chem-dft-orca-optimization
Run DFT geometry optimization (minimization or TS search) on a molecular structure using ORCA via SCINE/ReaDuct wrapper.
Skill by learningmatter-mit
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/learningmatter-mit-chem-dft-orca-optimization
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/chem-dft-orca-optimization
What's inside
- Skill name
chem-dft-orca-optimization- Description
- Run DFT geometry optimization (minimization or TS search) on a molecular structure using ORCA via SCINE/ReaDuct wrapper.
- License
- MIT
- Instructions
- 148 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Files (6)
- SKILL.md
- README.md · example/
- h2o.xyz · example/
- initial_structure.xyz · example/
- ts_guess.xyz · example/
- run_optimization.py · scripts/
SKILL.md (first lines)
# DFT Geometry Optimization with ORCA
## Goal
Optimize the geometry of a molecular structure at the DFT level using the ORCA quantum chemistry program. Supports two modes: **minimization** (finding the nearest local minimum) and **transition state (TS) optimization** (single-ended saddle point search). The calculation uses the SCINE/ReaDuct wrapper for robust optimizer management.
> [!IMPORTANT]
> This skill provides **single-ended TS optimization** only. For reaction pathway methods (NEB, IRC), consider using the MLIP-based [NEB skill](../chem-neb-barrier/SKILL.md) or [IRC skill](../chem-irc-verification/SKILL.md) with MLIP pre-screening, then refine with DFT. For advanced ORCA features, use the [advanced ORCA skill](../chem-dft-orca-advanced-calculation/SKILL.md).
## Background
Geometry optimization iteratively adjusts nuclear positions to minimize (or, for TS search, to find a first-order saddle point of) the potential energy surface $E(\mathbf{R})$. The SCINE/ReaDuct optimizer handles step control, coordinate transformations, and convergence criteria internally.
- **Minimization** seeks a stationary point where $\nabla E = 0$ and the Hessian has all positive eigenvalues.
- **TS optimization** seeks a first-order saddle point where $\nabla E = 0$ and the Hessian has exactly one negative eigenvalue.
## 1. Prerequisites
- **Environment:** `cpu` (commands run through `venv/run cpu ...`), which includes `scine_utilities`, `scine_readuct` (x86_64 only), and `ase`
- **ORCA binary:** The environment variable `ORCA_BINARY_PATH` must point to the ORCA executable
```bash
export ORCA_BINARY_PATH=/path/to/orca
```
- **Input structure:** A molecular structure file readable by ASE (`.xyz`, `.cif`, `.mol`, etc.)
- For **TS optimization:** Provide a reasonable TS guess geometry. Poor initial guesses will likely fail to converge to the correct saddle point.
## 2. Parameters
| Parameter | Default | Description |
|-----------|---------|-------------|
| `--structure` | (required) | Path to input structure file |
| `--opt_type` | `min` | `min` for minimization, `ts` for transition state search |
| `--charge` | `0` | Molecular charge |
| `--spin_multiplicity` | `1` | Spin multiplicity (2S+1) |
| `--functional` | `PBE` | DFT functional (e.g. `PBE`, `B3LYP`, `wB97X-V`) |
| `--basis_set` | `def2-SVP` | Basis set (e.g. `def2-SVP`, `def2-TZVP`) |
| `--dispersion` | None | Dispersion correction (e.g. `D3BJ`, `D4`) |
| `--solvation` | None | Implicit solvation model: `CPCM` or `SMD` |
| `--solvent` | None | Solvent name; required if `--solvation` is set |
| `--special_option` | `NOSOSCF` | ORCA special option passed to SCINE calculator. Set to empty string to disable. |
| `--nprocs` | `1` | Number of CPU cores for ORCA |
| `--convergence_max_iterations` | `200` | Maximum optimization steps |
| `--calculate_final_hessian` | off | Compute Hessian at optimized geometry (for TS verification) |
| `--calculator_settings` | None | Extra SCINE calculator settings as a JSON string (see below) |
| `--optimizer_settings` | None | Extra ReaDuct optimizer kwargs as a JSON string (see below) |
| `--output_dir` | auto | Output directory |
## 3. Running an Optimization
### Geometry minimization
```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/run_optimization.py \
--structure molecule.xyz \
--functional B3LYP \
--basis_set def2-TZVP \
--dispersion D3BJ \
--nprocs 4 \
--output_dir research/my_project/optimization
```
### Transition state optimization
```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/run_optimization.py \
--structure ts_guess.xyz \
--opt_type ts \
--functional B3LYP \
--basis_set def2-TZVP \
--dispersion D3BJ \
--calculate_final_hessian \
--nprocs 4 \
--output_dir research/my_project/ts_optimization
```
### With extra settings (calculator + optimizer)
For settings not exposed as dedicated flags, pass JSON strings. `--calculator_settings` applies to the SCINE/ORCA calculator, `--optimizer_settings` applies to the ReaDuct optimization task. SCINE is strict about types, so JSON ensures values are passed with the correct type (int, float, string).
