layout, add bottles/carriers/plates, or mentions 物料/资源/resource/bottle/carrier/deck/plate/warehouse.

Guide for adding new workstations to Uni-Lab-OS (接入新工作站). Uses @device decorator + AST auto-scanning. Walks through workstation type, sub-device composition, driver creation, deck setup, and graph file. Use when the user wants to add a workstation, create a workstation driver, configure a station with sub-devices, or mentions 工作站/工站/station/workstation.

Batch insert reagents into Uni-Lab platform — add chemicals with CAS, SMILES, supplier info. Use when the user wants to add reagents, insert chemicals, batch register reagents, or mentions 录入试剂/添加试剂/试剂入库/reagent.

Batch submit experiments (notebooks) to the Uni-Lab cloud platform (leap-lab) — list workflows, generate node_params from registry schemas, submit multiple rounds, check notebook status. Use when the user wants to submit experiments, create notebooks, batch run workflows, check experiment status, or mentions 提交实验/批量实验/notebook/实验轮次/实验状态.

Create a skill for any Uni-Lab device by extracting action schemas from the device registry. Use when the user wants to create a new device skill, add device API documentation, or set up action schemas for a device.

Train DeePMD-kit models with progressive disclosure. Use when the user wants to train a DeePMD-kit potential, prepare an input.json, choose between model families such as se_e2_a/DeepPot-SE and DPA3, run `dp train`, monitor learning curves, freeze checkpoints, or test trained models. Start with model selection and read only the selected model reference under `models/` when model-specific configuration is needed.

Fine-tune a DPA3 model in DeePMD-kit using the PyTorch backend. Use when the user wants to adapt a pre-trained DPA3 model to a new downstream dataset. Supports fine-tuning from a self-trained DPA3 model (.pt checkpoint), from a multi-task pre-trained model, or from a built-in pretrained model downloaded via `dp pretrained download` (e.g., DPA-3.1-3M, DPA-3.2-5M, DPA-3.3-1M). Covers single-task and multi-task fine-tuning workflows.

Run Python inference with DeePMD-kit models using the DeepPot API. Use when the user wants to load a trained/frozen DeePMD model (.pth or .pb) or a built-in pretrained model (e.g., DPA-3.2-5M) in Python, predict energy/force/virial for atomic configurations, evaluate descriptors, or calculate model deviation between multiple models. Also covers using `dp test` CLI for batch evaluation against labeled data.

A tool and knowledge base for running molecular dynamics (MD) simulations in LAMMPS with the DeePMD-kit plugin. It handles input script preparation, ensemble selection (NVE/NVT/NPT), and job execution via `uv` or offline binaries. USE WHEN you need to set up, write, explain, or execute a LAMMPS molecular dynamics simulation using a DeePMD machine learning potential (e.g., `graph.pb`).

Diagnose gradient flow issues in training, especially for compiled models (torch.compile/make_fx). Systematically isolates which loss components (energy, force, virial) contribute gradients to which parameters, and identifies where the gradient chain breaks.

Guides through adding a new descriptor type to deepmd-kit. Covers implementing in dpmodel (array-API-compatible), wrapping for JAX/pt_expt backends, hard-coding for PT/PD, registering arguments, and writing all required tests.

Create and install dpdata plugins (especially custom Format readers/writers) using Format.register(...) and pyproject.toml entry_points under 'dpdata.plugins'. Use when extending dpdata with new formats or distributing plugins as separate Python packages.

Use dpdata Python Driver plugins to label systems (energies/forces/virials) via System.predict(), list available drivers, and build Driver objects (ase/deepmd/gaussian/sqm/hybrid). Use when working with dpdata Python API (not CLI) and you need driver-based energy/force prediction, plugin registration keys, or examples of using dpdata with ASE calculators or DeePMD models.

Minimize geometries with dpdata minimizer plugins via System.minimize(), including how minimizers relate to drivers (ASEMinimizer needs a dpdata Driver) and how to list supported minimizers (ase/sqm). Use when doing geometry optimization/minimization through dpdata Python API.

Convert and manipulate atomic simulation data formats using dpdata CLI. Use when converting between DFT/MD output formats (VASP, LAMMPS, QE, CP2K, Gaussian, ABACUS, etc.), preparing training data for DeePMD-kit, or working with DeePMD formats. Supports 50+ formats including deepmd/raw, deepmd/comp, deepmd/npy, deepmd/hdf5.

Prepare, explain, validate, and run DP-GEN simplify workflows for reducing repeated or redundant DeepMD datasets. Use when the user wants to generate or modify `param.json` and `machine.json`, run `dpgen simplify param.json machine.json`, organize repeated simplify experiments, or inspect simplify outputs.