Skip to content

Adapt Optimisation¶

adapt_optimisation — mvp.adapt_optimisation

Cluster: ML & Optimisation | Type: component | MCP Tools: 27

Overview¶

adapt_optimisation provides a curated optimisation adapter with audited backend-native extension operations. The curated tier exposes stable scipy/grid/random, metaheuristic, multi-objective, function registry, result, and advisory operations; the native tier exposes selected backend-native capabilities through explicit MCP operations with schemas, allowlists, resource bounds, safety guards, capability discovery, and normalised degradation reporting. It is not a full unrestricted adapter over SciPy, pyMetaheuristic, or pyMultiobjective.

Launch caveat

adapt_optimisation is a numerical optimisation utility, not a domain workflow by itself. It is most useful when a recipe, agent, or user has already translated the business problem into an objective function, bounds, and constraints. For non-technical users, prefer workflow templates or job agents that frame the optimisation problem in domain language, then call this component under the hood.

When to use:

Minimising objective functions with scipy or metaheuristic solvers
Hyperparameter tuning via grid, random, or population-based search
Benchmarking optimisation methods against standard test functions
Solving a domain problem after a workflow has expressed the target as a measurable objective

Example:

from mvp.adapt_optimisation import AdaptOptimisationBlock, OptimisationInput

block = AdaptOptimisationBlock(name="opt")
result = block.infer(OptimisationInput(fn_name="sphere", bounds=[[-5, 5], [-5, 5]]))
# result.ok → True; result.value → OptimisationOutput with x_opt and f_opt

Works well with: adapt_pygad, adapt_sklearn, formal_methods

Public API¶

`AdaptOptimisationBlock(AIBlock[OptimisationInput, OptimisationOutput, None])`¶

Wraps scipy.optimize.minimize (if available) with a grid-search fallback,

Field	Type	Default
`name`	`str`	'adapt_optimisation'
`resource_bounds`	`ResourceBounds \\| None`	None
`usage`	`ResourceUsage`	field(default_factory=ResourceUsage)
`agentic_planner`	`OptimisationPlanner \\| None`	None

Methods:

`infer(data: OptimisationInput) -> Result[OptimisationOutput]`¶

`bias() -> dict`¶

`OptimisationInput(BaseModel)`¶

Input to AdaptOptimisationBlock.

Field	Type	Default
`op`	`Literal['solve', 'optimise', 'optimize', 'info', 'recommend']`	'solve'
`fn_name`	`str`	'sphere'
`bounds`	`list[list[float]]`	Field(default_factory=lambda: [[-1.0, 1.0]])
`method`	`Literal['scipy', 'grid']`	'scipy'
`max_evals`	`int`	200
`minimize`	`bool`	True
`n_objectives`	`int`	1
`multimodal_hint`	`bool`	False
`expression`	`str`	''
`pts_per_dim`	`int`	0
`run_mode`	`str`	'beta'
`reviewer_signature`	`str`	''

`OptimisationOutput(BaseModel)`¶

Output from AdaptOptimisationBlock.

Field	Type	Default
`x_opt`	`list[float]`	Field(default_factory=list)
`f_opt`	`float`	0.0
`n_evals`	`int`	0
`converged`	`bool`	False
`method`	`str`	''
`fn_name`	`str`	''
`degraded`	`bool`	False
`degradation_reason`	`str`	''
`solver_rationale`	`str`	''
`convergence_tolerance`	`float`	0.0
`constraint_violations`	`list[str]`	Field(default_factory=list)
`backend_used`	`str`	''
`fallback_reason`	`str`	''
`seed`	`int \\| None`	None
`acceptance`	`Literal['acceptable', 'incomplete', 'degraded', 'infeasible']`	'acceptable'
`ranked_algorithms`	`list[str]`	Field(default_factory=list)
`recommendation_rationale`	`str`	''
`eligible_algorithms`	`list[str]`	Field(default_factory=list)
`agentic_evidence`	`dict`	Field(default_factory=dict)
`requires_fallback`	`bool`	False
`diverged_from_floor`	`bool`	False
`deterministic_floor_rank`	`list[str]`	Field(default_factory=list)
`completion_state`	`str`	''

Methods:

`optimal_x() -> list[float]`¶

Backward-compatible alias used by older job recipes.

`optimal_f() -> float`¶

Backward-compatible alias used by older job recipes.