```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/run_optimization.py \
--structure molecule.xyz \
--functional B3LYP \
--basis_set def2-TZVP \
--calculator_settings '{"max_scf_iterations": 128}' \
--optimizer_settings '{"convergence_delta_value": 1e-6}' \
--output_dir research/my_project/opt_custom
```
### With implicit solvation
```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/run_optimization.py \
--structure molecule.xyz \
--functional PBE0 \
--basis_set def2-TZVP \
--solvation SMD \
--solvent water \
--nprocs 4 \
--output_dir research/my_project/opt_solvated
```
## 4. Output Files
- `optimization_results.json`: Structured results containing:
- `converged`: Boolean indicating whether the optimization converged
- `final_energy_hartree`, `final_energy_eV`: Final electronic energy
- `final_max_force_eV_per_Ang`, `final_rms_force_eV_per_Ang`: Residual force information
- `opt_type`: Whether this was a minimization or TS search
- If `--calculate_final_hessian` was used: `hessian_eV_per_Ang2`, `hessian_wave_numbers_cm-1`, and `n_imaginary_modes`
- All input parameters for reproducibility
- `initial_structure.xyz`: Copy of the input structure
- `optimized_structure.xyz`: The optimized geometry
## 5. Interpreting Results
### Minimization
- Check `converged: true` in the results JSON.
- Residual forces should be small (max force < 0.01 eV/A for typical convergence).From learningmatter-mit/AtomisticSkills, MIT.
More from learningmatter-mit/AtomisticSkills
- 0.681score
drug-mmpbsa-gbsa
learningmatter-mit
Compute single-trajectory MM-GBSA and / or MM-PBSA binding free energy estimates from a protein-ligand MD trajectory.
SkillAFreshness: A★ 183Claude Code +9 - 0.681score
drug-pocket-detection
learningmatter-mit
Identify and rank ligandable pockets on a protein structure or model using geometry (fpocket) or an ML predictor (P2Rank).
SkillAFreshness: A★ 183Claude Code +9 - 0.681score
drug-binding-site-definition
learningmatter-mit
Define a docking search box (center coordinates + box dimensions in Angstroms) from a co-crystal ligand, binding-site residues, or a saved JSON specification.
SkillAFreshness: A★ 183Claude Code +9 - 0.681score
drug-complex-system-builder
learningmatter-mit
Build a solvated, charge-neutralized protein-ligand complex for OpenMM molecular dynamics simulation.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
chem-hazard-toxicity
learningmatter-mit
Extract explicit safety warnings, GHS classifications, LD50 toxicity profiles, and acute oral toxicity triage from PubChem PUG VIEW.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
drug-redocking-rmsd
learningmatter-mit
Compute symmetry-corrected heavy-atom RMSD between docked poses and a reference crystal ligand to validate docking protocols.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
chem-dft-orca-singlepoint
learningmatter-mit
Run a DFT or Coupled Cluster single-point energy calculation (with optional gradients/Hessian) on a molecular structure with ORCA through SCINE wrapper.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
drug-pose-validation
learningmatter-mit
Validate docked or generated ligand poses for physical plausibility using PoseBusters, filtering out chemically invalid or clashing poses before downstream…
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
chem-db-mof
learningmatter-mit
Query multiple MOF databases (QMOF via MPContribs; ARC-MOF DB7/Majumdar et al.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
chem-bond-dissociation
learningmatter-mit
Calculate homolytic and heterolytic bond dissociation energies (BDEs) for all single bonds in a molecule using MLIPs with RDKit fragmentation.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
drug-protein-ligand-md
learningmatter-mit
Run a protein-ligand MD simulation in OpenMM with energy minimization, restrained equilibration, and production NPT, producing trajectory and checkpoint files…
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
chem-sorption-relax
learningmatter-mit
Prepares supercells for porous frameworks based on minimum interplanar distance and relaxes them using standard MLIP relaxation tools.
SkillAFreshness: A★ 183Claude Code +9
Signals
- Stars
- 183
- Forks
- 31
- Open issues
- 0
- Last commit (this folder)
- Oct 5, 2026 (22 days ago)
- Last release
- Oct 6, 2026 (22 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Website
- arxiv.org/abs/2605.24002
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from learningmatter-mit
- 0.681score
drug-binding-site-definition
learningmatter-mit
Define a docking search box (center coordinates + box dimensions in Angstroms) from a co-crystal ligand, binding-site residues, or a saved JSON specification.
SkillAFreshness: A★ 183Claude Code +9 - 0.681score
drug-complex-system-builder
learningmatter-mit
Build a solvated, charge-neutralized protein-ligand complex for OpenMM molecular dynamics simulation.
SkillAFreshness: A★ 183Claude Code +9 - 0.681score
drug-pocket-detection
learningmatter-mit
Identify and rank ligandable pockets on a protein structure or model using geometry (fpocket) or an ML predictor (P2Rank).
SkillAFreshness: A★ 183Claude Code +9 - 0.681score
drug-mmpbsa-gbsa
learningmatter-mit
Compute single-trajectory MM-GBSA and / or MM-PBSA binding free energy estimates from a protein-ligand MD trajectory.
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
drug-protein-ligand-md
learningmatter-mit
Run a protein-ligand MD simulation in OpenMM with energy minimization, restrained equilibration, and production NPT, producing trajectory and checkpoint files…
SkillAFreshness: A★ 183Claude Code +9 - 0.678score
chem-hazard-toxicity
learningmatter-mit
Extract explicit safety warnings, GHS classifications, LD50 toxicity profiles, and acute oral toxicity triage from PubChem PUG VIEW.
SkillAFreshness: A★ 183Claude Code +9
Cite this page
chem-dft-orca-optimization skill. Agent Mart. https://agentmart.dev/skills/learningmatter-mit-chem-dft-orca-optimization (last updated 2026-10-24)