API reference
Everything exported from the top-level bouquet namespace, grouped by role.
Docstrings in the source are authoritative; this page is a map.
Class API (recommended)
| Class / Method |
Description |
Bouquet |
Stateful driver: setup_solver → prepare/reconstruct → generate → filter → export (or just .run()) |
Bouquet.from_geqdsk() / Bouquet.from_imas() |
Minimal constructors for the two baseline sources; auto-apply the validated workflow preset for each path |
Bouquet.describe() |
Print the configuration, non-default knobs only |
Bouquet.verify_sigma0_consistency() |
σ=0 regression guard on the draw-pipeline bootstrap split (one solve) |
Bouquet.run_slices() |
Multi-slice IMAS sweep into one archive, one scan_key per slice |
Bouquet.export_bundle() / Bouquet.export_ids() |
Per-draw file bundle (geqdsk / pfile / profiles JSON), or one IMAS/OMAS IDS per draw (fidelity="exact" uses the captured eq_fsa geometry) |
Bouquet.selected_indices() / Bouquet.output_spread() |
Post-generation introspection |
Bouquet.plot_baseline() / .plot_bouquet() / .plot_traces() / .plot_coil_currents() / .plot_spec_summary() |
Bound plotting |
Bouquet.archive |
The run’s BouquetArchive |
Configuration and sources
| Class / Function |
Description |
BouquetConfig |
Top-level typed config; JSON-serializable via to_dict / from_dict |
SolverConfig, UncertaintyConfig, GenerationConfig, FilterConfig, FixedComponentsConfig |
The sub-objects — see workflows.md |
ReconstructionSource / ImasSource |
Baseline source definitions (g-file + profiles, or IMAS/OMAS JSON) |
Baseline / resolve_baseline() |
The common separated-current product every source resolves to |
resolve_uncertainty() |
Resolve sigma envelopes + GPR length scales from the config |
Archive
| Class / Function |
Description |
BouquetArchive |
High-level reader: ar[scan_key] → ScanView → DrawView (profiles, flags, parsed equilibria, extraction) |
ScanView / DrawView |
The view objects those return |
write_provenance() / load_config() |
Stamp / recover the exact BouquetConfig stored in an archive |
initialize_equilibrium_database() |
Create/open an archive (stamps schema_version) |
load_equilibrium() / load_equilibrium_by_path() |
Read one draw |
load_baseline_profiles() |
Read the per-scan baseline |
load_eq_fsa() |
Read the per-draw live-equilibrium flux-surface-average block |
discover_scan_keys() / count_equilibria() / list_equilibrium_indices() |
Archive introspection |
select_indices() / read_filter_flags() |
Filter-flag queries |
filter_coil_currents() / filter_boundaries() / export_filtered() |
Post-generation selection of the machine-realizable subset |
Layout spec: archive-schema.md. Code source of truth:
bouquet/schema.py.
Parallel generation
| Function |
Description |
parallel_generate() |
Process-parallel driver (backend="laptop" pool or "slurm" job array) |
run_shard() |
The per-worker serial pipeline on one shard |
merge_archives() |
Concatenate shards, verifying every shard hit the same baseline |
emit_slurm_script() |
Write the SLURM job-array + submit scripts |
I/O
| Class / Function |
Description |
GEQDSKEquilibrium / read_geqdsk() |
Full-featured COCOS-aware GEQDSK reader with flux-surface analysis |
read_eqdsk_from_bytes() |
Parse a g-file from in-memory bytes |
bouquet.io.write_geqdsk() |
Write a raw g-file dict to disk (import from bouquet.io) |
PFile / read_pfile() |
Osborne p-file reader/writer with rotation computation |
read_ida() / IDAProfiles |
IDA netCDF kinetic fits (direct + ensemble layouts) |
read_ida_cer() / IDACERProfiles |
Impurity CER channels |
read_imas_baseline() / read_imas_geometry() |
IMAS/OMAS data-dictionary JSON reader |
write_imas_draw() / export_imas_drawset() |
Reconstruct one / all perturbed IMAS/OMAS IDS from an archive |
Details: io-and-plotting.md.
Physics helpers
| Function |
Description |
parallel_to_toroidal() / toroidal_to_parallel() |
Current-convention conversion, both directions, FSA-geometry aware |
isotropize_fast_pressure() |
Anisotropic fast-pressure reduction for the isotropic GS solve |
fast_pressure_residual() / infer_fast_pressure() |
Fast-ion pressure accounting |
radial_field_from_impurity_force_balance() |
Impurity radial-force-balance E_r with propagated uncertainty |
Hmode_profiles() |
Synthetic H-mode profile generator |
Sampling and uncertainty
| Function / Class |
Description |
GPRProfilePerturber |
Gaussian process profile perturbation engine (Gibbs non-stationary kernels) |
generate_perturbed_GPR() |
One-call wrapper for perturbing a 1-D profile |
sigmoid_length_scale() |
Spatially varying correlation length for Gibbs kernels |
verify_gpr_statistics() |
Monte Carlo validation of the GPR sampling statistics |
calc_cylindrical_li_proxy() |
Fast cylindrical l_i proxy (no GS solve required) |
new_uncertainty_profiles() |
Build 1-D uncertainty envelopes (power-law or flat+tail) |
synthetic_ida_sigma() |
IDA-shaped fractional sigma envelopes for synthetic studies |
Plotting
plot_bouquet, plot_bouquet_timeseries, plot_traces,
plot_boundary_point_traces, plot_geqdsk_bouquet, plot_pfile_bouquet,
plot_kinetic_profiles, plot_jphi_profiles, plot_jphi,
plot_aux_profiles, plot_transport_profiles, plot_coil_currents,
plot_spec_summary, plot_tokamaker_comparison, plot_input_vs_recon, the
draw_* axes-level variants, set_plot_style, and WONG. Signatures and
selection semantics: io-and-plotting.md.
Environment
| Function |
Description |
add_oft_to_path() |
Make OpenFUSIONToolkit importable (OFT_PYTHONPATH → known locations → walk-up) |
find_mesh() |
Locate the TokaMaker mesh (BOUQUET_MESH → walk-up → bundled example) |
find_ida() |
Locate an IDA .cdf (BOUQUET_IDA-as-file → extra → walk-up → BOUQUET_IDA-as-directory, searched recursively; differing same-named vintages raise rather than guess). Data lives outside the repo — nothing is bundled |
All three raise with the full list of locations tried, so failures are actionable on
a new machine.
Functional API (pre-class, still supported)
Existing scripts written against the functional API keep working; new work
should use Bouquet.
| Function |
Description |
generate_bouquet() |
Batch driver: draw N perturbations, solve GS, archive to HDF5. Supports psi_N_kinetic for SOL-aware profiles |
perturb_kinetic_equilibrium() |
Single perturbation: draw profiles, match pressure and l_i, corrective j_phi iteration |
reconstruct_equilibrium() |
Reconstruct one GS equilibrium from a g-file + profiles, with corrective iteration |
classify_jphi_profile() |
Classify the edge current profile (H_mode / Lmode_like_jphi / L_mode) |
fit_inductive_profile() |
Smoothing-spline + PCHIP fit of the inductive current, scaled to a target l_i |
A few knobs are reachable only here — notably p_thresh (pressure acceptance
band, default 0.05) and coil_drift_floor_A (absolute coil-bound floor,
default 50 A).
All tolerance arguments across both APIs are fractions
(l_i_tolerance=0.01 is 1%), never percentages.