Packing

Evaluate candidate semicircle packings against the fixed geometry scorer. Supports single evaluation, batch scoring, and built-in optimization methods.

Current public support: shared placement on CPU workers only. Submit via POST /api/v1/recipes/packing/jobs or POST /api/v1/jobs with "recipe": "packing".

Operations

Operation	Description
`eval`	Score one 45-dimensional candidate vector
`eval_batch`	Score a batch of candidates in one step
`feasible_optimize`	Local search with feasibility-first scoring
`basin_hopping`	Basin-hopping with bounded Gaussian hops and local refinement
`differential_evolution`	Bounded differential evolution

Payload Fields

Field	Type	Required	Description
`operation`	string	yes	One of the operations above
`candidate`	array (45 numbers)	for eval	Single candidate vector
`candidate_ref`	string	alt for eval	Artifact ID of a candidate vector
`candidates`	array of arrays	for eval_batch	Batch of candidate vectors
`candidates_ref`	string	alt for eval_batch	Artifact ID of a batch
`penalty`	number	no	Penalty coefficient
`seed`	integer	no	Random seed for optimization
`max_iters`	integer	no	Max iterations for optimization
`n_restarts`	integer	no	Number of restarts (feasible_optimize)
`population_size`	integer	no	Population size (differential_evolution)
`hop_scale`	number	no	Hop scale (basin_hopping)

Examples

Score a single candidate


POST /api/v1/recipes/packing/jobs
{
  "payload": {
    "operation": "eval",
    "candidate": [0.12, 0.31, 0.88, 0.45, 0.67, 0.23, 0.91, 0.14, 0.56, 0.78, 0.33, 0.62, 0.85, 0.19, 0.44, 0.71, 0.28, 0.93, 0.16, 0.59, 0.82, 0.37, 0.65, 0.11, 0.48, 0.74, 0.26, 0.89, 0.52, 0.18, 0.69, 0.41, 0.95, 0.32, 0.76, 0.22, 0.58, 0.84, 0.13, 0.47, 0.73, 0.39, 0.66, 0.21, 0.87]
  },
  "timeout_s": 300
}

Batch evaluation


POST /api/v1/recipes/packing/jobs
{
  "payload": {
    "operation": "eval_batch",
    "candidates_ref": "art_batch_123"
  },
  "timeout_s": 600
}

Optimize from a starting point


POST /api/v1/recipes/packing/jobs
{
  "payload": {
    "operation": "feasible_optimize",
    "candidate_ref": "art_seed_123",
    "seed": 42,
    "max_iters": 5000,
    "n_restarts": 3
  },
  "timeout_s": 900
}

Result Fields

Eval: success, score, radius, overlap, elapsed_s

Batch eval: success, scores, radii, overlaps, n_candidates, elapsed_s

Optimization: success, best_score, best_radius, best_overlap, best_candidate, evaluations, iterations, elapsed_s