Package gov.nasa.jpl.aerie.scheduler.solver

Interface Solver

All Known Implementing Classes:

PrioritySolver

public interface Solver

interface to a scheduling algorithm that produces schedules for input plans depending on the kind of solver and its configuration, the algorithm may produce one or many different solutions to the same planning problem: eg by iteratively improving on a solution, by providing different high level options, etc

Method Summary

Modifier and Type	Method	Description
Optional<Plan>	getNextSolution()	calculates the next solution plan based on solver configuration operates according to the solver's controlling configuration to determine a new solution, which may range from an iterative improvement on a prior solution up to a contrasting high level alternative for consideration in general, the solution trajectory is not guaranteed to be identical across different executions on the same problem, though some specific algorithms may optionally support deterministic solution trajectories in general, the solver method is not re-entrant safe, though some specific algorithms may optionally support parallel solution requests in general, the solver requires that any previously specified input planning problem descriptor and solver configuration remain unchanged between successive calls, though some algorithms may optionally support changing problem specification the algorithm may return no solution in the event the solver has expended all of its solutions (eg by reaching a requested quality threshold, proving an optimum solution, exhausting the configured alternative search space, etc).

Method Details

getNextSolution

Optional<Plan> getNextSolution()
                        throws SchedulingInterruptedException

calculates the next solution plan based on solver configuration operates according to the solver's controlling configuration to determine a new solution, which may range from an iterative improvement on a prior solution up to a contrasting high level alternative for consideration in general, the solution trajectory is not guaranteed to be identical across different executions on the same problem, though some specific algorithms may optionally support deterministic solution trajectories in general, the solver method is not re-entrant safe, though some specific algorithms may optionally support parallel solution requests in general, the solver requires that any previously specified input planning problem descriptor and solver configuration remain unchanged between successive calls, though some algorithms may optionally support changing problem specification the algorithm may return no solution in the event the solver has expended all of its solutions (eg by reaching a requested quality threshold, proving an optimum solution, exhausting the configured alternative search space, etc). in general, the solver will return no solution for all requests thereafter, but some algorithms may optionally support further solutions (eg on some input/configuration modification)

Returns:

an output schedule/plan that (possibly partially) solves the previously specified planning problem, or empty if the solver cannot provide any more solutions right now

Throws:

SchedulingInterruptedException