hmatrix-0.19.0.0: Numeric Linear Algebra

Copyright(c) Alberto Ruiz 2006-15
LicenseBSD3
MaintainerAlberto Ruiz
Stabilityprovisional
Safe HaskellNone
LanguageHaskell98

Numeric.LinearAlgebra

Contents

Description

 

Synopsis

Basic types and data manipulation

This package works with 2D (Matrix) and 1D (Vector) arrays of real (R) or complex (C) double precision numbers. Single precision and machine integers are also supported for basic arithmetic and data manipulation.

Numeric classes

The standard numeric classes are defined elementwise (commonly referred to as the Hadamard product or the Schur product):

>>> vector [1,2,3] * vector [3,0,-2]
fromList [3.0,0.0,-6.0]
>>> matrix 3 [1..9] * ident 3
(3><3)
 [ 1.0, 0.0, 0.0
 , 0.0, 5.0, 0.0
 , 0.0, 0.0, 9.0 ]

Autoconformable dimensions

In most operations, single-element vectors and matrices (created from numeric literals or using scalar), and matrices with just one row or column, automatically expand to match the dimensions of the other operand:

>>> 5 + 2*ident 3 :: Matrix Double
(3><3)
 [ 7.0, 5.0, 5.0
 , 5.0, 7.0, 5.0
 , 5.0, 5.0, 7.0 ]
>>> (4><3) [1..] + row [10,20,30]
(4><3)
 [ 11.0, 22.0, 33.0
 , 14.0, 25.0, 36.0
 , 17.0, 28.0, 39.0
 , 20.0, 31.0, 42.0 ]

Products

Dot

dot :: Numeric t => Vector t -> Vector t -> t Source #

(<.>) :: Numeric t => Vector t -> Vector t -> t infixr 8 Source #

An infix synonym for dot

>>> vector [1,2,3,4] <.> vector [-2,0,1,1]
5.0
>>> let 𝑖 = 0:+1 :: C
>>> fromList [1+𝑖,1] <.> fromList [1,1+𝑖]
2.0 :+ 0.0

Matrix-vector

(#>) :: Numeric t => Matrix t -> Vector t -> Vector t infixr 8 Source #

dense matrix-vector product

>>> let m = (2><3) [1..]
>>> m
(2><3)
 [ 1.0, 2.0, 3.0
 , 4.0, 5.0, 6.0 ]
>>> let v = vector [10,20,30]
>>> m #> v
fromList [140.0,320.0]

(<#) :: Numeric t => Vector t -> Matrix t -> Vector t infixl 8 Source #

dense vector-matrix product

(!#>) :: GMatrix -> Vector Double -> Vector Double infixr 8 Source #

general matrix - vector product

>>> let m = mkSparse [((0,999),1.0),((1,1999),2.0)]
>>> m !#> vector [1..2000]
fromList [1000.0,4000.0]

Matrix-matrix

(<>) :: Numeric t => Matrix t -> Matrix t -> Matrix t infixr 8 Source #

dense matrix product

>>> let a = (3><5) [1..]
>>> a
(3><5)
 [  1.0,  2.0,  3.0,  4.0,  5.0
 ,  6.0,  7.0,  8.0,  9.0, 10.0
 , 11.0, 12.0, 13.0, 14.0, 15.0 ]
>>> let b = (5><2) [1,3, 0,2, -1,5, 7,7, 6,0]
>>> b
(5><2)
 [  1.0, 3.0
 ,  0.0, 2.0
 , -1.0, 5.0
 ,  7.0, 7.0
 ,  6.0, 0.0 ]
>>> a <> b
(3><2)
 [  56.0,  50.0
 , 121.0, 135.0
 , 186.0, 220.0 ]

The matrix product is also implemented in the Data.Monoid instance, where single-element matrices (created from numeric literals or using scalar) are used for scaling.

>>> import Data.Monoid as M
>>> let m = matrix 3 [1..6]
>>> m M.<> 2 M.<> diagl[0.5,1,0]
(2><3)
 [ 1.0,  4.0, 0.0
 , 4.0, 10.0, 0.0 ]

mconcat uses optimiseMult to get the optimal association order.

Other

outer :: Product t => Vector t -> Vector t -> Matrix t Source #

Outer product of two vectors.

