Numeric.AD.Mode.Reverse

Reverse-mode automatic differentiation using Wengert lists and Data.Reflection

Documentation

data Reverse s a

Instances

(Reifies s Tape, Num a) => Jacobian (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Associated Types

type D (Reverse s a)

Methods

unary :: (Scalar (Reverse s a) -> Scalar (Reverse s a)) -> D (Reverse s a) -> Reverse s a -> Reverse s a

lift1 :: (Scalar (Reverse s a) -> Scalar (Reverse s a)) -> (D (Reverse s a) -> D (Reverse s a)) -> Reverse s a -> Reverse s a

lift1_ :: (Scalar (Reverse s a) -> Scalar (Reverse s a)) -> (D (Reverse s a) -> D (Reverse s a) -> D (Reverse s a)) -> Reverse s a -> Reverse s a

binary :: (Scalar (Reverse s a) -> Scalar (Reverse s a) -> Scalar (Reverse s a)) -> D (Reverse s a) -> D (Reverse s a) -> Reverse s a -> Reverse s a -> Reverse s a

lift2 :: (Scalar (Reverse s a) -> Scalar (Reverse s a) -> Scalar (Reverse s a)) -> (D (Reverse s a) -> D (Reverse s a) -> (D (Reverse s a), D (Reverse s a))) -> Reverse s a -> Reverse s a -> Reverse s a

lift2_ :: (Scalar (Reverse s a) -> Scalar (Reverse s a) -> Scalar (Reverse s a)) -> (D (Reverse s a) -> D (Reverse s a) -> D (Reverse s a) -> (D (Reverse s a), D (Reverse s a))) -> Reverse s a -> Reverse s a -> Reverse s a

(Reifies s Tape, Num a) => Mode (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Associated Types

type Scalar (Reverse s a)

Methods

isKnownConstant :: Reverse s a -> Bool

asKnownConstant :: Reverse s a -> Maybe (Scalar (Reverse s a))

isKnownZero :: Reverse s a -> Bool

auto :: Scalar (Reverse s a) -> Reverse s a

(*^) :: Scalar (Reverse s a) -> Reverse s a -> Reverse s a

(^*) :: Reverse s a -> Scalar (Reverse s a) -> Reverse s a

(^/) :: Reverse s a -> Scalar (Reverse s a) -> Reverse s a

zero :: Reverse s a

(Reifies s Tape, Num a, Bounded a) => Bounded (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

minBound :: Reverse s a

maxBound :: Reverse s a

(Reifies s Tape, Num a, Enum a) => Enum (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

succ :: Reverse s a -> Reverse s a

pred :: Reverse s a -> Reverse s a

toEnum :: Int -> Reverse s a

fromEnum :: Reverse s a -> Int

enumFrom :: Reverse s a -> [Reverse s a]

enumFromThen :: Reverse s a -> Reverse s a -> [Reverse s a]

enumFromTo :: Reverse s a -> Reverse s a -> [Reverse s a]

enumFromThenTo :: Reverse s a -> Reverse s a -> Reverse s a -> [Reverse s a]

(Reifies s Tape, Floating a) => Floating (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

pi :: Reverse s a

exp :: Reverse s a -> Reverse s a

log :: Reverse s a -> Reverse s a

sqrt :: Reverse s a -> Reverse s a

(**) :: Reverse s a -> Reverse s a -> Reverse s a

logBase :: Reverse s a -> Reverse s a -> Reverse s a

sin :: Reverse s a -> Reverse s a

cos :: Reverse s a -> Reverse s a

tan :: Reverse s a -> Reverse s a

asin :: Reverse s a -> Reverse s a

acos :: Reverse s a -> Reverse s a

atan :: Reverse s a -> Reverse s a

sinh :: Reverse s a -> Reverse s a

cosh :: Reverse s a -> Reverse s a

tanh :: Reverse s a -> Reverse s a

asinh :: Reverse s a -> Reverse s a

acosh :: Reverse s a -> Reverse s a

atanh :: Reverse s a -> Reverse s a

log1p :: Reverse s a -> Reverse s a

expm1 :: Reverse s a -> Reverse s a

log1pexp :: Reverse s a -> Reverse s a

log1mexp :: Reverse s a -> Reverse s a

(Reifies s Tape, RealFloat a) => RealFloat (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

floatRadix :: Reverse s a -> Integer

floatDigits :: Reverse s a -> Int

floatRange :: Reverse s a -> (Int, Int)

decodeFloat :: Reverse s a -> (Integer, Int)

encodeFloat :: Integer -> Int -> Reverse s a

exponent :: Reverse s a -> Int

significand :: Reverse s a -> Reverse s a

scaleFloat :: Int -> Reverse s a -> Reverse s a

isNaN :: Reverse s a -> Bool

isInfinite :: Reverse s a -> Bool

isDenormalized :: Reverse s a -> Bool

isNegativeZero :: Reverse s a -> Bool

isIEEE :: Reverse s a -> Bool

atan2 :: Reverse s a -> Reverse s a -> Reverse s a

(Reifies s Tape, Num a) => Num (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

