--------------------------------------------------------------------------------- -- | -- Module : Data.SBV -- Copyright : (c) Levent Erkok -- License : BSD3 -- Maintainer : erkokl@gmail.com -- Stability : experimental -- -- (The sbv library is hosted at <http://github.com/LeventErkok/sbv>. -- Comments, bug reports, and patches are always welcome.) -- -- SBV: SMT Based Verification -- -- Express properties about Haskell programs and automatically prove -- them using SMT solvers. -- -- >>> prove $ \x -> x `shiftL` 2 .== 4 * (x :: SWord8) -- Q.E.D. -- -- >>> prove $ \x -> x `shiftL` 2 .== 2 * (x :: SWord8) -- Falsifiable. Counter-example: -- s0 = 32 :: Word8 -- -- The function 'prove' has the following type: -- -- @ -- 'prove' :: 'Provable' a => a -> 'IO' 'ThmResult' -- @ -- -- The class 'Provable' comes with instances for n-ary predicates, for arbitrary n. -- The predicates are just regular Haskell functions over symbolic types listed below. -- Functions for checking satisfiability ('sat' and 'allSat') are also -- provided. -- -- The sbv library introduces the following symbolic types: -- -- * 'SBool': Symbolic Booleans (bits). -- -- * 'SWord8', 'SWord16', 'SWord32', 'SWord64': Symbolic Words (unsigned). -- -- * 'SInt8', 'SInt16', 'SInt32', 'SInt64': Symbolic Ints (signed). -- -- * 'SInteger': Unbounded signed integers. -- -- * 'SReal': Algebraic-real numbers -- -- * 'SFloat': IEEE-754 single-precision floating point values -- -- * 'SDouble': IEEE-754 double-precision floating point values -- -- * 'SArray', 'SFunArray': Flat arrays of symbolic values. -- -- * Symbolic polynomials over GF(2^n), polynomial arithmetic, and CRCs. -- -- * Uninterpreted constants and functions over symbolic values, with user -- defined SMT-Lib axioms. -- -- * Uninterpreted sorts, and proofs over such sorts, potentially with axioms. -- -- The user can construct ordinary Haskell programs using these types, which behave -- very similar to their concrete counterparts. In particular these types belong to the -- standard classes 'Num', 'Bits', custom versions of 'Eq' ('EqSymbolic') -- and 'Ord' ('OrdSymbolic'), along with several other custom classes for simplifying -- programming with symbolic values. The framework takes full advantage of Haskell's type -- inference to avoid many common mistakes. -- -- Furthermore, predicates (i.e., functions that return 'SBool') built out of -- these types can also be: -- -- * proven correct via an external SMT solver (the 'prove' function) -- -- * checked for satisfiability (the 'sat', 'allSat' functions) -- -- * used in synthesis (the `sat` function with existentials) -- -- * quick-checked -- -- If a predicate is not valid, 'prove' will return a counterexample: An -- assignment to inputs such that the predicate fails. The 'sat' function will -- return a satisfying assignment, if there is one. The 'allSat' function returns -- all satisfying assignments, lazily. -- -- The sbv library uses third-party SMT solvers via the standard SMT-Lib interface: -- <http://smtlib.cs.uiowa.edu/> -- -- The SBV library is designed to work with any SMT-Lib compliant SMT-solver. -- Currently, we support the following SMT-Solvers out-of-the box: -- -- * ABC from University of Berkeley: <http://www.eecs.berkeley.edu/~alanmi/abc/> -- -- * CVC4 from New York University and University of Iowa: <http://cvc4.cs.nyu.edu/> -- -- * Boolector from Johannes Kepler University: <http://fmv.jku.at/boolector/> -- -- * MathSAT from Fondazione Bruno Kessler and DISI-University of Trento: <http://mathsat.fbk.eu/> -- -- * Yices from SRI: <http://yices.csl.sri.com/> -- -- * Z3 from Microsoft: <http://github.com/Z3Prover/z3/wiki> -- -- SBV also allows calling these solvers in parallel, either getting results from multiple solvers -- or returning the fastest one. (See 'proveWithAll', 'proveWithAny', etc.) -- -- Support for other compliant solvers can be added relatively easily, please -- get in touch if there is a solver you'd like to see included. --------------------------------------------------------------------------------- {-# LANGUAGE FlexibleInstances #-} {-# OPTIONS_GHC -fno-warn-orphans #-} module Data.SBV ( -- * Programming with symbolic values -- $progIntro -- ** Symbolic types -- *** Symbolic bit SBool -- *** Unsigned symbolic bit-vectors , SWord8, SWord16, SWord32, SWord64 -- *** Signed symbolic bit-vectors , SInt8, SInt16, SInt32, SInt64 -- *** Signed unbounded integers -- $unboundedLimitations , SInteger -- *** IEEE-floating point numbers -- $floatingPoints , SFloat, SDouble, IEEEFloating(..), IEEEFloatConvertable(..), RoundingMode(..), SRoundingMode, nan, infinity, sNaN, sInfinity -- **** Rounding modes , sRoundNearestTiesToEven, sRoundNearestTiesToAway, sRoundTowardPositive, sRoundTowardNegative, sRoundTowardZero, sRNE, sRNA, sRTP, sRTN, sRTZ -- **** Bit-pattern conversions , sFloatAsSWord32, sWord32AsSFloat, sDoubleAsSWord64, sWord64AsSDouble, blastSFloat, blastSDouble -- *** Signed algebraic reals -- $algReals , SReal, AlgReal, sRealToSInteger -- ** Creating a symbolic variable -- $createSym , sBool, sWord8, sWord16, sWord32, sWord64, sInt8, sInt16, sInt32, sInt64, sInteger, sReal, sFloat, sDouble -- ** Creating a list of symbolic variables -- $createSyms , sBools, sWord8s, sWord16s, sWord32s, sWord64s, sInt8s, sInt16s, sInt32s, sInt64s, sIntegers, sReals, sFloats, sDoubles -- *** Abstract SBV type , SBV -- *** Arrays of symbolic values , SymArray(..), SArray, SFunArray, mkSFunArray -- ** Operations on symbolic values -- *** Word level , sTestBit, sExtractBits, sPopCount, sShiftLeft, sShiftRight, sRotateLeft, sRotateRight, sSignedShiftArithRight, sFromIntegral, setBitTo, oneIf , lsb, msb, label -- *** Addition and Multiplication with high-bits , fullAdder, fullMultiplier -- *** Exponentiation , (.^) -- *** Blasting/Unblasting , blastBE, blastLE, FromBits(..) -- *** Splitting, joining, and extending , Splittable(..) -- ** Conditionals: Mergeable values , Mergeable(..), ite, iteLazy -- ** Symbolic integral numbers , SIntegral -- ** Division , SDivisible(..) -- ** The Boolean class , Boolean(..) -- *** Generalizations of boolean operations , bAnd, bOr, bAny, bAll -- ** Pretty-printing and reading numbers in Hex & Binary , PrettyNum(..), readBin -- * Checking satisfiability in path conditions , isSatisfiableInCurrentPath -- * Uninterpreted sorts, constants, and functions -- $uninterpreted , Uninterpreted(..), addAxiom -- * Symbolic Equality and Comparisons , EqSymbolic(..), OrdSymbolic(..) -- * Cardinality constraints -- $cardIntro , pbAtMost, pbAtLeast, pbExactly, pbLe, pbGe, pbEq, pbMutexed, pbStronglyMutexed -- * Enumerations -- $enumerations -- * Properties, proofs, satisfiability, and safety -- $proveIntro -- $noteOnNestedQuantifiers -- ** Predicates and Goals , Predicate, Goal, Provable(..), Equality(..) -- ** Proving properties , prove, proveWith, isTheorem, isTheoremWith -- ** Checking satisfiability , sat, satWith, isSatisfiable, isSatisfiableWith -- ** Checking safety -- $safeIntro , sAssert, safe, safeWith, isSafe, SExecutable(..) -- ** Finding all satisfying assignments , allSat, allSatWith -- ** Satisfying a sequence of boolean conditions , solve -- ** Adding constraints -- $constrainIntro , constrain, namedConstraint, pConstrain -- ** Checking constraint vacuity , isVacuous, isVacuousWith -- ** Quick-checking , sbvQuickCheck -- * Proving properties using multiple solvers -- $multiIntro , proveWithAll, proveWithAny, satWithAll, satWithAny -- * Tactics -- $tacticIntro , Tactic(..), tactic -- * Optimization -- $optiIntro , OptimizeStyle(..), Penalty(..), Objective(..), minimize, maximize, assertSoft, optimize, optimizeWith , ExtCW(..), GeneralizedCW(..) -- * Model extraction -- $modelExtraction -- ** Inspecting proof results -- $resultTypes , ThmResult(..), SatResult(..), AllSatResult(..), SafeResult(..), OptimizeResult(..), SMTResult(..) -- ** Programmable model extraction -- $programmableExtraction , SatModel(..), Modelable(..), displayModels, extractModels , getModelDictionaries, getModelValues, getModelUninterpretedValues -- * SMT Interface: Configurations and solvers , SMTConfig(..), SMTLibVersion(..), SMTLibLogic(..), Logic(..), Solver(..), SMTSolver(..) , boolector, cvc4, yices, z3, mathSAT, abc, defaultSolverConfig, sbvCurrentSolver, defaultSMTCfg, sbvCheckSolverInstallation, sbvAvailableSolvers , Timing(..), TimedStep(..), TimingInfo, showTDiff, CW(..), HasKind(..), Kind(..), cwToBool -- * Symbolic computations , Symbolic, output, SymWord(..) -- * Module exports -- $moduleExportIntro , module Data.Bits , module Data.Word , module Data.Int , module Data.Ratio ) where import Control.Monad (filterM) import Control.Concurrent.Async (async, waitAny, waitAnyCancel) import System.IO.Unsafe (unsafeInterleaveIO) -- only used safely! import Data.SBV.Core.AlgReals import Data.SBV.Core.Data import Data.SBV.Core.Model import Data.SBV.Core.Floating import Data.SBV.Core.Splittable import Data.SBV.Provers.Prover import Data.SBV.Utils.Boolean import Data.SBV.Utils.TDiff import Data.SBV.Utils.PrettyNum import Data.Bits import Data.Int import Data.Ratio import Data.Word -- | The currently active solver, obtained by importing "Data.SBV". -- To have other solvers /current/, import one of the bridge -- modules "Data.SBV.Bridge.ABC", "Data.SBV.Bridge.Boolector", "Data.SBV.Bridge.CVC4", -- "Data.SBV.Bridge.Yices", or "Data.SBV.Bridge.Z3" directly. sbvCurrentSolver :: SMTConfig sbvCurrentSolver = z3 -- | Form the symbolic conjunction of a given list of boolean conditions. Useful in expressing -- problems with constraints, like the following: -- -- @ -- do [x, y, z] <- sIntegers [\"x\", \"y\", \"z\"] -- solve [x .> 5, y + z .< x] -- @ solve :: [SBool] -> Symbolic SBool solve = return . bAnd -- | Check whether the given solver is installed and is ready to go. This call does a -- simple call to the solver to ensure all is well. sbvCheckSolverInstallation :: SMTConfig -> IO Bool sbvCheckSolverInstallation cfg = do ThmResult r <- proveWith cfg $ \x -> (x+x) .