>>> fromList [1,2,3] `outer` fromList [5,2,3]
(3><3)
 [  5.0, 2.0, 3.0
 , 10.0, 4.0, 6.0
 , 15.0, 6.0, 9.0 ]

kronecker :: Product t => Matrix t -> Matrix t -> Matrix t Source #

Kronecker product of two matrices.

m1=(2><3)
 [ 1.0,  2.0, 0.0
 , 0.0, -1.0, 3.0 ]
m2=(4><3)
 [  1.0,  2.0,  3.0
 ,  4.0,  5.0,  6.0
 ,  7.0,  8.0,  9.0
 , 10.0, 11.0, 12.0 ]
>>> kronecker m1 m2
(8><9)
 [  1.0,  2.0,  3.0,   2.0,   4.0,   6.0,  0.0,  0.0,  0.0
 ,  4.0,  5.0,  6.0,   8.0,  10.0,  12.0,  0.0,  0.0,  0.0
 ,  7.0,  8.0,  9.0,  14.0,  16.0,  18.0,  0.0,  0.0,  0.0
 , 10.0, 11.0, 12.0,  20.0,  22.0,  24.0,  0.0,  0.0,  0.0
 ,  0.0,  0.0,  0.0,  -1.0,  -2.0,  -3.0,  3.0,  6.0,  9.0
 ,  0.0,  0.0,  0.0,  -4.0,  -5.0,  -6.0, 12.0, 15.0, 18.0
 ,  0.0,  0.0,  0.0,  -7.0,  -8.0,  -9.0, 21.0, 24.0, 27.0
 ,  0.0,  0.0,  0.0, -10.0, -11.0, -12.0, 30.0, 33.0, 36.0 ]

cross :: Product t => Vector t -> Vector t -> Vector t Source #

cross product (for three-element vectors)

scale :: Linear t c => t -> c t -> c t Source #

add :: Additive c => c -> c -> c Source #

sumElements :: Container c e => c e -> e Source #

the sum of elements

prodElements :: Container c e => c e -> e Source #

the product of elements

Linear systems

General

(<\>) :: (LSDiv c, Field t) => Matrix t -> c t -> c t infixl 7 Source #

Least squares solution of a linear system, similar to the \ operator of Matlab/Octave (based on linearSolveSVD)

a = (3><2)
 [ 1.0,  2.0
 , 2.0,  4.0
 , 2.0, -1.0 ]
v = vector [13.0,27.0,1.0]
>>> let x = a <\> v
>>> x
fromList [3.0799999999999996,5.159999999999999]
>>> a #> x
fromList [13.399999999999999,26.799999999999997,1.0]

It also admits multiple right-hand sides stored as columns in a matrix.

linearSolveLS :: Field t => Matrix t -> Matrix t -> Matrix t Source #

Least squared error solution of an overconstrained linear system, or the minimum norm solution of an underconstrained system. For rank-deficient systems use linearSolveSVD.

linearSolveSVD :: Field t => Matrix t -> Matrix t -> Matrix t Source #

Minimum norm solution of a general linear least squares problem Ax=B using the SVD. Admits rank-deficient systems but it is slower than linearSolveLS. The effective rank of A is determined by treating as zero those singular valures which are less than eps times the largest singular value.

Determined

linearSolve :: Field t => Matrix t -> Matrix t -> Maybe (Matrix t) Source #

Solve a linear system (for square coefficient matrix and several right-hand sides) using the LU decomposition, returning Nothing for a singular system. For underconstrained or overconstrained systems use linearSolveLS or linearSolveSVD.

a = (2><2)
 [ 1.0, 2.0
 , 3.0, 5.0 ]
b = (2><3)
 [  6.0, 1.0, 10.0
 , 15.0, 3.0, 26.0 ]
>>> linearSolve a b
Just (2><3)
 [ -1.4802973661668753e-15,     0.9999999999999997, 1.999999999999997
 ,       3.000000000000001, 1.6653345369377348e-16, 4.000000000000002 ]
>>> let Just x = it
>>> disp 5 x
2x3
-0.00000  1.00000  2.00000
 3.00000  0.00000  4.00000
>>> a <> x
(2><3)
 [  6.0, 1.0, 10.0
 , 15.0, 3.0, 26.0 ]

luSolve :: Field t => LU t -> Matrix t -> Matrix t Source #

Solution of a linear system (for several right hand sides) from the precomputed LU factorization obtained by luPacked.

luPacked :: Field t => Matrix t -> LU t Source #

Obtains the LU decomposition of a matrix in a compact data structure suitable for luSolve.

luSolve' :: (Fractional t, Container Vector t) => LU t -> Matrix t -> Matrix t Source #

Experimental implementation of luSolve for any Fractional element type, including Mod n I and Mod n Z.