(+) :: Reverse s a -> Reverse s a -> Reverse s a

(-) :: Reverse s a -> Reverse s a -> Reverse s a

(*) :: Reverse s a -> Reverse s a -> Reverse s a

negate :: Reverse s a -> Reverse s a

abs :: Reverse s a -> Reverse s a

signum :: Reverse s a -> Reverse s a

fromInteger :: Integer -> Reverse s a

(Reifies s Tape, Fractional a) => Fractional (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

(/) :: Reverse s a -> Reverse s a -> Reverse s a

recip :: Reverse s a -> Reverse s a

fromRational :: Rational -> Reverse s a

(Reifies s Tape, Real a) => Real (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

toRational :: Reverse s a -> Rational

(Reifies s Tape, RealFrac a) => RealFrac (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

properFraction :: Integral b => Reverse s a -> (b, Reverse s a)

truncate :: Integral b => Reverse s a -> b

round :: Integral b => Reverse s a -> b

ceiling :: Integral b => Reverse s a -> b

floor :: Integral b => Reverse s a -> b

Show a => Show (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

showsPrec :: Int -> Reverse s a -> ShowS

show :: Reverse s a -> String

showList :: [Reverse s a] -> ShowS

(Reifies s Tape, Erf a) => Erf (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

erf :: Reverse s a -> Reverse s a

erfc :: Reverse s a -> Reverse s a

erfcx :: Reverse s a -> Reverse s a

normcdf :: Reverse s a -> Reverse s a

(Reifies s Tape, InvErf a) => InvErf (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

inverf :: Reverse s a -> Reverse s a

inverfc :: Reverse s a -> Reverse s a

invnormcdf :: Reverse s a -> Reverse s a

(Reifies s Tape, Num a, Eq a) => Eq (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

(==) :: Reverse s a -> Reverse s a -> Bool

(/=) :: Reverse s a -> Reverse s a -> Bool

(Reifies s Tape, Num a, Ord a) => Ord (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

Methods

compare :: Reverse s a -> Reverse s a -> Ordering

(<) :: Reverse s a -> Reverse s a -> Bool

(<=) :: Reverse s a -> Reverse s a -> Bool

(>) :: Reverse s a -> Reverse s a -> Bool

(>=) :: Reverse s a -> Reverse s a -> Bool

max :: Reverse s a -> Reverse s a -> Reverse s a

min :: Reverse s a -> Reverse s a -> Reverse s a

type D (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

type D (Reverse s a) = Id a

type Scalar (Reverse s a)

Defined in Numeric.AD.Internal.Reverse

type Scalar (Reverse s a) = a

auto :: Mode t => Scalar t -> t

Embed a constant

Gradient

grad :: (Traversable f, Num a) => (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> Reverse s a) -> f a -> f a

The grad function calculates the gradient of a non-scalar-to-scalar function with reverse-mode AD in a single pass.

>>> grad (\[x,y,z] -> x*y+z) [1,2,3]
[2,1,1]

>>> grad (\[x,y] -> x**y) [0,2]
[0.0,NaN]

grad' :: (Traversable f, Num a) => (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> Reverse s a) -> f a -> (a, f a)

The grad' function calculates the result and gradient of a non-scalar-to-scalar function with reverse-mode AD in a single pass.

>>> grad' (\[x,y,z] -> x*y+z) [1,2,3]
(5,[2,1,1])

gradWith :: (Traversable f, Num a) => (a -> a -> b) -> (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> Reverse s a) -> f a -> f b

grad g f function calculates the gradient of a non-scalar-to-scalar function f with reverse-mode AD in a single pass. The gradient is combined element-wise with the argument using the function g.

grad == gradWith (_ dx -> dx)
id == gradWith const

gradWith' :: (Traversable f, Num a) => (a -> a -> b) -> (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> Reverse s a) -> f a -> (a, f b)

grad' g f calculates the result and gradient of a non-scalar-to-scalar function f with reverse-mode AD in a single pass the gradient is combined element-wise with the argument using the function g.

grad' == gradWith' (_ dx -> dx)

Jacobian

jacobian :: (Traversable f, Functor g, Num a) => (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> g (Reverse s a)) -> f a -> g (f a)

The jacobian function calculates the jacobian of a non-scalar-to-non-scalar function with reverse AD lazily in m passes for m outputs.

