Sparse.Matrix

Sparse Matrices in Morton order

Sparse Matrices

data Mat a

Constructors

Mat !Int !(Vector Word) !(Vector Word) !(Vector a)

Instances

(Functor f, Contravariant f, Vectored a) => Contains f (Mat a)

(Applicative f, Vectored a) => Ixed f (Mat a)

(Applicative f, Vectored a, ~ * a b) => Each f (Mat a) (Mat b) a b

(Vectored a, Eq (Vector a)) => Eq (Mat a)

(Vectored a, Eq0 a) => Num (Mat a)

(Vectored a, Ord (Vector a)) => Ord (Mat a)

(Vectored a, Read a) => Read (Mat a)

(Vectored a, Show a) => Show (Mat a)

NFData (Vector a) => NFData (Mat a)

Vectored a => Vectored (Mat a)

(Vectored a, Eq0 a) => Eq0 (Mat a)

Keys

data Key

Key i j logically orders the keys as if the bits of the keys i and j were interleaved. This is equivalent to storing the keys in "Morton Order".

>>> Key 100 200 ^. _1
100

>>> Key 100 200 ^. _2
200

Constructors

Key !Word !Word

Instances

Eq Key

Ord Key

Read Key

Show Key

Unbox Key

Vectored Key

Vector Vector Key

FunctorWithIndex Key Heap

FoldableWithIndex Key Heap

TraversableWithIndex Key Heap

MVector MVector Key

(~ * a Word, ~ * b Word) => Field1 Key Key a b

(~ * a Word, ~ * b Word) => Field2 Key Key a b

Construction

fromList :: Vectored a => [(Key, a)] -> Mat a

Build a sparse matrix.

singleton :: Vectored a => Key -> a -> Mat a

singleton makes a matrix with a singleton value at a given location

transpose :: Vectored a => Mat a -> Mat a

Transpose a matrix

ident :: (Vectored a, Num a) => Int -> Mat a

ident n makes an n x n identity matrix

>>> ident 4
fromList [(Key 0 0,1),(Key 1 1,1),(Key 2 2,1),(Key 3 3,1)]

empty :: Vectored a => Mat a

The empty matrix

>>> empty :: Mat Int
fromList []

Consumption

size :: Mat a -> Int

Count the number of non-zero entries in the matrix

>>> size (ident 4)
4

null :: Mat a -> Bool

>>> null (empty :: Mat Int)
True

Distinguishable Zero

class (Vectored a, Num a) => Eq0 a where

Methods

isZero :: a -> Bool

Return whether or not the element is 0.

It may be okay to never return True, but you won't be able to thin spurious zeroes introduced into your matrix.

nonZero :: (x -> y -> a) -> x -> y -> Maybe a

Remove results that are equal to zero from a simpler function.

When used with addWith or multiplyWith's additive argument this can help retain the sparsity of the matrix.

addMats :: Mat a -> Mat a -> Mat a

Add two matrices. By default this assumes isZero can possibly return True after an addition. For some ring-like structures, this doesn't hold. There you can use:

 addMats = addWith (+)

By default this will use

 addMats = addWith0 $ nonZero (+)

addHeap :: Maybe (Heap a) -> Stream (Key, a)

Convert from a Heap to a Stream.

If addition of non-zero valus in your ring-like structure cannot yield zero, then you can use

 addHeap = streamHeapWith (+)

instead of the default definition:

 addHeap = streamHeapWith0 $ nonZero (+)

Instances

Eq0 Double

Eq0 Float

Eq0 Int

Eq0 Integer

Eq0 Word

(RealFloat a, Eq0 a) => Eq0 (Complex a)

(Vectored a, Eq0 a) => Eq0 (Mat a)

Customization

addWith :: Vectored a => (a -> a -> a) -> Mat a -> Mat a -> Mat a

Merge two matrices where the indices coincide into a new matrix. This provides for generalized addition, but where the summation of two non-zero entries is necessarily non-zero.

multiplyWith :: Vectored a => (a -> a -> a) -> (Maybe (Heap a) -> Stream (Key, a)) -> Mat a -> Mat a -> Mat a

Multiply two matrices using the specified multiplication and addition operation.

Storage

class (Vector (Vec a) a, Monoid (Vec a a)) => Vectored a

Associated Types

type Vec a :: * -> *

Instances

Vectored Double

Vectored Float

Vectored Int

Vectored Int8

Vectored Int16

Vectored Int32

Vectored Int64

Vectored Integer

Vectored Word

Vectored Word8

Vectored Word16

Vectored Word32

Vectored Word64

Vectored ()

Vectored Key

(Vectored a, RealFloat a) => Vectored (Complex a)

Vectored a => Vectored (Mat a)

Lenses

_Mat :: Vectored a => Iso' (Mat a) (Vector Vector (Vec a) (Key, a))

bundle up the matrix in a form suitable for vector-algorithms

keys :: Lens' (Mat a) (Vector Key)

Access the keys of a matrix

values :: Lens (Mat a) (Mat b) (Vector a) (Vector b)

Access the keys of a matrix