== ((x*2) :: SWord8) case r of Unsatisfiable{} -> return True _ -> return False -- | The default configs corresponding to supported SMT solvers defaultSolverConfig :: Solver -> SMTConfig defaultSolverConfig Z3 = z3 defaultSolverConfig Yices = yices defaultSolverConfig Boolector = boolector defaultSolverConfig CVC4 = cvc4 defaultSolverConfig MathSAT = mathSAT defaultSolverConfig ABC = abc -- | Return the known available solver configs, installed on your machine. sbvAvailableSolvers :: IO [SMTConfig] sbvAvailableSolvers = filterM sbvCheckSolverInstallation (map defaultSolverConfig [minBound .. maxBound]) sbvWithAny :: [SMTConfig] -> (SMTConfig -> a -> IO b) -> a -> IO (Solver, b) sbvWithAny [] _ _ = error "SBV.withAny: No solvers given!" sbvWithAny solvers what a = snd `fmap` (mapM try solvers >>= waitAnyCancel) where try s = async $ what s a >>= \r -> return (name (solver s), r) sbvWithAll :: [SMTConfig] -> (SMTConfig -> a -> IO b) -> a -> IO [(Solver, b)] sbvWithAll solvers what a = mapM try solvers >>= (unsafeInterleaveIO . go) where try s = async $ what s a >>= \r -> return (name (solver s), r) go [] = return [] go as = do (d, r) <- waitAny as -- The following filter works because the Eq instance on Async -- checks the thread-id; so we know that we're removing the -- correct solver from the list. This also allows for -- running the same-solver (with different options), since -- they will get different thread-ids. rs <- unsafeInterleaveIO $ go (filter (/= d) as) return (r : rs) -- | Prove a property with multiple solvers, running them in separate threads. The -- results will be returned in the order produced. proveWithAll :: Provable a => [SMTConfig] -> a -> IO [(Solver, ThmResult)] proveWithAll = (`sbvWithAll` proveWith) -- | Prove a property with multiple solvers, running them in separate threads. Only -- the result of the first one to finish will be returned, remaining threads will be killed. proveWithAny :: Provable a => [SMTConfig] -> a -> IO (Solver, ThmResult) proveWithAny = (`sbvWithAny` proveWith) -- | Find a satisfying assignment to a property with multiple solvers, running them in separate threads. The -- results will be returned in the order produced. satWithAll :: Provable a => [SMTConfig] -> a -> IO [(Solver, SatResult)] satWithAll = (`sbvWithAll` satWith) -- | Find a satisfying assignment to a property with multiple solvers, running them in separate threads. Only -- the result of the first one to finish will be returned, remaining threads will be killed. satWithAny :: Provable a => [SMTConfig] -> a -> IO (Solver, SatResult) satWithAny = (`sbvWithAny` satWith) -- If we get a program producing nothing (i.e., Symbolic ()), pretend it simply returns True. -- This is useful since min/max calls and constraints will provide the context instance Provable Goal where forAll_ a = forAll_ ((a >> return true) :: Predicate) forAll ns a = forAll ns ((a >> return true) :: Predicate) forSome_ a = forSome_ ((a >> return true) :: Predicate) forSome ns a = forSome ns ((a >> return true) :: Predicate) -- | Equality as a proof method. Allows for -- very concise construction of equivalence proofs, which is very typical in -- bit-precise proofs. infix 4 === class Equality a where (===) :: a -> a -> IO ThmResult instance {-# OVERLAPPABLE #-} (SymWord a, EqSymbolic z) => Equality (SBV a -> z) where k === l = prove $ \a -> k a .== l a instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, EqSymbolic z) => Equality (SBV a -> SBV b -> z) where k === l = prove $ \a b -> k a b .== l a b instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, EqSymbolic z) => Equality ((SBV a, SBV b) -> z) where k === l = prove $ \a b -> k (a, b) .== l (a, b) instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, EqSymbolic z) => Equality (SBV a -> SBV b -> SBV c -> z) where k === l = prove $ \a b c -> k a b c .== l a b c instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, EqSymbolic z) => Equality ((SBV a, SBV b, SBV c) -> z) where k === l = prove $ \a b c -> k (a, b, c) .== l (a, b, c) instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, EqSymbolic z) => Equality (SBV a -> SBV b -> SBV c -> SBV d -> z) where k === l = prove $ \a b c d -> k a b c d .== l a b c d instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, EqSymbolic z) => Equality ((SBV a, SBV b, SBV c, SBV d) -> z) where k === l = prove $ \a b c d -> k (a, b, c, d) .== l (a, b, c, d) instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, SymWord e, EqSymbolic z) => Equality (SBV a -> SBV b -> SBV c -> SBV d -> SBV e -> z) where k === l = prove $ \a b c d e -> k a b c d e .== l a b c d e instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, SymWord e, EqSymbolic z) => Equality ((SBV a, SBV b, SBV c, SBV d, SBV e) -> z) where k === l = prove $ \a b c d e -> k (a, b, c, d, e) .