>>> jacobian (\[x,y] -> [y,x,x*y]) [2,1]
[[0,1],[1,0],[1,2]]

jacobian' :: (Traversable f, Functor g, Num a) => (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> g (Reverse s a)) -> f a -> g (a, f a)

The jacobian' function calculates both the result and the Jacobian of a nonscalar-to-nonscalar function, using m invocations of reverse AD, where m is the output dimensionality. Applying fmap snd to the result will recover the result of jacobian | An alias for gradF'

>>> jacobian' (\[x,y] -> [y,x,x*y]) [2,1]
[(1,[0,1]),(2,[1,0]),(2,[1,2])]

jacobianWith :: (Traversable f, Functor g, Num a) => (a -> a -> b) -> (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> g (Reverse s a)) -> f a -> g (f b)

'jacobianWith g f' calculates the Jacobian of a non-scalar-to-non-scalar function f with reverse AD lazily in m passes for m outputs.

Instead of returning the Jacobian matrix, the elements of the matrix are combined with the input using the g.

jacobian == jacobianWith (_ dx -> dx)
jacobianWith const == (f x -> const x <$> f x)

jacobianWith' :: (Traversable f, Functor g, Num a) => (a -> a -> b) -> (forall s. (Reifies s Tape, Typeable s) => f (Reverse s a) -> g (Reverse s a)) -> f a -> g (a, f b)

jacobianWith g f' calculates both the result and the Jacobian of a nonscalar-to-nonscalar function f, using m invocations of reverse AD, where m is the output dimensionality. Applying fmap snd to the result will recover the result of jacobianWith

Instead of returning the Jacobian matrix, the elements of the matrix are combined with the input using the g.

jacobian' == jacobianWith' (_ dx -> dx)

Hessian

hessian :: (Traversable f, Num a) => (forall s s'. (Reifies s Tape, Typeable s, Reifies s' Tape, Typeable s') => f (On (Reverse s (Reverse s' a))) -> On (Reverse s (Reverse s' a))) -> f a -> f (f a)

Compute the hessian via the jacobian of the gradient. gradient is computed in reverse mode and then the jacobian is computed in reverse mode.

However, since the grad f :: f a -> f a is square this is not as fast as using the forward-mode Jacobian of a reverse mode gradient provided by hessian.

>>> hessian (\[x,y] -> x*y) [1,2]
[[0,1],[1,0]]

hessianF :: (Traversable f, Functor g, Num a) => (forall s s'. (Reifies s Tape, Typeable s, Reifies s' Tape, Typeable s') => f (On (Reverse s (Reverse s' a))) -> g (On (Reverse s (Reverse s' a)))) -> f a -> g (f (f a))

Compute the order 3 Hessian tensor on a non-scalar-to-non-scalar function via the reverse-mode Jacobian of the reverse-mode Jacobian of the function.

Less efficient than hessianF.

>>> hessianF (\[x,y] -> [x*y,x+y,exp x*cos y]) [1,2 :: RDouble]
[[[0.0,1.0],[1.0,0.0]],[[0.0,0.0],[0.0,0.0]],[[-1.131204383757,-2.471726672005],[-2.471726672005,1.131204383757]]]

Derivatives

diff :: Num a => (forall s. (Reifies s Tape, Typeable s) => Reverse s a -> Reverse s a) -> a -> a

Compute the derivative of a function.

>>> diff sin 0
1.0

diff' :: Num a => (forall s. (Reifies s Tape, Typeable s) => Reverse s a -> Reverse s a) -> a -> (a, a)

The diff' function calculates the result and derivative, as a pair, of a scalar-to-scalar function.

>>> diff' sin 0
(0.0,1.0)

>>> diff' exp 0
(1.0,1.0)

diffF :: (Functor f, Num a) => (forall s. (Reifies s Tape, Typeable s) => Reverse s a -> f (Reverse s a)) -> a -> f a

Compute the derivatives of each result of a scalar-to-vector function with regards to its input.

>>> diffF (\a -> [sin a, cos a]) 0
[1.0,0.0]

diffF' :: (Functor f, Num a) => (forall s. (Reifies s Tape, Typeable s) => Reverse s a -> f (Reverse s a)) -> a -> f (a, a)

Compute the derivatives of each result of a scalar-to-vector function with regards to its input along with the answer.

>>> diffF' (\a -> [sin a, cos a]) 0
[(0.0,1.0),(1.0,0.0)]