Params Analysis
Analyze optimization-parameter updates in force-field training workflows.
This module extracts and normalizes optimization-parameter change records, with section-aware mapping to force-field parameter groups. It is bounded to parameter-update analysis and does not execute optimization itself.
Usage context
- Parameter change review: Inspect optimization updates by section and key.
- Cross-check workflows: Compare optimization parameters against base force field.
- Training diagnostics: Export normalized parameter-update tables for reports.
Request: FFieldOptimizationParameterRequest
Request payload for optimization-parameter extraction and interpretation.
This request controls duplicate handling and whether params pointer tuples are interpreted into concrete force-field section/component/value metadata.
Fields
| Field | Type | Default | Help | Choices |
|---|---|---|---|---|
drop_duplicate |
bool |
True | Remove duplicated params rows by (ff_section, ff_section_line, ff_parameter). | True, False |
interpret |
bool |
False | If true, resolve params pointers to named force-field parameters loaded from task required_data. | True, False |
Examples
request = ForceFieldOptimizationParameterRequest(
drop_duplicate=True,
interpret=True,
)
The request returns deduplicated rows with interpreted component mappings.
Task: FFieldOptimizationParameterTask
Return raw or interpreted optimization-parameter definitions from params.
Method: recommended_presentations(_result: FFieldOptimizationParameterResult, payload: dict[str, Any])
Recommend the default tabular presentation for params outputs.
Parameter definitions are naturally table-oriented, so this analyzer exposes a single stable table recommendation.
Works on
Analyzer task output for force_field_optimization_parameters.
Parameters
| Name | Type | Description |
|---|---|---|
_result |
FFieldOptimizationParameterResult |
Typed analyzer result instance (unused by current logic). |
payload |
dict[str, Any] |
Serialized payload (unused for fixed recommendations). |
Returns
| Type | Description |
|---|---|
list[PresentationSpec] |
Single table presentation specification. |
Examples
specs = ForceFieldOptimizationParameterTask.recommended_presentations(_result, {})
The returned list contains one table view specification.
Method: run(data: ForceFieldOptimizationParameterBundleData, request: FFieldOptimizationParameterRequest, reporter=None)
Run extraction of optimization parameters in raw or interpreted mode.
Loads optimization parameter rows from the bundle, optionally resolves
them to force-field component/value context, and enforces a stable
component column in the returned table.
Works on
ForceFieldOptimizationParameterBundleData.
Parameters
| Name | Type | Description |
|---|---|---|
data |
ForceFieldOptimizationParameterBundleData |
Bundle containing optimization parameter rows and optional force-field parameter tables for interpretation. |
request |
FFieldOptimizationParameterRequest |
Request controlling deduplication and interpretation behavior. |
reporter |
Any, optional |
Progress callback accepted by analyzer tasks; unused here. |
Returns
| Type | Description |
|---|---|
FFieldOptimizationParameterResult |
Result containing normalized parameter-update rows. |
Examples
result = ForceFieldOptimizationParameterTask().run(
data,
ForceFieldOptimizationParameterRequest(interpret=False),
)
result.table contains raw params rows with a stable component column.
Result: FFieldOptimizationParameterResult
Result payload for optimization-parameter extraction.
The analyzer returns raw or interpreted parameter-update rows with a stable
component label column for downstream presentation and plotting.
Fields
| Field | Type | Default | Help | Choices |
|---|---|---|---|---|
table |
pd.DataFrame |
|||
request |
FFieldOptimizationParameterRequest |
Notes
The component column is always present: interpreted outputs use resolved
parameter names, while raw outputs use synthetic labels such as p1.
Examples
row = {
"component": "De_sigma",
"ff_section": 3,
"ff_section_line": 12,
"ff_parameter": 1,
"search_interval": 0.2,
}
The sample row represents one optimization parameter target.