== l (a, b, c, d, e) instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, SymWord e, SymWord f, EqSymbolic z) => Equality (SBV a -> SBV b -> SBV c -> SBV d -> SBV e -> SBV f -> z) where k === l = prove $ \a b c d e f -> k a b c d e f .== l a b c d e f instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, SymWord e, SymWord f, EqSymbolic z) => Equality ((SBV a, SBV b, SBV c, SBV d, SBV e, SBV f) -> z) where k === l = prove $ \a b c d e f -> k (a, b, c, d, e, f) .== l (a, b, c, d, e, f) instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, SymWord e, SymWord f, SymWord g, EqSymbolic z) => Equality (SBV a -> SBV b -> SBV c -> SBV d -> SBV e -> SBV f -> SBV g -> z) where k === l = prove $ \a b c d e f g -> k a b c d e f g .== l a b c d e f g instance {-# OVERLAPPABLE #-} (SymWord a, SymWord b, SymWord c, SymWord d, SymWord e, SymWord f, SymWord g, EqSymbolic z) => Equality ((SBV a, SBV b, SBV c, SBV d, SBV e, SBV f, SBV g) -> z) where k === l = prove $ \a b c d e f g -> k (a, b, c, d, e, f, g) .== l (a, b, c, d, e, f, g) -- Haddock section documentation {- $progIntro The SBV library is really two things: * A framework for writing symbolic programs in Haskell, i.e., programs operating on symbolic values along with the usual concrete counterparts. * A framework for proving properties of such programs using SMT solvers. The programming goal of SBV is to provide a /seamless/ experience, i.e., let people program in the usual Haskell style without distractions of symbolic coding. While Haskell helps in some aspects (the 'Num' and 'Bits' classes simplify coding), it makes life harder in others. For instance, @if-then-else@ only takes 'Bool' as a test in Haskell, and comparisons ('>' etc.) only return 'Bool's. Clearly we would like these values to be symbolic (i.e., 'SBool'), thus stopping us from using some native Haskell constructs. When symbolic versions of operators are needed, they are typically obtained by prepending a dot, for instance '==' becomes '.=='. Care has been taken to make the transition painless. In particular, any Haskell program you build out of symbolic components is fully concretely executable within Haskell, without the need for any custom interpreters. (They are truly Haskell programs, not AST's built out of pieces of syntax.) This provides for an integrated feel of the system, one of the original design goals for SBV. -} {- $proveIntro The SBV library provides a "push-button" verification system via automated SMT solving. The design goal is to let SMT solvers be used without any knowledge of how SMT solvers work or how different logics operate. The details are hidden behind the SBV framework, providing Haskell programmers with a clean API that is unencumbered by the details of individual solvers. To that end, we use the SMT-Lib standard (<http://smtlib.cs.uiowa.edu/>) to communicate with arbitrary SMT solvers. -} {- $multiIntro On a multi-core machine, it might be desirable to try a given property using multiple SMT solvers, using parallel threads. Even with machines with single-cores, threading can be helpful if you want to try out multiple-solvers but do not know which one would work the best for the problem at hand ahead of time. The functions in this section allow proving/satisfiability-checking with multiple backends at the same time. Each function comes in two variants, one that returns the results from all solvers, the other that returns the fastest one. The @All@ variants, (i.e., 'proveWithAll', 'satWithAll') run all solvers and return all the results. SBV internally makes sure that the result is lazily generated; so, the order of solvers given does not matter. In other words, the order of results will follow the order of the solvers as they finish, not as given by the user. These variants are useful when you want to make sure multiple-solvers agree (or disagree!) on a given problem. The @Any@ variants, (i.e., 'proveWithAny', 'satWithAny') will run all the solvers in parallel, and return the results of the first one finishing. The other threads will then be killed. These variants are useful when you do not care if the solvers produce the same result, but rather want to get the solution as quickly as possible, taking advantage of modern many-core machines. Note that the function 'sbvAvailableSolvers' will return all the installed solvers, which can be used as the first argument to all these functions, if you simply want to try all available solvers on a machine. -} {- $safeIntro The 'sAssert' function allows users to introduce invariants to make sure certain properties hold at all times. This is another mechanism to provide further documentation/contract info into SBV code. The functions 'safe' and 'safeWith' can be used to statically discharge these proof assumptions. If a violation is found, SBV will print a model showing which inputs lead to the invariant being violated. Here's a simple example. Let's assume we have a function that does subtraction, and requires its first argument to be larger than the second: >>> let sub x y = sAssert Nothing "sub: x >= y must hold!" (x .