>>> let a = (2><2) [1,2,3,5] :: Matrix (Z ./. 13)
(2><2)
 [ 1, 2
 , 3, 5 ]
>>> b
(2><3)
 [ 5, 1, 3
 , 8, 6, 3 ]
>>> luSolve' (luPacked' a) b
(2><3)
 [ 4,  7, 4
 , 7, 10, 6 ]

luPacked' :: (Num (Vector t), Normed (Vector t), Container Vector t, Fractional t) => Matrix t -> LU t Source #

Experimental implementation of luPacked for any Fractional element type, including Mod n I and Mod n Z.

>>> let m = ident 5 + (5><5) [0..] :: Matrix (Z ./. 17)
(5><5)
 [  1,  1,  2,  3,  4
 ,  5,  7,  7,  8,  9
 , 10, 11, 13, 13, 14
 , 15, 16,  0,  2,  2
 ,  3,  4,  5,  6,  8 ]
>>> let (l,u,p,s) = luFact $ luPacked' m
>>> l
(5><5)
 [  1,  0, 0,  0, 0
 ,  6,  1, 0,  0, 0
 , 12,  7, 1,  0, 0
 ,  7, 10, 7,  1, 0
 ,  8,  2, 6, 11, 1 ]
>>> u
(5><5)
 [ 15, 16,  0,  2,  2
 ,  0, 13,  7, 13, 14
 ,  0,  0, 15,  0, 11
 ,  0,  0,  0, 15, 15
 ,  0,  0,  0,  0,  1 ]

Symmetric indefinite

ldlSolve :: Field t => LDL t -> Matrix t -> Matrix t Source #

Solution of a linear system (for several right hand sides) from a precomputed LDL factorization obtained by ldlPacked.

Note: this can be slower than the general solver based on the LU decomposition.

ldlPacked :: Field t => Herm t -> LDL t Source #

Obtains the LDL decomposition of a matrix in a compact data structure suitable for ldlSolve.

Positive definite

cholSolve Source #

Arguments

:: Field t 
=> Matrix t

Cholesky decomposition of the coefficient matrix

-> Matrix t

right hand sides

-> Matrix t

solution

Solve a symmetric or Hermitian positive definite linear system using a precomputed Cholesky decomposition obtained by chol.

Triangular

data UpLo Source #

Constructors

Lower 
Upper 

triSolve Source #

Arguments

:: Field t 
=> UpLo

Lower or Upper

-> Matrix t

coefficient matrix

-> Matrix t

right hand sides

-> Matrix t

solution

Solve a triangular linear system. If Upper is specified then all elements below the diagonal are ignored; if Lower is specified then all elements above the diagonal are ignored.

