Reference Family Vs Feedforward Support

Support utilities for TE Curve Verification Pipeline reference-family vs feedforward comparison.

class scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.ReferenceModelEntry(target_name, target_kind, harmonic_order, python_model_path, feature_name_list)

Bases: object

One archived reference target model from the curated family inventory.

Parameters:

• target_name (str)

• target_kind (str)

• harmonic_order (int)

• python_model_path (Path)

• feature_name_list (list[str])

target_name: str

target_kind: str

harmonic_order: int

python_model_path: Path

feature_name_list: list[str]

class scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.Track2Candidate(candidate_id, candidate_family, candidate_kind, candidate_source_label, candidate_surface, allowed_direction_list, source_path, selected_harmonic_list, model_entry_list, model_dictionary, registry_entry, training_config, model_object)

Bases: object

One model candidate in the direction-aware TE Curve Verification Pipeline comparison matrix.

Parameters:

• candidate_id (str)

• candidate_family (str)

• candidate_kind (str)

• candidate_source_label (str)

• candidate_surface (str)

• allowed_direction_list (list[str])

• source_path (Path)

• selected_harmonic_list (list[int])

• model_entry_list (list[ReferenceModelEntry] | None)

• model_dictionary (dict[str, Any] | None)

• registry_entry (dict[str, Any] | None)

• training_config (dict[str, Any] | None)

• model_object (Any | None)

candidate_id: str

candidate_family: str

candidate_kind: str

candidate_source_label: str

candidate_surface: str

allowed_direction_list: list[str]

source_path: Path

selected_harmonic_list: list[int]

model_entry_list: list[ReferenceModelEntry] | None

model_dictionary: dict[str, Any] | None

registry_entry: dict[str, Any] | None

training_config: dict[str, Any] | None

model_object: Any | None

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_reference_family_comparison_config(config_path)

Load one TE Curve Verification Pipeline reference-family comparison configuration file.

Parameters:

config_path (str | Path)

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_yaml_dictionary(yaml_path)

Load one YAML dictionary from disk.

Parameters:

yaml_path (Path)

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_comparison_report_path(training_config)

Build the canonical Markdown report path for one comparison run.

Parameters:

training_config (dict[str, Any])

Return type:

Path

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_canonical_track2_report_path(training_config)

Resolve the stable TE curve-verification report path for canonical full-matrix runs.

Parameters:

training_config (dict[str, Any])

Return type:

Path

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_reference_inventory(reference_inventory_path)

Load one curated RCIM Model-Bank Reproduction family reference inventory.

Parameters:

reference_inventory_path (str | Path)

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_selected_harmonic_list(reference_inventory)

Resolve the harmonic orders covered by the curated reference inventory.

Parameters:

reference_inventory (dict[str, Any])

Return type:

list[int]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_reference_model_entries(reference_inventory)

Load and validate the reference target-model inventory entries.

Parameters:

reference_inventory (dict[str, Any])

Return type:

list[ReferenceModelEntry]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.find_reference_model_entry(reference_model_entry_list, target_kind, harmonic_order)

Find one target entry in a reference inventory.

Parameters:

• reference_model_entry_list (list[ReferenceModelEntry])

• target_kind (str)

• harmonic_order (int)

Return type:

ReferenceModelEntry

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_reference_model_dictionary(reference_model_entry_list)

Load the archived Python estimators for one curated reference bank.

Parameters:

reference_model_entry_list (list[ReferenceModelEntry])

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_feedforward_best_entry(feedforward_leaderboard_path)

Resolve the current canonical best feedforward registry entry.

Parameters:

feedforward_leaderboard_path (str | Path)

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_family_best_entry(registry_path)

Resolve one family-best registry entry from a registry YAML file.

Parameters:

registry_path (str | Path)

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_lightning_regression_module_for_inference(checkpoint_path, training_config)

Load one Lightning TE checkpoint without constructing the full datamodule.

Parameters:

• checkpoint_path (Path)

• training_config (dict[str, Any])

Return type:

TransmissionErrorRegressionModule

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_feedforward_regression_module(feedforward_best_entry)

Load the canonical best feedforward checkpoint plus its config snapshot.

Parameters:

feedforward_best_entry (dict[str, Any])

Return type:

tuple[TransmissionErrorRegressionModule, dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_wave1_registry_model(registry_entry)

Load one Wave 1 registry-backed model artifact.

Parameters:

registry_entry (dict[str, Any])

Return type:

tuple[Any, dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_wave1_exported_model(export_inventory)

Load one Wave 1 exported model directly from the models/ tree.

