Copyright | (c) Michal Konecny Pieter Collins |
---|---|
License | BSD3 |
Maintainer | mikkonecny@gmail.com |
Stability | experimental |
Portability | portable |
Safe Haskell | None |
Language | Haskell2010 |
MixedTypesNumPrelude
provides a version of Prelude
where
unary and binary operations such as not
, +
, ==
have their result type derived from the parameter type(s).
This module facilitates a single-line import for the package mixed-types-num. See the re-exported modules for further details.
Synopsis
- module Numeric.MixedTypes.PreludeHiding
- module Data.Convertible.Base
- module Numeric.MixedTypes.Literals
- module Numeric.MixedTypes.Bool
- module Numeric.MixedTypes.Eq
- module Numeric.MixedTypes.Ord
- module Numeric.MixedTypes.MinMaxAbs
- module Numeric.MixedTypes.AddSub
- module Numeric.MixedTypes.Round
- module Numeric.MixedTypes.Ring
- module Numeric.MixedTypes.Field
- module Numeric.MixedTypes.Elementary
- module Numeric.MixedTypes.Complex
- module Numeric.CollectErrors
- module Utils.TH.DeclForTypes
- module Utils.Test.EnforceRange
- (%) :: Integral a => a -> a -> Ratio a
Feature highlights
Basics
To replicate the below in ghci using stack, start it as follows:
> stack ghci mixed-types-num:lib ...> :add MixedTypesNumPrelude
Literals have a fixed type
...> :t 1 ... Integer
...> :t 1.0 ... Rational
...> 1 :: Rational ... Couldn't match type ‘Integer’ with ‘GHC.Real.Ratio Integer’ ...
Mixed-type operations
...> :t 1.5 + 1 ... :: Rational
...> :t 1.5 * (length [[]]) ... :: Rational
Dividing integers, dealing with potential error
...> :t let n = 1 in n/(n+1) ... :: CollectErrors [(ErrorCertaintyLevel, NumError)] Rational
A shorter synonym of this type is CN Rational
.
We use the shorter form below for better readability of this documentation
although ghci usually prints the longer version:
...> :t let n = 1 in n/(n+1) ... :: CN Rational
The CN
wrapper here indicates that integer division can fail for some values:
...> 1/0 {[(ERROR,division by 0)]}
Note that when evaluating 1/0
, it evaluates to the error value printed above.
This is not an exception, but a special value.
When one is certain the division is well defined, one can remove CN
as follows:
...> :t (1/!2) ... :: Rational
Note that if one gets it wrong, it can lead to an exception:
...> :t (1/!0) *** Exception: Ratio has zero denominator
More generally, one can remove CN
as follows:
...> :t (~!) (1/2) ... :: Rational
The operator (/!)
stands for division which throws an exception is the
denominator is 0. It "propagates" any potential errors
from the sub-expressions. For example:
...> :t 1/!(1 - 1/n) ... :: CN Rational
The above expression will throw an error exception when evaluated with n=1
but when n=0
, it will not throw an excetion but return an error value.
The (~!)
operator removes CN from any type, throwing an exception if some errors have certainly occurred:
...> :t (~!) (1/(1 - 1/n)) ... :: Rational
The following examples require also package aern2-real. To get access to this via stack, you can start ghci eg as follows:
stack ghci aern2-real:lib ...> :add AERN2.Real
Also other harmless potential errors can be ignored using (~!)
:
...> (~!) $ sqrt (pi-pi) [7.395570986446986e-32 ± <2^(-103)]
...> sqrt (pi-pi) [7.395570986446986e-32 ± <2^(-103)]{[(POTENTIAL ERROR,out of range: sqrt: argument must be >= 0: [0 ± <2^(-240)])]}
Natural, integer and fractional powers
...> :t 2^2 ...CN Integer
...> :t 2.0^(-2) ...CN Rational
...> :t (double 2)^(1/!2) ...Double
The following examples require package aern2-real:
...> :t 2^(1/2) ...CauchyRealCN
...> :t pi ...CauchyReal
...> :t sqrt 2 ...CauchyRealCN
Comparing an integer with an (exact) real number
...> let abs2 x = if x < 0 then -x else x in (abs2 (pi - pi)) ? (bitsS 100) [0 ± <2^(-103)]{[(POTENTIAL ERROR,numeric error: union of enclosures: not enclosing the same value)]}
The potential error means that both branches were executed in parallel because the condition could not be decided, and it was moreover impossible to guarantee (in general) that both branches will return the same number. If we make a mistake, this error may appear with certainty, eg:
...> let abs2 x = if x < 0 then 1-x else x in (abs2 (pi - pi)) ? (bitsS 100) *** Exception: WithGlobalParam ensureNoCE: [(ERROR,numeric error: union of enclosures: not enclosing the same value)]
If we are certain such errors will never appear, we can silence the potential error warnings:
...> let abs2 x = (~!) (if x < 0 then -x else x) in (abs2 (pi - pi)) ? (bitsS 100) [0 ± <2^(-103)]
In these examples, if
is overloaded so that it works for conditions
of other types than Bool
. Here the condition has the type Sequence (Maybe Bool)
.
