Packages

package solve

Linear Supertypes
AnyRef, Any
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. solve
  2. AnyRef
  3. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. All

Type Members

  1. class ConstantStrategy extends RaisingStrategy

    A solving strategy that applies the same solver and raising criteria to every problem, including nested problems.

  2. type RaisingCriteria = (NestedProblem[_]) ⇒ Boolean

    A raising criteria is a decision function that chooses whether or not to eliminate variables at the level of this nested problem, or to raise factors to the next higher problem.

    A raising criteria is a decision function that chooses whether or not to eliminate variables at the level of this nested problem, or to raise factors to the next higher problem. Returns true if the factors should be raised, or false if a solver should perform elimination.

  3. abstract class RaisingStrategy extends SolvingStrategy

    A solving strategy that handles subproblems.

    A solving strategy that handles subproblems. It either recursively raises subproblems, or solves them and raises their solutions.

  4. abstract class SolvingStrategy extends AnyRef

    A solving strategy solves an inference problem after a series of refinements have been made.

    A solving strategy solves an inference problem after a series of refinements have been made. This can involve solving subproblems and collecting their factors, but this class does not account for subproblems by default. Executing the strategy then eliminates variables that do not belong in the solution.

    By default, solving strategies do not use parameterized factors, but subclasses can override this.

  5. class VEBPGibbsStrategy extends RaisingStrategy

    A solving strategy that chooses between VE, BP, and Gibbs based on a score of the elminiation order and determinism

  6. class VEBPStrategy extends RaisingStrategy

    A solving strategy that chooses between VE and BP based on a score of the elminiation order

  7. class VEGibbsStrategy extends RaisingStrategy

    A solving strategy that chooses between VE and Gibbs based on a score of the elminiation order

Value Members

  1. def flatRaising(problem: NestedProblem[_]): Boolean

    Always raises a nested problem.

    Always raises a nested problem. This has the effect of "flattening" because all factors of problems strictly contained in the top-level problem are raised.

  2. def raiseIfGlobal(problem: NestedProblem[_]): Boolean

    Raises a nested problem if any of its components are global.

  3. def structuredRaising(problem: NestedProblem[_]): Boolean

    Never raises a nested problem without solving it.

Inherited from AnyRef

Inherited from Any

Ungrouped