rgpycrumbs.chemgp.plt_gp

Plot ChemGP figures from HDF5 data files.

Added in version 1.6.0.

CLI for generating publication figures from ChemGP HDF5 outputs. Reads grids, tables, paths, and point sets, then delegates to chemparseplot for visualization.

HDF5 layout (mirrors Julia common_plot.jl helpers):

• grids/<name>: 2D arrays with attrs x_range, y_range, x_length, y_length

• table/<name>: group of same-length 1D arrays

• paths/<name>: point sequences (x, y or rAB, rBC)

• points/<name>: point sets (x, y or pc1, pc2)

• Root attrs: metadata scalars

Attributes

log
_CLAMP_PRESETS

Functions

h5_read_table(→ pandas.DataFrame)	Read a group of same-length vectors as a DataFrame.
h5_read_grid(→ tuple[numpy.ndarray, ...)	Read a 2D grid with optional axis ranges.
h5_read_path(→ dict[str, numpy.ndarray])	Read a path (ordered point sequence).
h5_read_points(→ dict[str, numpy.ndarray])	Read a point set.
h5_read_metadata(→ dict)	Read root-level metadata attributes.
save_plot(→ None)	Save a plotnine ggplot or matplotlib Figure to PDF.
_detect_clamp(→ tuple[float | None, float | None, ...)	Detect energy clamping preset from filename.
common_options(func)	Shared options for all subcommands.
cli()	ChemGP figure generation from HDF5 data.
convergence(input_path, output_path, width, height, dpi)	Force/energy convergence vs oracle calls.
surface(input_path, output_path, width, height, dpi, ...)	2D PES contour plot.
quality(input_path, output_path, width, height, dpi, ...)	GP surrogate quality progression (multi-panel).
rff(input_path, output_path, width, height, dpi)	RFF approximation quality vs exact GP.
nll(input_path, output_path, width, height, dpi)	MAP-NLL landscape in hyperparameter space.
sensitivity(input_path, output_path, width, height, dpi)	Hyperparameter sensitivity grid (3x3).
trust(input_path, output_path, width, height, dpi)	Trust region illustration (1D slice).
variance(input_path, output_path, width, height, dpi)	GP variance overlaid on PES.
fps(input_path, output_path, width, height, dpi)	FPS subset visualization (PCA scatter).
profile(input_path, output_path, width, height, dpi)	NEB energy profile (image index vs delta E).
landscape(dat_pattern, con_pattern, source_dir, ...)	2D NEB reaction landscape via plt-neb (RMSD coordinates).
batch(config_path, base_dir, dpi)	Generate multiple plots from a TOML config.
main()

Module Contents

rgpycrumbs.chemgp.plt_gp.log

rgpycrumbs.chemgp.plt_gp.h5_read_table(f: h5py.File, name: str = 'table') → pandas.DataFrame

Read a group of same-length vectors as a DataFrame.

rgpycrumbs.chemgp.plt_gp.h5_read_grid(f: h5py.File, name: str) → tuple[numpy.ndarray, numpy.ndarray | None, numpy.ndarray | None]

Read a 2D grid with optional axis ranges.

Returns (data, x_coords, y_coords).

rgpycrumbs.chemgp.plt_gp.h5_read_path(f: h5py.File, name: str) → dict[str, numpy.ndarray]

Read a path (ordered point sequence).

rgpycrumbs.chemgp.plt_gp.h5_read_points(f: h5py.File, name: str) → dict[str, numpy.ndarray]

Read a point set.

rgpycrumbs.chemgp.plt_gp.h5_read_metadata(f: h5py.File) → dict

Read root-level metadata attributes.

rgpycrumbs.chemgp.plt_gp.save_plot(fig, output: pathlib.Path, dpi: int) → None

Save a plotnine ggplot or matplotlib Figure to PDF.

rgpycrumbs.chemgp.plt_gp._CLAMP_PRESETS

rgpycrumbs.chemgp.plt_gp._detect_clamp(filename: str) → tuple[float | None, float | None, float | None]

Detect energy clamping preset from filename.

rgpycrumbs.chemgp.plt_gp.common_options(func)

Shared options for all subcommands.

rgpycrumbs.chemgp.plt_gp.cli()

ChemGP figure generation from HDF5 data.

rgpycrumbs.chemgp.plt_gp.convergence(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

Force/energy convergence vs oracle calls.

rgpycrumbs.chemgp.plt_gp.surface(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int, clamp_lo: float | None, clamp_hi: float | None, contour_step: float | None)

2D PES contour plot.

rgpycrumbs.chemgp.plt_gp.quality(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int, n_points: tuple[int, Ellipsis] | None)

GP surrogate quality progression (multi-panel).

rgpycrumbs.chemgp.plt_gp.rff(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

RFF approximation quality vs exact GP.

rgpycrumbs.chemgp.plt_gp.nll(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

MAP-NLL landscape in hyperparameter space.

rgpycrumbs.chemgp.plt_gp.sensitivity(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

Hyperparameter sensitivity grid (3x3).

rgpycrumbs.chemgp.plt_gp.trust(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

Trust region illustration (1D slice).

rgpycrumbs.chemgp.plt_gp.variance(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

GP variance overlaid on PES.

rgpycrumbs.chemgp.plt_gp.fps(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

FPS subset visualization (PCA scatter).

rgpycrumbs.chemgp.plt_gp.profile(input_path: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int)

NEB energy profile (image index vs delta E).

rgpycrumbs.chemgp.plt_gp.landscape(dat_pattern: str | None, con_pattern: str | None, source_dir: pathlib.Path, output_path: pathlib.Path, width: float, height: float, dpi: int, surface_type: str, landscape_mode: str, plot_structures: str, project_path: bool, additional_con: tuple | None, augment_dat: str | None, augment_con: str | None, zoom_ratio: float | None, ase_rotation: str | None)

2D NEB reaction landscape via plt-neb (RMSD coordinates).

rgpycrumbs.chemgp.plt_gp.batch(config_path: pathlib.Path, base_dir: pathlib.Path | None, dpi: int)

Generate multiple plots from a TOML config.

rgpycrumbs.chemgp.plt_gp.main()