Parameters:

export_inventory (dict[str, Any])

Return type:

tuple[Any, dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_curve_record_list(training_config, selected_harmonic_list)

Build the held-out TE-curve record list used by the comparison.

Parameters:

• training_config (dict[str, Any])

• selected_harmonic_list (list[int])

Return type:

tuple[list[HarmonicCurveRecord], dict[str, int], dict[str, int], Path]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_feature_matrix(curve_record_list)

Build the reference-bank feature matrix aligned with the archived models.

Parameters:

curve_record_list (list[HarmonicCurveRecord])

Return type:

DataFrame

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.predict_reference_model_in_batches(model_object, reference_feature_matrix)

Predict one archived reference target with a bounded inference batch size.

Parameters:

• model_object (Any)

• reference_feature_matrix (DataFrame)

Return type:

ndarray

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.predict_reference_bank_target_dictionary(curve_record_list, reference_model_entry_list, reference_model_dictionary)

Predict all archived amplitude and phase targets for the held-out curves.

Parameters:

• curve_record_list (list[HarmonicCurveRecord])

• reference_model_entry_list (list[ReferenceModelEntry])

• reference_model_dictionary (dict[str, Any] | None)

Return type:

dict[str, ndarray]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_coefficient_dictionary(predicted_target_dictionary, sample_index, selected_harmonic_list)

Convert one amplitude/phase prediction set into harmonic coefficients.

Parameters:

• predicted_target_dictionary (dict[str, ndarray])

• sample_index (int)

• selected_harmonic_list (list[int])

Return type:

tuple[dict[str, float], dict[str, float]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_prediction_lookup(reference_model_entry_list, predicted_target_dictionary)

Build a target lookup keyed by target kind and harmonic order.

Parameters:

• reference_model_entry_list (list[ReferenceModelEntry])

• predicted_target_dictionary (dict[str, ndarray])

Return type:

dict[tuple[str, int], ndarray]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_coefficient_dictionary_from_entries(prediction_lookup, sample_index, selected_harmonic_list, h0_sign_multiplier=1.0)

Convert one generic RCIM Model-Bank Reproduction bank prediction into harmonic coefficients.

Parameters:

• prediction_lookup (dict[tuple[str, int], ndarray])

• sample_index (int)

• selected_harmonic_list (list[int])

• h0_sign_multiplier (float)

Return type:

tuple[dict[str, float], dict[str, float]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_reference_h0_sign_multiplier(candidate)

Resolve source-specific h0 sign compatibility for reference banks.

Parameters:

candidate (Track2Candidate)

Return type:

float

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_target_metric_dictionary(curve_record_list, predicted_target_dictionary, selected_harmonic_list)

Build compact amplitude and phase diagnostics for the archived bank.

Parameters:

• curve_record_list (list[HarmonicCurveRecord])

• predicted_target_dictionary (dict[str, ndarray])

• selected_harmonic_list (list[int])

Return type:

dict[str, float]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_target_metric_dictionary_from_entries(curve_record_list, reference_model_entry_list, predicted_target_dictionary, selected_harmonic_list)

Build target diagnostics for a generic forward or backward reference bank.

Parameters:

• curve_record_list (list[HarmonicCurveRecord])

• reference_model_entry_list (list[ReferenceModelEntry])

• predicted_target_dictionary (dict[str, ndarray])

• selected_harmonic_list (list[int])

Return type:

dict[str, float]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_model_dataset_name(training_config)

Resolve the dataset schema expected by one model artifact.

Parameters:

training_config (dict[str, Any] | None)

Return type:

str

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_feedforward_input_tensor(curve_record, training_config=None)

Build the pointwise feedforward input tensor for one curve record.

Parameters:

• curve_record (HarmonicCurveRecord)

• training_config (dict[str, Any] | None)

Return type:

Tensor

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_temporal_sequence_input_tensor(curve_record, training_config)

Build full-curve sequence windows for one temporal registry model.

Parameters:

• curve_record (HarmonicCurveRecord)

• training_config (dict[str, Any])

Return type:

Tensor

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.predict_temporal_sequence_curve_in_batches(model_object, curve_record, training_config)

Predict one temporal TE curve without materializing every sequence window at once.

Parameters:

• model_object (TransmissionErrorRegressionModule)

• curve_record (HarmonicCurveRecord)

• training_config (dict[str, Any])

Return type:

ndarray

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.predict_feedforward_curve(regression_module, curve_record)

Predict one TE curve with the canonical feedforward checkpoint.

