Configuration Reference

Date:

2026-03-13

1 Configuration Reference

Complete reference for config/config.yaml parameters.

2 System Definitions

2.1 systems

Dictionary of chemical systems to process.

2.1.1 Structure

systems:
  system_name:
    reactant: structures/reactant.con  # Path to reactant .con file
    product: structures/product.con    # Path to product .con file
    use_ira: true
    saddle: structures/saddle.con  # Optional: for 2D plots

2.1.2 Parameters

Parameter	Type	Required	Default	Description
reactant	string	Yes		Path to reactant .con file
product	string	Yes		Path to product .con file
use_ira	boolean	No	true	Enable IRA alignment
saddle	string	No		Path to saddle point (for 2D plots)

2.1.3 Example

systems:
  hcn_isom:
    reactant: /data/hcn.con
    product: /data/hnc.con
    use_ira: true

  proton_transfer:
    reactant: /data/pt_reactant.con
    product: /data/pt_product.con
    use_ira: true
    saddle: /data/pt_saddle.con

3 Model Configuration

3.1 model

Machine learning potential settings.

3.1.1 Parameters

Parameter	Type	Required	Default	Description
name	string	Yes	pet-mad-v1.1.0	Model filename
version	string	Yes	v1.1.0	HuggingFace version tag
type	string	No	pet-mad	Model type (pet-mad or upet)

3.1.2 Available Models

• pet-mad-v1.1.0: PET-MAD v1.1.0 (recommended)

• pet-mad-v1.0.0: PET-MAD v1.0.0 (legacy)

• upet: Universal PET (experimental)

3.1.3 Example

model:
  name: pet-mad-v1.1.0
  version: v1.1.0
  type: pet-mad

4 Compute Settings

4.1 compute

Hardware and device configuration.

4.1.1 Parameters

Parameter	Type	Required	Default	Options
device	string	No	cuda	cuda, cpu, mps

4.1.2 Device Selection

• cuda: NVIDIA GPU (recommended for performance)

• cpu: CPU-only (slower but universally available)

• mps: Apple Silicon (M1/M2/M3)

4.1.3 Example

compute:
  device: cuda

5 Alignment Parameters

5.1 alignment

IRA (Iterative Rotational Alignment) settings.

5.1.1 Parameters

Parameter	Type	Required	Default	Range	Description
kmax	float	No	1.8	1.0-14.0	Rotation matching cutoff (Å⁻¹)

5.1.2 Guidelines

• Small molecules (<20 atoms): 1.5-2.0 Å⁻¹

• Medium molecules (20-50 atoms): 2.0-3.0 Å⁻¹

• Difficult systems: 3.0-5.0 Å⁻¹

• 2D visualization: 14 Å⁻¹ (fixed)

5.1.3 Example

alignment:
  kmax: 1.8

6 NEB Parameters

6.1 neb

Complete NEB calculation settings.

6.1.1 neb.minimization

Endpoint geometry optimization.

Parameter	Type	Default	Description
max_iterations	int	2000	Maximum optimization steps
converged_force	float	0.0514221	Force threshold (eV/Å)
opt_method	string	lbfgs	Optimizer method
max_move	float	0.1	Maximum step size (Å)

6.1.2 neb.optimization

NEB calculation parameters.

Parameter	Type	Default	Description
images	int	18	Number of NEB images
max_iterations	int	1000	Maximum NEB iterations
converged_force	float	0.0514221	Force threshold (eV/Å)
opt_method	string	lbfgs	Optimizer method
max_move	float	0.1	Maximum step size (Å)

6.1.2.1 Energy-Weighted Springs

Parameter	Type	Default	Description
energy_weighted	bool	true	Enable energy weighting
ew_ksp_min	float	0.972	Minimum spring constant (eV/Ų)
ew_ksp_max	float	9.72	Maximum spring constant (eV/Ų)
ew_trigger	float	0.5	Energy threshold (eV)

6.1.2.2 Climbing Image

Parameter	Type	Default	Description
climbing_image_method	bool	true	Enable CI method
ci_after_rel	float	0.8	Relative convergence for CI

6.1.2.3 MMF Refinement

Parameter	Type	Default	Description
ci_mmf	bool	true	Enable MMF at CI
ci_mmf_nsteps	int	1000	MMF refinement steps

6.1.2.4 SIDPP Initializer

Parameter	Type	Default	Description
sidpp_growth_alpha	float	0.33	Path growth step size

6.1.3 neb.idpp

Legacy IDPP path generation (optional).