>= y) (x - y) Clearly, this function is not safe, as there's nothing that stops us from passing it a larger second argument. We can use 'safe' to statically see if such a violation is possible before we use this function elsewhere. >>> safe (sub :: SInt8 -> SInt8 -> SInt8) [sub: x >= y must hold!: Violated. Model: s0 = -128 :: Int8 s1 = -127 :: Int8] What happens if we make sure to arrange for this invariant? Consider this version: >>> let safeSub x y = ite (x .>= y) (sub x y) 0 Clearly, 'safeSub' must be safe. And indeed, SBV can prove that: >>> safe (safeSub :: SInt8 -> SInt8 -> SInt8) [sub: x >= y must hold!: No violations detected] Note how we used 'sub' and 'safeSub' polymorphically. We only need to monomorphise our types when a proof attempt is done, as we did in the 'safe' calls. If required, the user can pass a 'CallStack' through the first argument to 'sAssert', which will be used by SBV to print a diagnostic info to pinpoint the failure. Also see "Data.SBV.Examples.Misc.NoDiv0" for the classic div-by-zero example. -} {- $tacticIntro In certain cases, the prove/sat calls can benefit from user guidance, in terms of tactics. From a semantic view, a tactic has no effect on the meaning of a predicate. It is merely guidance for SBV to guide the proof. It is also used for executing cases in parallel ('ParallelCase'), or picking the logic to use ('UseLogic'), or specifying a timeout ('StopAfter'). For most users, default values of these should suffice. -} {- $optiIntro SBV can optimize metric functions, i.e., those that generate both bounded 'SIntN', 'SWordN', and unbounded 'SInteger' types, along with those produce 'SReal's. That is, it can find models satisfying all the constraints while minimizing or maximizing user given metrics. Currently, optimization requires the use of the z3 SMT solver as the backend, and a good review of these features is given in this paper: <http://www.easychair.org/publications/download/Z_-_Maximal_Satisfaction_with_Z3>. Goals can be lexicographically (default), independently, or pareto-front optimized. The relevant functions are: * 'minimize': Minimize a given arithmetic goal * 'maximize': Minimize a given arithmetic goal Goals can be optimized at a regular or an extended value: An extended value is either positive or negative infinity (for unbounded integers and reals) or positive or negative epsilon differential from a real value (for reals). For instance, a call of the form @ 'minimize' "name-of-goal" $ x + 2*y @ minimizes the arithmetic goal @x+2*y@, where @x@ and @y@ can be signed\/unsigned bit-vectors, reals, or integers. == A simple example Here's an optimization example in action: >>> optimize $ \x y -> minimize "goal" (x+2*(y::SInteger)) Optimal in an extension field: goal = -oo :: Integer Of course, this becomes more useful when the result is not in an extension field: @ optimize $ do x <- sInteger "x" y <- sInteger "y" constrain $ x .> 0 constrain $ x .< 6 constrain $ y .> 2 constrain $ y .< 12 minimize "goal" (x+2*(y::SInteger)) @ This will produce: @ Optimal model: x = 1 :: Integer y = 3 :: Integer goal = 7 :: Integer @ As usual, the programmatic API can be used to extract the values of objectives and model-values ('getModelObjectives', 'getModel', etc.) to access these values and program with them further. == Multiple optimization goals Multiple goals can be specified, using the same syntax. In this case, the user gets to pick what style of optimization to perform: * The default is lexicographic. That is, solver will optimize the goals in the given order, optimizing the latter ones under the model that optimizes the previous ones. This is the default behavior, but can also be explicitly specified by: @ 'tactic' $ 'OptimizePriority' 'Lexicographic' @ * Goals can also be independently optimized. In this case the user will be presented a model for each goal given. To enable this, use the tactic: @ 'tactic' $ 'OptimizePriority' 'Independent' @ * Finally, the user can query for pareto-fronts. A pareto front is an model such that no goal can be made "better" without making some other goal "worse." To enable this style, use: @ 'tactic' $ 'OptimizePriority' 'Pareto' @ == Soft Assertions Related to optimization, SBV implements soft-asserts via 'assertSoft' calls. A soft assertion is a hint to the SMT solver that we would like a particular condition to hold if **possible*. That is, if there is a solution satisfying it, then we would like it to hold, but it can be violated if there is no way to satisfy it. Each soft-assertion can be associated with a numeric penalty for not satisfying it, hence turning it into an optimization problem. Note that 'assertSoft' works well with optimization goals ('minimize'/'maximize' etc.), and are most useful when we are optimizing a metric and thus some of the constraints can be relaxed with a penalty to obtain a good solution. Again see <http://www.easychair.org/publications/download/Z_-_Maximal_Satisfaction_with_Z3> for a good overview of the features in Z3 that SBV is providing the bridge for. A soft assertion can be specified in one of the following three main ways: @ 'assertSoft' "bounded_x" (x .