The MSplit primitive from LogicT paper.
Moduleextensible-effects-5.0.0.1Haskell2010
Control.Eff.Logic.Core
Logic primitives. See LogicT paper for details.
- 1 type
- 2 classes
- 11 values
- Packageextensible-effects-5.0.0.1
- Exports14
- LanguageHaskell2010
- LicenceMIT
- SourceCore.hs
Embed a pure value into MSplit
Laws for reflect:
msplit (lift m >> mzero) >>= reflect = lift m >> mzero
msplit (lift m `mplus` ma) >>= reflect = lift m `mplus` (msplit ma >>= reflect)Soft-cut: non-deterministic if-then-else, aka Prolog's *->
Declaratively,
ifte t th el = (t >>= th) `mplus` ((not t) >> el)However, t is evaluated only once. In other words, ifte t th el
is equivalent to t >>= th if t has at least one solution.
If t fails, ifte t th el is the same as el. Laws:
ifte (return a) th el = th a
ifte mzero th el = el
ifte (return a `mplus` m) th el = th a `mplus` (m >>= th)Another pruning operation (ifte is the other). This selects one solution out of possibly many.
Negation as failure
Fair (i.e., avoids starvation) disjunction. It obeys the following laws:
interleave mzero m = m
interleave (return a `mplus` m1) m2 = return a `mplus` (interleave m2 m1)corollary:
interleave m mzero = mFair (i.e., avoids starvation) conjunction. It obeys the following laws:
mzero >>- k = mzero
(return a `mplus` m) >>- k = interleave (k a) (m >>- k)Collect all solutions. This is from Hinze's Backtr monad
class. Unsurprisingly, this can be implemented in terms of msplit.
Non-determinism with control (cut).
For the explanation of cut, see Section 5 of Hinze ICFP 2000 paper:
Hinze suggests expressing cut in terms of cutfalse:
= return () `mplus` cutfalse
where
cutfalse :: m asatisfies the following laws:
cutfalse >>= k = cutfalse (F1)
cutfalse | m = cutfalse (F2)(note: m `mplus` cutfalse is different from cutfalse `mplus` m).
In other words, cutfalse is the left zero of both bind and mplus.
Hinze also introduces the operation call :: m a -> m a that
delimits the effect of cut: call m executes m. If the cut is
invoked in m, it discards only the choices made since m was called.
Hinze postulates the axioms of call:
call false = false (C1)
call (return a | m) = return a | call m (C2)
call (m | cutfalse) = call m (C3)
call (lift m >>= k) = lift m >>= (call . k) (C4)call m behaves like m except any cut inside m has only a local effect,
he says.
Hinze noted a problem with the "mechanical" derivation of backtracing monad transformer with cut: no axiom specifying the interaction of call with bind; no way to simplify nested invocations of call.
Methods
call :: MonadPlus (Eff r) => Eff (Exc CutFalse ': r) a -> Eff r aMapping
Backtrinterface to MonadPlus and using exceptions forcutfalse, every instance should ensure that the following laws hold:cutfalse `mplus` m = cutfalse --(F2) call mzero = mzero --(C1) call (return a `mplus` m) = return a `mplus` call m --(C2) call (m `mplus` cutfalse) = call m --(C3) call (lift m >>= k) = lift m >>= (call . k) --(C4)
Instances2Call
Member NDet r => Call rDefined in extensible-effects-5.0.0.1 · Control.Eff.Logic.NDet · orphanThe call interpreter -- it is like reify . reflect with a twist. Compare this implementation with the huge implementation of call in Hinze 2000 (Figure 9). Each clause corresponds to the axiom of call or cutfalse. All axioms are covered.
The code clearly expresses the intuition that call watches the choice points of its argument computation. When it encounteres a cutfalse request, it discards the remaining choicepoints. It completely handles CutFalse effects but not non-determinism
(MonadPlus (Eff (Exc CutFalse ': r)), MSplit (Eff (Exc CutFalse ': r))) => Call rDefined in extensible-effects-5.0.0.1 · Control.Eff.Logic.Experimental · orphan
We use exceptions for cutfalse
Therefore, the law cutfalse >>= k = cutfalse
is satisfied automatically since all exceptions have the above property.
Prolog cut, taken from Hinze 2000 (Deriving backtracking monad
transformers).
Case analysis for lists