Parameters:

• regression_module (TransmissionErrorRegressionModule)

• curve_record (HarmonicCurveRecord)

Return type:

ndarray

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.predict_wave1_registry_curve(model_object, training_config, curve_record)

Predict one TE curve with a loaded registry-backed model.

Parameters:

• model_object (Any)

• training_config (dict[str, Any])

• curve_record (HarmonicCurveRecord)

Return type:

ndarray

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.normalize_allowed_direction_list(candidate_configuration)

Resolve allowed evaluation directions for one candidate configuration.

Parameters:

candidate_configuration (dict[str, Any])

Return type:

list[str]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_legacy_candidate_configuration_list(training_config)

Build candidate configurations for the historical single-bank config.

Parameters:

training_config (dict[str, Any])

Return type:

list[dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_family_folder_lookup(family_configuration_list)

Build a paper-family to archive-folder lookup from compact config rows.

Parameters:

family_configuration_list (list[dict[str, Any]])

Return type:

dict[str, str]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_composite_reference_candidate_configuration_list(generation_configuration)

Build configured composed reference-bank candidates.

Parameters:

generation_configuration (dict[str, Any])

Return type:

list[dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_registry_candidate_configuration_list(registry_group_configuration, default_source_label)

Generate registry-backed TE Curve Verification Pipeline candidates from compact family metadata.

Parameters:

• registry_group_configuration (dict[str, Any])

• default_source_label (str)

Return type:

list[dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_generated_candidate_configuration_list(training_config)

Generate a full TE Curve Verification Pipeline candidate matrix from compact config metadata.

Parameters:

training_config (dict[str, Any])

Return type:

list[dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_track2_candidate_configuration_list(training_config)

Resolve the configured TE Curve Verification Pipeline candidate list.

Parameters:

training_config (dict[str, Any])

Return type:

list[dict[str, Any]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.load_track2_candidate(candidate_configuration)

Load one configured TE Curve Verification Pipeline candidate.

Parameters:

candidate_configuration (dict[str, Any])

Return type:

Track2Candidate

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.filter_curve_records_for_candidate(curve_record_list, candidate)

Filter held-out curves to the directions valid for one candidate.

Parameters:

• curve_record_list (list[HarmonicCurveRecord])

• candidate (Track2Candidate)

Return type:

list[HarmonicCurveRecord]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.evaluate_track2_candidate(candidate, curve_record_list, percentage_error_denominator, include_curve_payload=True)

Evaluate one TE Curve Verification Pipeline candidate on its valid held-out curve records.

Parameters:

• candidate (Track2Candidate)

• curve_record_list (list[HarmonicCurveRecord])

• percentage_error_denominator (str)

• include_curve_payload (bool)

Return type:

tuple[list[dict[str, Any]], dict[str, float] | None]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.summarize_metric_dictionary(metric_dictionary_list)

Average one metric-dictionary list and add a p95 percentage statistic.

Parameters:

metric_dictionary_list (list[dict[str, float]])

Return type:

dict[str, float]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_group_metric_summary(per_sample_entry_list, group_key_name)

Aggregate per-model metrics by one chosen grouping key.

Parameters:

• per_sample_entry_list (list[dict[str, Any]])

• group_key_name (str)

Return type:

dict[str, dict[str, dict[str, float]]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_candidate_metric_summary(per_candidate_entry_list)

Summarize metrics for every evaluated TE Curve Verification Pipeline candidate.

Parameters:

per_candidate_entry_list (list[dict[str, Any]])

Return type:

dict[str, dict[str, float]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_generic_group_metric_summary(per_candidate_entry_list, group_key_name)

Summarize candidate metrics by one grouping key.

Parameters:

• per_candidate_entry_list (list[dict[str, Any]])

• group_key_name (str)

Return type:

dict[str, dict[str, dict[str, float]]]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.save_track2_per_condition_metrics_csv(output_directory, per_candidate_entry_list)

Save the direction-aware TE Curve Verification Pipeline per-condition metric table.

Parameters:

• output_directory (Path)

• per_candidate_entry_list (list[dict[str, Any]])

Return type:

Path

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.save_per_condition_metrics_csv(output_directory, per_sample_entry_list)

Save one per-condition comparison table for downstream inspection.

Parameters:

• output_directory (Path)

• per_sample_entry_list (list[dict[str, Any]])

Return type:

Path

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.maybe_generate_preview_plots(output_directory, per_sample_entry_list, preview_curve_count)

Generate a few representative overlay plots when matplotlib is available.