< 5) 'DefaultPenalty' 'assertSoft' "bounded_x" (x .< 5) ('Penalty' 2.3 Nothing) 'assertSoft' "bounded_x" (x .< 5) ('Penalty' 4.7 (Just "group-1")) @ In the first form, we are saying that the constraint @x .< 5@ must be satisfied, if possible, but if this constraint can not be satisfied to find a model, it can be violated with the default penalty of 1. In the second case, we are associating a penalty value of @2.3@. Finally in the third case, we are also associating this constraint with a group. The group name is only needed if we have classes of soft-constraints that should be considered together. == Optimization examples The following examples illustrate the use of basic optimization routines: * "Data.SBV.Examples.Optimization.LinearOpt": Simple linear-optimization example. * "Data.SBV.Examples.Optimization.Production": Scheduling machines in a shop * "Data.SBV.Examples.Optimization.VM": Scheduling virtual-machines in a data-center -} {- $modelExtraction The default 'Show' instances for prover calls provide all the counter-example information in a human-readable form and should be sufficient for most casual uses of sbv. However, tools built on top of sbv will inevitably need to look into the constructed models more deeply, programmatically extracting their results and performing actions based on them. The API provided in this section aims at simplifying this task. -} {- $resultTypes 'ThmResult', 'SatResult', and 'AllSatResult' are simple newtype wrappers over 'SMTResult'. Their main purpose is so that we can provide custom 'Show' instances to print results accordingly. -} {- $programmableExtraction While default 'Show' instances are sufficient for most use cases, it is sometimes desirable (especially for library construction) that the SMT-models are reinterpreted in terms of domain types. Programmable extraction allows getting arbitrarily typed models out of SMT models. -} {- $moduleExportIntro The SBV library exports the following modules wholesale, as user programs will have to import these modules to make any sensible use of the SBV functionality. -} {- $createSym These functions simplify declaring symbolic variables of various types. Strictly speaking, they are just synonyms for 'free' (specialized at the given type), but they might be easier to use. -} {- $createSyms These functions simplify declaring a sequence symbolic variables of various types. Strictly speaking, they are just synonyms for 'mapM' 'free' (specialized at the given type), but they might be easier to use. -} {- $unboundedLimitations The SBV library supports unbounded signed integers with the type 'SInteger', which are not subject to overflow/underflow as it is the case with the bounded types, such as 'SWord8', 'SInt16', etc. However, some bit-vector based operations are /not/ supported for the 'SInteger' type while in the verification mode. That is, you can use these operations on 'SInteger' values during normal programming/simulation. but the SMT translation will not support these operations since there corresponding operations are not supported in SMT-Lib. Note that this should rarely be a problem in practice, as these operations are mostly meaningful on fixed-size bit-vectors. The operations that are restricted to bounded word/int sizes are: * Rotations and shifts: 'rotateL', 'rotateR', 'shiftL', 'shiftR' * Bitwise logical ops: '.&.', '.|.', 'xor', 'complement' * Extraction and concatenation: 'split', '#', and 'extend' (see the 'Splittable' class) Usual arithmetic ('+', '-', '*', 'sQuotRem', 'sQuot', 'sRem', 'sDivMod', 'sDiv', 'sMod') and logical operations ('.<', '.<=', '.>', '.>=', '.==', './=') operations are supported for 'SInteger' fully, both in programming and verification modes. -} {- $algReals Algebraic reals are roots of single-variable polynomials with rational coefficients. (See <http://en.wikipedia.org/wiki/Algebraic_number>.) Note that algebraic reals are infinite precision numbers, but they do not cover all /real/ numbers. (In particular, they cannot represent transcendentals.) Some irrational numbers are algebraic (such as @sqrt 2@), while others are not (such as pi and e). SBV can deal with real numbers just fine, since the theory of reals is decidable. (See <http://smtlib.cs.uiowa.edu/theories-Reals.shtml>.) In addition, by leveraging backend solver capabilities, SBV can also represent and solve non-linear equations involving real-variables. (For instance, the Z3 SMT solver, supports polynomial constraints on reals starting with v4.0.) -} {- $floatingPoints Floating point numbers are defined by the IEEE-754 standard; and correspond to Haskell's 'Float' and 'Double' types. For SMT support with floating-point numbers, see the paper by Rummer and Wahl: <http://www.philipp.ruemmer.org/publications/smt-fpa.pdf>. -} {- $constrainIntro A constraint is a means for restricting the input domain of a formula. Here's a simple example: @ do x <- 'exists' \"x\" y <- 'exists' \"y\" 'constrain' $ x .