----------------------------------------------------------------------------- -- | -- Module : Documentation.SBV.Examples.Queries.CaseSplit -- Copyright : (c) Levent Erkok -- License : BSD3 -- Maintainer : erkokl@gmail.com -- Stability : experimental -- -- A couple of demonstrations for the 'caseSplit' function. ----------------------------------------------------------------------------- module Documentation.SBV.Examples.Queries.CaseSplit where import Data.SBV import Data.SBV.Control -- | A simple floating-point problem, but we do the sat-analysis via a case-split. -- Due to the nature of floating-point numbers, a case-split on the characteristics -- of the number (such as NaN, negative-zero, etc. is most suitable.) -- -- We have: -- -- >>> csDemo1 -- Case fpIsNegativeZero: Starting -- Case fpIsNegativeZero: Unsatisfiable -- Case fpIsPositiveZero: Starting -- Case fpIsPositiveZero: Unsatisfiable -- Case fpIsNormal: Starting -- Case fpIsNormal: Unsatisfiable -- Case fpIsSubnormal: Starting -- Case fpIsSubnormal: Unsatisfiable -- Case fpIsPoint: Starting -- Case fpIsPoint: Unsatisfiable -- Case fpIsNaN: Starting -- Case fpIsNaN: Satisfiable -- ("fpIsNaN",NaN) csDemo1 :: IO (String, Float) csDemo1 = runSMT $ do x <- sFloat "x" constrain $ x ./= x -- yes, in the FP land, this does hold query $ do mbR <- caseSplit True [ ("fpIsNegativeZero", fpIsNegativeZero x) , ("fpIsPositiveZero", fpIsPositiveZero x) , ("fpIsNormal", fpIsNormal x) , ("fpIsSubnormal", fpIsSubnormal x) , ("fpIsPoint", fpIsPoint x) , ("fpIsNaN", fpIsNaN x) ] case mbR of Nothing -> error "Cannot find a FP number x such that x == x + 1" -- Won't happen! Just (s, _) -> do xv <- getValue x return (s, xv) -- | Demonstrates the "coverage" case. -- -- We have: -- -- >>> csDemo2 -- Case negative: Starting -- Case negative: Unsatisfiable -- Case less than 8: Starting -- Case less than 8: Unsatisfiable -- Case Coverage: Starting -- Case Coverage: Satisfiable -- ("Coverage",10) csDemo2 :: IO (String, Integer) csDemo2 = runSMT $ do x <- sInteger "x" constrain $ x .== 10 query $ do mbR <- caseSplit True [ ("negative" , x .< 0) , ("less than 8", x .< 8) ] case mbR of Nothing -> error "Cannot find a solution!" -- Won't happen! Just (s, _) -> do xv <- getValue x return (s, xv)