discrete_optimization.singlemachine.transformations package

Submodules

discrete_optimization.singlemachine.transformations.to_jsp module

Transformation from Single Machine Scheduling to Job Shop Problem.

This is a LOSSY transformation because: - Due dates are lost (JSP doesn’t model deadlines) - Weights are lost (JSP optimizes makespan, not weighted tardiness) - Release dates may not be fully enforced (JSP doesn’t have built-in release constraints)

class discrete_optimization.singlemachine.transformations.to_jsp.SingleMachineToJspTransformation[source]

Bases: ProblemTransformation[WeightedTardinessProblem, WTSolution, JobShopProblem, JobShopSolution]

Transform Single Machine Scheduling to Job Shop Problem.

Mapping: - Single machine → JSP with 1 machine - Each job → JSP job with 1 subjob - All subjobs assigned to machine 0 - Processing times map directly

Information Loss: - Due dates: Lost (JSP doesn’t model deadlines) - Weights: Lost (JSP only optimizes makespan) - Release dates: Partially lost (not enforced in standard JSP)

This transformation is LOSSY but still useful for: - Reusing powerful JSP solvers for single machine scheduling - Focusing on minimizing makespan (ignoring tardiness) - Benchmarking JSP solvers on single-machine instances

back_transform_solution(solution: JobShopSolution, source_problem: WeightedTardinessProblem) → WTSolution[source]

Transform JSP solution back to Single Machine solution.

Parameters:

  • solution – JSP solution (schedule for jobs on 1 machine)

  • source_problem – Original WeightedTardinessProblem

Returns:

Equivalent WTSolution (permutation + schedule)

forward_transform_solution(solution: WTSolution, target_problem: JobShopProblem) → JobShopSolution | None[source]

Transform Single Machine solution to JSP solution (for warmstart).

Parameters:

  • solution – WTSolution (permutation + schedule)

  • target_problem – Target JobShopProblem

Returns:

Equivalent JobShopSolution

get_forward_metadata() → TransformationMetadata[source]

Metadata for forward transformation (SingleMachine → JSP).

transform_problem(source_problem: WeightedTardinessProblem) → JobShopProblem[source]

Transform Single Machine problem to JSP.

Parameters:

source_problem – WeightedTardinessProblem instance

Returns:

Equivalent JobShopProblem (single machine, single subjob per job)

Module contents

Transformations from Single Machine Scheduling to other problems.

class discrete_optimization.singlemachine.transformations.SingleMachineToJspTransformation[source]

Bases: ProblemTransformation[WeightedTardinessProblem, WTSolution, JobShopProblem, JobShopSolution]

Transform Single Machine Scheduling to Job Shop Problem.

Mapping: - Single machine → JSP with 1 machine - Each job → JSP job with 1 subjob - All subjobs assigned to machine 0 - Processing times map directly

Information Loss: - Due dates: Lost (JSP doesn’t model deadlines) - Weights: Lost (JSP only optimizes makespan) - Release dates: Partially lost (not enforced in standard JSP)

This transformation is LOSSY but still useful for: - Reusing powerful JSP solvers for single machine scheduling - Focusing on minimizing makespan (ignoring tardiness) - Benchmarking JSP solvers on single-machine instances

back_transform_solution(solution: JobShopSolution, source_problem: WeightedTardinessProblem) → WTSolution[source]

Transform JSP solution back to Single Machine solution.

Parameters:

  • solution – JSP solution (schedule for jobs on 1 machine)

  • source_problem – Original WeightedTardinessProblem

Returns:

Equivalent WTSolution (permutation + schedule)

forward_transform_solution(solution: WTSolution, target_problem: JobShopProblem) → JobShopSolution | None[source]

Transform Single Machine solution to JSP solution (for warmstart).

Parameters:

  • solution – WTSolution (permutation + schedule)

  • target_problem – Target JobShopProblem

Returns:

Equivalent JobShopSolution

get_forward_metadata() → TransformationMetadata[source]

Metadata for forward transformation (SingleMachine → JSP).

transform_problem(source_problem: WeightedTardinessProblem) → JobShopProblem[source]

Transform Single Machine problem to JSP.

Parameters:

source_problem – WeightedTardinessProblem instance

Returns:

Equivalent JobShopProblem (single machine, single subjob per job)