> y 'constrain' $ x + y .>= 12 'constrain' $ y .>= 3 ... @ The first constraint requires @x@ to be larger than @y@. The scond one says that sum of @x@ and @y@ must be at least @12@, and the final one says that @y@ to be at least @3@. Constraints provide an easy way to assert additional properties on the input domain, right at the point of the introduction of variables. Note that the proper reading of a constraint depends on the context: * In a 'sat' (or 'allSat') call: The constraint added is asserted conjunctively. That is, the resulting satisfying model (if any) will always satisfy all the constraints given. * In a 'prove' call: In this case, the constraint acts as an implication. The property is proved under the assumption that the constraint holds. In other words, the constraint says that we only care about the input space that satisfies the constraint. * In a 'quickCheck' call: The constraint acts as a filter for 'quickCheck'; if the constraint does not hold, then the input value is considered to be irrelevant and is skipped. Note that this is similar to 'prove', but is stronger: We do not accept a test case to be valid just because the constraints fail on them, although semantically the implication does hold. We simply skip that test case as a /bad/ test vector. * In a 'genTest' call: Similar to 'quickCheck' and 'prove': If a constraint does not hold, the input value is ignored and is not included in the test set. A good use case (in fact the motivating use case) for 'constrain' is attaching a constraint to a 'forall' or 'exists' variable at the time of its creation. Also, the conjunctive semantics for 'sat' and the implicative semantics for 'prove' simplify programming by choosing the correct interpretation automatically. However, one should be aware of the semantic difference. For instance, in the presence of constraints, formulas that are /provable/ are not necessarily /satisfiable/. To wit, consider: @ do x <- 'exists' \"x\" 'constrain' $ x .< x return $ x .< (x :: 'SWord8') @ This predicate is unsatisfiable since no element of 'SWord8' is less than itself. But it's (vacuously) true, since it excludes the entire domain of values, thus making the proof trivial. Hence, this predicate is provable, but is not satisfiable. To make sure the given constraints are not vacuous, the functions 'isVacuous' (and 'isVacuousWith') can be used. Also note that this semantics imply that test case generation ('genTest') and quick-check can take arbitrarily long in the presence of constraints, if the random input values generated rarely satisfy the constraints. (As an extreme case, consider @'constrain' 'false'@.) A probabilistic constraint (see 'pConstrain') attaches a probability threshold for the constraint to be considered. For instance: @ 'pConstrain' 0.8 c @ will make sure that the condition @c@ is satisfied 80% of the time (and correspondingly, falsified 20% of the time), in expectation. This variant is useful for 'genTest' and 'quickCheck' functions, where we want to filter the test cases according to some probability distribution, to make sure that the test-vectors are drawn from interesting subsets of the input space. For instance, if we were to generate 100 test cases with the above constraint, we'd expect about 80 of them to satisfy the condition @c@, while about 20 of them will fail it. The following properties hold: @ 'constrain' = 'pConstrain' 1 'pConstrain' t c = 'pConstrain' (1-t) (not c) @ Note that while 'constrain' can be used freely, 'pConstrain' is only allowed in the contexts of 'genTest' or 'quickCheck'. Calls to 'pConstrain' in a prove/sat call will be rejected as SBV does not deal with probabilistic constraints when it comes to satisfiability and proofs. Also, both 'constrain' and 'pConstrain' calls during code-generation will also be rejected, for similar reasons. === Named constraints and unsat cores Constraints can be given names: @ 'namedConstraint' "a is at least 5" $ a .