Parameters:

• output_directory (Path)

• per_sample_entry_list (list[dict[str, Any]])

• preview_curve_count (int)

Return type:

list[str]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.maybe_generate_track2_preview_plots(output_directory, per_candidate_entry_list, preview_curve_count)

Generate direction-aware TE Curve Verification Pipeline overlay plots.

Parameters:

• output_directory (Path)

• per_candidate_entry_list (list[dict[str, Any]])

• preview_curve_count (int)

Return type:

list[str]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.resolve_track2_report_plot_root(training_config)

Resolve the optional report-facing grouped PNG root.

Parameters:

training_config (dict[str, Any])

Return type:

Path | None

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_track2_plot_source_folder_name(candidate_source_label)

Map a candidate source label to the requested report-folder name.

Parameters:

candidate_source_label (str)

Return type:

str

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_track2_condition_slug(per_candidate_entry)

Build a stable condition slug for one plotted TE Curve Verification Pipeline curve.

Parameters:

per_candidate_entry (dict[str, Any])

Return type:

str

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.maybe_generate_track2_grouped_report_plots(report_plot_root, per_candidate_entry_list, preview_curve_count_per_candidate)

Generate report-facing TE Curve Verification Pipeline PNG overlays for every evaluated model.

Parameters:

• report_plot_root (Path | None)

• per_candidate_entry_list (list[dict[str, Any]])

• preview_curve_count_per_candidate (int)

Return type:

list[str]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_comparison_summary(resolved_config_path, output_directory, training_config, reference_inventory, feedforward_best_entry, curve_record_list, reference_target_metric_dictionary, aggregate_metric_dictionary, per_sample_entry_list, direction_metric_summary, temperature_metric_summary, preview_plot_path_list, per_condition_metrics_csv_path, selected_harmonic_list)

Build the machine-readable comparison summary.

Parameters:

• resolved_config_path (Path)

• output_directory (Path)

• training_config (dict[str, Any])

• reference_inventory (dict[str, Any])

• feedforward_best_entry (dict[str, Any])

• curve_record_list (list[HarmonicCurveRecord])

• reference_target_metric_dictionary (dict[str, float])

• aggregate_metric_dictionary (dict[str, dict[str, float]])

• per_sample_entry_list (list[dict[str, Any]])

• direction_metric_summary (dict[str, dict[str, dict[str, float]]])

• temperature_metric_summary (dict[str, dict[str, dict[str, float]]])

• preview_plot_path_list (list[str])

• per_condition_metrics_csv_path (Path)

• selected_harmonic_list (list[int])

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_reference_family_vs_feedforward_report_markdown(comparison_summary)

Build the Markdown comparison report.

Parameters:

comparison_summary (dict[str, Any])

Return type:

str

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_track2_directional_comparison_summary(resolved_config_path, output_directory, training_config, curve_record_list, candidate_list, target_metric_dictionary, per_candidate_entry_list, preview_plot_path_list, report_plot_root, report_plot_path_list, per_condition_metrics_csv_path, dataset_root, candidate_metric_summary_override=None, direction_metric_summary_override=None, temperature_metric_summary_override=None, sample_preview_list_override=None)

Build the direction-aware TE Curve Verification Pipeline comparison summary.

Parameters:

• resolved_config_path (Path)

• output_directory (Path)

• training_config (dict[str, Any])

• curve_record_list (list[HarmonicCurveRecord])

• candidate_list (list[Track2Candidate])

• target_metric_dictionary (dict[str, dict[str, float]])

• per_candidate_entry_list (list[dict[str, Any]])

• preview_plot_path_list (list[str])

• report_plot_root (Path | None)

• report_plot_path_list (list[str])

• per_condition_metrics_csv_path (Path)

• dataset_root (Path)

• candidate_metric_summary_override (dict[str, dict[str, float]] | None)

• direction_metric_summary_override (dict[str, dict[str, dict[str, float]]] | None)

• temperature_metric_summary_override (dict[str, dict[str, dict[str, float]]] | None)

• sample_preview_list_override (list[dict[str, object]] | None)

Return type:

dict[str, Any]

scripts.paper_reimplementation.rcim_ml_compensation.reference_family_vs_feedforward.reference_family_vs_feedforward_support.build_track2_directional_comparison_report_markdown(comparison_summary)

Build the direction-aware TE Curve Verification Pipeline Markdown report.

Parameters:

comparison_summary (dict[str, Any])

Return type:

str