The whole expression is the sequence of balls in which those balls for which the condition
is inconclusive are the union of the balls computed by both branches.
Type classes
Arithmetic operations are provided via multi-parameter type classes and the result type is given by associated type families. For example:
(+) :: (CanAddAsymmetric t1 t2) => t1 -> t2 -> AddType t1 t2
The type constraint CanAdd t1 t2
implies both
CanAddAsymmetric t1 t2
and CanAddAsymmetric t2 t1
.
For convenience there are other aggregate type constraints such as
CanAddThis t1 t2
, which implies that the result is of type t1
,
and CanAddSameType t
, which is a shortcut for CanAddThis t t
.
Notably, there are convenience classes Ring
and Field
.
For types that instantiate Prelude classes such as Num
, one can
define instances of the new classes using the default implementation, eg:
newtype II = II Integer deriving (Eq, Ord, Num) -- assuming -XGeneralizedNewtypeDeriving instance CanAddAsymmetric II II
Conversely, if one defines instances for classes such as CanAddAsymmetric
,
one can then trivially define also instances of Num
etc:
instance Prelude.Num T where (+) = (+) ...
Testable specifications
The arithmetic type classes are accompanied by generic hspec test suites, which are specialised to concrete instance types for their testing. These test suites include the expected algebraic properties of operations, such as commutativity and associativity of addition.
Limitations
- Not all numerical operations are supported yet.
Eg
tan
,atan
are missing at the moment. - Not all Prelude numerical types are supported yet.
Eg
Natural
andFloat
are not supported at present, butDouble
is supported. - Many common operations such as
fromEnum
,threadDelay
give or require anInt
value, which means we sometimes need to convert:
threadDelay (int 1000000) integer (fromEnum True)
Prelude functions such as take
, !!
and length
that use Int
in Prelude
are shadowed in MixedTypesNumPrelude
with more compatible/flexible versions.
Beware that Data.List.length
clashes with length
in MixedTypesNumPrelude
.
- Inferred types can be very large. Eg for
f a b c = sqrt (a + b * c + 1)
the inferred type is:
f: (CanMulAsymmetric t1 t2, CanAddAsymmetric t4 (MulType t1 t2), CanAddAsymmetric (AddType t4 (MulType t1 t2)) Integer, CanSqrt (AddType (AddType t4 (MulType t1 t2)) Integer)) => t4 -> t1 -> t2 -> SqrtType (AddType (AddType t4 (MulType t1 t2)) Integer)
- Due to limitations of some versions of ghc, type inferrence sometimes fails.
Eg
add1 = (+ 1)
fails (eg with ghc 8.0.2) unless we explicitly declare the typeadd1 :: (CanAdd Integer t) => t -> AddType t Integer
or use an explicit parameter, egadd1 x = x + 1
.
Origin
The idea of having numeric expressions in Haskell with types derived bottom-up was initially suggested and implemented by Pieter Collins. This version is a fresh rewrite by Michal Konečný.
Re-exporting Prelude, hiding the operators we are changing
A part of package `convertible'
module Data.Convertible.Base
Modules with Prelude alternatives
module Numeric.MixedTypes.Literals
module Numeric.MixedTypes.Bool
module Numeric.MixedTypes.Eq
module Numeric.MixedTypes.Ord
module Numeric.MixedTypes.MinMaxAbs
module Numeric.MixedTypes.AddSub
module Numeric.MixedTypes.Round
module Numeric.MixedTypes.Ring
module Numeric.MixedTypes.Field
module Numeric.MixedTypes.Complex
module Numeric.CollectErrors
module Utils.TH.DeclForTypes
module Utils.Test.EnforceRange