>= 5@ Such constraints are useful when used in conjunction with 'getUnsatCore', and 'extractUnsatCore' features, where the backend solver can be queried to obtain an unsat core in case the constraints are unsatisfiable: @ satWith z3{getUnsatCore=True} $ do ... @ See "Data.SBV.Examples.Misc.UnsatCore" for an example use case. === Constraint vacuity SBV does not check that a given constraints is not vacuous. That is, that it can never be satisfied. This is usually the right behavior, since checking vacuity can be costly. The functions 'isVacuous' and 'isVacuousWith' should be used to explicitly check for constraint vacuity if desired. Alternatively, the tactic: @ 'tactic' $ 'CheckConstrVacuity' True @ can be given which will force SBV to run an explicit check that constraints are not vacuous. (And complain if they are!) Note that this adds an extra call to the solver for each constraint, and thus can be rather costly. -} {- $uninterpreted Users can introduce new uninterpreted sorts simply by defining a data-type in Haskell and registering it as such. The following example demonstrates: @ data B = B () deriving (Eq, Ord, Show, Read, Data, SymWord, HasKind, SatModel) @ (Note that you'll also need to use the language pragmas @DeriveDataTypeable@, @DeriveAnyClass@, and import @Data.Generics@ for the above to work.) This is all it takes to introduce 'B' as an uninterpreted sort in SBV, which makes the type @SBV B@ automagically become available as the type of symbolic values that ranges over 'B' values. Note that the @()@ argument is important to distinguish it from enumerations. Uninterpreted functions over both uninterpreted and regular sorts can be declared using the facilities introduced by the 'Uninterpreted' class. -} {- $enumerations If the uninterpreted sort definition takes the form of an enumeration (i.e., a simple data type with all nullary constructors), then SBV will actually translate that as just such a data-type to SMT-Lib, and will use the constructors as the inhabitants of the said sort. A simple example is: @ data X = A | B | C deriving (Eq, Ord, Show, Read, Data, SymWord, HasKind, SatModel) @ Now, the user can define @ type SX = SBV X @ and treat @SX@ as a regular symbolic type ranging over the values @A@, @B@, and @C@. Such values can be compared for equality, and with the usual other comparison operators, such as @.==@, @./=@, @.>@, @.>=@, @<@, and @<=@. Note that in this latter case the type is no longer uninterpreted, but is properly represented as a simple enumeration of the said elements. A simple query would look like: @ allSat $ \x -> x .== (x :: SX) @ which would list all three elements of this domain as satisfying solutions. @ Solution #1: s0 = A :: X Solution #2: s0 = B :: X Solution #3: s0 = C :: X Found 3 different solutions. @ Note that the result is properly typed as @X@ elements; these are not mere strings. So, in a 'getModel' scenario, the user can recover actual elements of the domain and program further with those values as usual. -} {- $noteOnNestedQuantifiers === A note on reasoning in the presence of quantifers Note that SBV allows reasoning with quantifiers: Inputs can be existentially or universally quantified. Predicates can be built with arbitrary nesting of such quantifiers as well. However, SBV always /assumes/ that the input is in prenex-normal form: <https://en.wikipedia.org/wiki/Prenex_normal_form>. That is, all the input declarations are treated as happening at the beginning of a predicate, followed by the actual formula. Unfortunately, the way predicates are written can be misleading at times, since symbolic inputs can be created at arbitrary points; interleaving them with other code. The rule is simple, however: All inputs are assumed at the top, in the order declared, regardless of their quantifiers. SBV will apply skolemization to get rid of existentials before sending predicates to backend solvers. However, if you do want nested quantification, you will manually have to first convert to prenex-normal form (which produces an equisatisfiable but not necessarily equivalent formula), and code that explicitly in SBV. See <https://github.com/LeventErkok/sbv/issues/256> for a detailed discussion of this issue. -} {- $cardIntro A pseudo-boolean function (<http://en.wikipedia.org/wiki/Pseudo-Boolean_function>) is a function from booleans to reals, basically treating 'True' as @1@ and 'False' as @0@. They are typically expressed in polynomial form. Such functions can be used to express cardinality constraints, where we want to /count/ how many things satisfy a certain condition. One can code such constraints using regular SBV programming: Simply walk over the booleans and the corresponding coefficients, and assert the required relation. For instance: > [b0, b1, b2, b3] `pbAtMost` 2 is precisely equivalent to: > sum (map (\b -> ite b 1 0) [b0, b1, b2, b3]) .<= 2 and they both express that at most /two/ of @b0@, @b1@, @b2@, and @b3@ can be 'true'. However, the equivalent forms give rise to long formulas and the cardinality constraint can get lost in the translation. The idea here is that if you use these functions instead, SBV will produce better translations to SMTLib for more efficient solving of cardinality constraints, assuming the backend solver supports them. Currently, only Z3 supports pseudo-booleans directly. For all other solvers, SBV will translate these to equivalent terms that do not require special functions. -} {-# ANN module ("HLint: ignore Use import/export shortcut" :: String) #-}