Skip to content

task_charts__bar_3d__series_total_gap_value

Contract

  1. Domain: charts
  2. Scene id: bar_3d
  3. Source implementation scene: charts/bar_3d
  4. Query id: single
  5. Semantic query details are recorded in query_id and trace params.

Implementation

  1. Registered class: trace_tasks.tasks.charts.bar_3d.series_total_gap_value.ChartsThreeDBarSeriesTotalGapValueTask
  2. Prompt lookup scene: charts/bar_3d
  3. Generation is deterministic from instance_seed, explicit params, prompt bundle, renderer config, and code versions.
  4. Answers and annotation are produced from the same metadata execution trace.

Annotation Contract

  1. Answer schema: integer_value.
  2. Annotation schema: point_set_map.
  3. Annotation is keyed by the two compared series labels; each key maps to the top-center points for that series across displayed categories.
  4. Renderer context such as legends, axes, decorative labels, titles, and distractor text is metadata unless the task explicitly asks for it as annotation.

Program Contract

Program: difference(sum(value(category,series_a) for category in visible_categories), sum(value(category,series_b) for category in visible_categories)); output=integer_value; annotation=point_set_map(series_a_total_bars, series_b_total_bars); scene=bar_3d; scope=series_total_gap_value

Candidate set: the visible 3D bars grouped by category and series inside the series_total_gap_value objective scope. Operands: prompt-bound labels, categories, series names, thresholds, intervals, references, and encoded chart values, plus the task's prompt-bound target operands when present. Operation: evaluate difference over the candidate set using the filters, comparisons, aggregations, rankings, projections, or counterfactual edits named in the program expression; generation enforces a unique final answer. Output binding: answer is the integer_value value bound by integer_value. Annotation witnesses: point_set_map witnesses bound by point_set_map(series_a_total_bars, series_b_total_bars). Annotation is keyed by the two compared series labels; each key maps to the top-center points for that series across displayed categories. Renderer context such as legends, axes, decorative labels, titles, and distractor text is metadata unless the task explicitly asks for it as annotation. Query ids: single.

Reasoning Operations

Families: aggregation, formula_evaluation

Query Details

Query id Program signature Answer schema Annotation schema
single numeric.series_total_difference integer_value point_set_map