`AdaptOptimisationMCPBlock(AIBlock[MCPOptimisationInput, MCPOptimisationOutput, dict])`¶

Field	Type	Default
`name`	`str`	'adapt_optimisation_mcp'
`db_path`	`str`	':memory:'
`resource_bounds`	`ResourceBounds \\| None`	None
`usage`	`ResourceUsage`	field(default_factory=ResourceUsage)
`state`	`dict`	field(default_factory=dict)

Methods:

`infer(data: MCPOptimisationInput) -> Result[MCPOptimisationOutput]`¶

`MCPOptimisationInput(BaseModel)`¶

Field	Type	Default
`op`	`Literal['run_scipy', 'run_grid', 'run_random', 'run_pso', 'run_ga', 'run_de', 'run_sa', 'run_gwo', 'run_whale', 'run_firefly', 'run_nsga2', 'run_nsga3', 'run_moead', 'fn_register', 'fn_list', 'fn_evaluate', 'result_list', 'result_compare', 'result_delete', 'native_capability_info', 'native_describe_method', 'native_validate_problem', 'native_run_method', 'native_result_schema', 'get_info', 'recommend_algo', 'list_patterns']`	required
`fn_name`	`str`	'sphere'
`fn_names_json`	`str`	'[]'
`bounds`	`list[list[float]]`	Field(default_factory=lambda: [[-5.0, 5.0], [-5.0, 5.0]])
`minimize`	`bool`	True
`max_evals`	`int`	500
`n_samples`	`int`	200
`seed`	`int`	-1
`x0_json`	`str`	'[]'
`scipy_method`	`str`	'lbfgsb'
`pts_per_dim`	`int`	0
`iterations`	`int`	200
`population_size`	`int`	50
`swarm_size`	`int`	50
`w`	`float`	0.9
`c1`	`float`	2.0
`c2`	`float`	2.0
`mutation_rate`	`float`	0.1
`F`	`float`	0.9
`Cr`	`float`	0.5
`initial_temp`	`float`	1.0
`final_temp`	`float`	0.0001
`alpha`	`float`	0.9
`pack_size`	`int`	10
`n_fireflies`	`int`	25
`ff_alpha`	`float`	0.5
`beta`	`float`	1.0
`gamma`	`float`	1.0
`n_reference_points`	`int`	10
`n_neighbors`	`int`	5
`generations`	`int`	100
`expression`	`str`	''
`x_json`	`str`	'[]'
`notes`	`str`	''
`run_ids_json`	`str`	'[]'
`run_id`	`str`	''
`limit`	`int`	20
`algorithm`	`str`	''
`request_id`	`str`	''
`task_id`	`str`	''
`native_backend`	`str`	''
`native_action`	`str`	''
`native_method`	`str`	''
`native_options`	`dict[str, Any]`	Field(default_factory=dict)
`native_bounds`	`list[list[float]]`	Field(default_factory=list)
`max_iterations`	`int`	1000
`max_evaluations`	`int`	10000
`allow_fallback`	`bool`	False
`timeout_sec`	`float`	0.0
`stall_patience`	`int`	30
`stall_tol`	`float`	1e-08
`max_pareto_size`	`int`	100
`run_mode`	`str`	'beta'
`reviewer_signature`	`str`	''

`MCPOptimisationOutput(BaseModel)`¶

Field	Type	Default
`op`	`str`	required
`ok`	`bool`	True
`message`	`str`	''
`run_id`	`str`	''
`algorithm`	`str`	''
`fn_name`	`str`	''
`x_opt`	`list[float]`	Field(default_factory=list)
`f_opt`	`float`	0.0
`n_evals`	`int`	0
`converged`	`bool`	False
`runtime_ms`	`float`	0.0
`pareto_front`	`list[dict]`	Field(default_factory=list)
`runs`	`list[dict]`	Field(default_factory=list)
`functions`	`list[dict]`	Field(default_factory=list)
`value`	`float`	0.0
`metadata`	`dict`	Field(default_factory=dict)
`count`	`int`	0
`degraded`	`bool`	False
`degradation_reason`	`str`	''
`completion_state`	`Literal['verified', 'qualified-draft', 'blocked-escalated']`	'qualified-draft'
`warning_card`	`dict`	Field(default_factory=dict)
`evidence`	`list[dict]`	Field(default_factory=list)
`request_id`	`str`	''
`task_id`	`str`	''
`backend_requested`	`str`	''
`backend_used`	`str`	''
`fallback_used`	`bool`	False
`fallback_reason`	`str`	''
`acceptance`	`str`	'accepted'

MCP Tools¶

Operation	Source
`run_scipy`	optimisation_mcp
`run_grid`	optimisation_mcp
`run_random`	optimisation_mcp
`run_pso`	optimisation_mcp
`run_ga`	optimisation_mcp
`run_de`	optimisation_mcp
`run_sa`	optimisation_mcp
`run_gwo`	optimisation_mcp
`run_whale`	optimisation_mcp
`run_firefly`	optimisation_mcp
`run_nsga2`	optimisation_mcp
`run_nsga3`	optimisation_mcp
`run_moead`	optimisation_mcp
`fn_register`	optimisation_mcp
`fn_list`	optimisation_mcp
`fn_evaluate`	optimisation_mcp
`result_list`	optimisation_mcp
`result_compare`	optimisation_mcp
`result_delete`	optimisation_mcp
`native_capability_info`	optimisation_mcp
`native_describe_method`	optimisation_mcp
`native_validate_problem`	optimisation_mcp
`native_run_method`	optimisation_mcp
`native_result_schema`	optimisation_mcp
`get_info`	optimisation_mcp
`recommend_algo`	optimisation_mcp
`list_patterns`	optimisation_mcp