Parameter	Type	Default	Description
number_of_intermediate_imgs	int	10	Intermediate images

6.1.4 Example

neb:
  minimization:
    max_iterations: 2000
    converged_force: 0.0514221
    opt_method: lbfgs
    max_move: 0.1

  optimization:
    images: 18
    max_iterations: 1000
    converged_force: 0.0514221
    opt_method: lbfgs
    max_move: 0.1

    energy_weighted: true
    ew_ksp_min: 0.972
    ew_ksp_max: 9.72
    ew_trigger: 0.5

    climbing_image_method: true
    ci_after_rel: 0.8

    ci_mmf: true
    ci_mmf_nsteps: 1000

    sidpp_growth_alpha: 0.33

  idpp:
    number_of_intermediate_imgs: 10

7 Visualization Parameters

7.1 plotting

Plot generation settings.

7.1.1 General

Parameter	Type	Default	Description
plot_structures	bool	true	Overlay atomic structures
facecolor	string	white	Background color
title	string	""	Plot title prefix

7.1.2 Figure Settings

Parameter	Type	Default	Description
figsize	list	[8, 6]	Figure size (inches)
dpi	int	300	Output resolution
zoom_ratio	float	1.0	Structure zoom factor

7.1.3 Fonts

Parameter	Type	Default	Description
base	int	12	Base font size

7.1.4 Structure Overlays

Parameter	Type	Default	Description
draw_reactant.x	float	0.2	X position (normalized)
draw_reactant.y	float	0.1	Y position (normalized)
draw_reactant.rad	float	0.15	Circle radius
draw_product.x	float	0.8	X position
draw_product.y	float	0.1	Y position
draw_product.rad	float	0.15	Circle radius
draw_saddle.x	float	0.5	X position
draw_saddle.y	float	0.6	Y position
draw_saddle.rad	float	0.15	Circle radius

7.1.5 Advanced

Parameter	Type	Default	Description
ira_kmax	float	14	IRA kmax for 2D plots
aserot	float	0	ASE rotation (degrees)

7.1.6 Example

plotting:
  plot_structures: true
  facecolor: white
  title: "HCN → HNC"

  figure:
    figsize: [8, 6]
    dpi: 300
    zoom_ratio: 1.0

  fonts:
    base: 12

  draw_reactant:
    x: 0.2
    y: 0.1
    rad: 0.15

  draw_product:
    x: 0.8
    y: 0.1
    rad: 0.15

  draw_saddle:
    x: 0.5
    y: 0.6
    rad: 0.15

  ira_kmax: 14
  aserot: 0

8 Path Configuration

8.1 paths

Output directory structure.

8.1.1 Parameters

Parameter	Type	Default	Description
models	string	results/models	ML potential storage
endpoints	string	results/endpoints	Endpoint structures
neb	string	results/neb	NEB outputs
idpp	string	results/idpp	IDPP paths
plots	string	results/plots	Visualization outputs
cache	string	results/cache	Cache files

8.1.2 Example

paths:
  models: results/models
  endpoints: results/endpoints
  neb: results/neb
  idpp: results/idpp
  plots: results/plots
  cache: results/cache

9 Complete Configuration Example

systems:
  hcn_isom:
    reactant: /data/hcn.con
    product: /data/hnc.con
    use_ira: true

model:
  name: pet-mad-v1.1.0
  version: v1.1.0
  type: pet-mad

compute:
  device: cuda

alignment:
  kmax: 1.8

neb:
  minimization:
    max_iterations: 2000
    converged_force: 0.0514221
    opt_method: lbfgs
    max_move: 0.1

  optimization:
    images: 18
    max_iterations: 1000
    converged_force: 0.0514221
    energy_weighted: true
    ew_ksp_min: 0.972
    ew_ksp_max: 9.72
    climbing_image_method: true
    ci_mmf: true
    sidpp_growth_alpha: 0.33

plotting:
  plot_structures: true
  figure:
    figsize: [8, 6]
    dpi: 300

paths:
  models: results/models
  endpoints: results/endpoints
  neb: results/neb
  plots: results/plots