Tridiagonal

triDiagSolve Source #

Arguments

:: Field t 
=> Vector t

lower diagonal: \(n - 1\) elements

-> Vector t

diagonal: \(n\) elements

-> Vector t

upper diagonal: \(n - 1\) elements

-> Matrix t

right hand sides

-> Matrix t

solution

Solve a tridiagonal linear system. Suppose you wish to solve \(Ax = b\) where

\[ A = \begin{bmatrix} 1.0 & 4.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 \\ 3.0 & 1.0 & 4.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 \\ 0.0 & 3.0 & 1.0 & 4.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 \\ 0.0 & 0.0 & 3.0 & 1.0 & 4.0 & 0.0 & 0.0 & 0.0 & 0.0 \\ 0.0 & 0.0 & 0.0 & 3.0 & 1.0 & 4.0 & 0.0 & 0.0 & 0.0 \\ 0.0 & 0.0 & 0.0 & 0.0 & 3.0 & 1.0 & 4.0 & 0.0 & 0.0 \\ 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 3.0 & 1.0 & 4.0 & 0.0 \\ 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 3.0 & 1.0 & 4.0 \\ 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 3.0 & 1.0 \end{bmatrix} \quad b = \begin{bmatrix} 1.0 & 1.0 & 1.0 \\ 1.0 & -1.0 & 2.0 \\ 1.0 & 1.0 & 3.0 \\ 1.0 & -1.0 & 4.0 \\ 1.0 & 1.0 & 5.0 \\ 1.0 & -1.0 & 6.0 \\ 1.0 & 1.0 & 7.0 \\ 1.0 & -1.0 & 8.0 \\ 1.0 & 1.0 & 9.0 \end{bmatrix} \]

then

dL =  fromList [3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0, 3.0]
d  =  fromList [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
dU =  fromList [4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0, 4.0]

b = (9><3)
    [
      1.0,   1.0,   1.0,
      1.0,  -1.0,   2.0,
      1.0,   1.0,   3.0,
      1.0,  -1.0,   4.0,
      1.0,   1.0,   5.0,
      1.0,  -1.0,   6.0,
      1.0,   1.0,   7.0,
      1.0,  -1.0,   8.0,
      1.0,   1.0,   9.0
    ]

x = triDiagSolve dL d dU b

Sparse

cgSolve Source #

Arguments

:: Bool

is symmetric

-> GMatrix

coefficient matrix

-> Vector R

right-hand side

-> Vector R

solution

Solve a sparse linear system using the conjugate gradient method with default parameters.

cgSolve' Source #

Arguments

:: Bool

symmetric

-> R

relative tolerance for the residual (e.g. 1E-4)

-> R

relative tolerance for δx (e.g. 1E-3)

-> Int

maximum number of iterations

-> GMatrix

coefficient matrix

-> Vector R

initial solution

-> Vector R

right-hand side

-> [CGState]

solution

Solve a sparse linear system using the conjugate gradient method with default parameters.

Inverse and pseudoinverse

inv :: Field t => Matrix t -> Matrix t Source #

Inverse of a square matrix. See also invlndet.

pinv :: Field t => Matrix t -> Matrix t Source #

Pseudoinverse of a general matrix with default tolerance (pinvTol 1, similar to GNU-Octave).

pinvTol :: Field t => Double -> Matrix t -> Matrix t Source #

pinvTol r computes the pseudoinverse of a matrix with tolerance tol=r*g*eps*(max rows cols), where g is the greatest singular value.

m = (3><3) [ 1, 0,    0
           , 0, 1,    0
           , 0, 0, 1e-10] :: Matrix Double
>>> pinv m
1. 0.           0.
0. 1.           0.
0. 0. 10000000000.
>>> pinvTol 1E8 m
1. 0. 0.
0. 1. 0.
0. 0. 1.

Determinant and rank

rcond :: Field t => Matrix t -> Double Source #

Reciprocal of the 2-norm condition number of a matrix, computed from the singular values.

rank :: Field t => Matrix t -> Int Source #

Number of linearly independent rows or columns. See also ranksv

det :: Field t => Matrix t -> t Source #

Determinant of a square matrix. To avoid possible overflow or underflow use invlndet.

invlndet Source #

Arguments

:: Field t 
=> Matrix t 
-> (Matrix t, (t, t))

(inverse, (log abs det, sign or phase of det))

Joint computation of inverse and logarithm of determinant of a square matrix.

Norms

class Normed a where Source #

p-norm for vectors, operator norm for matrices

Minimal complete definition

norm_0, norm_1, norm_2, norm_Inf

Methods

norm_0 :: a -> R Source #

norm_1 :: a -> R Source #

norm_2 :: a -> R Source #

norm_Inf :: a -> R Source #

Instances

Normed (Vector Float) Source # 
Normed (Vector (Complex Float)) Source # 
Normed (Vector C) Source # 
Normed (Vector R) Source # 
Normed (Vector Z) Source # 
Normed (Vector I) Source # 
KnownNat m => Normed (Vector (Mod m Z)) Source # 

Methods

norm_0 :: Vector (Mod m Z) -> R Source #

norm_1 :: Vector (Mod m Z) -> R Source #

norm_2 :: Vector (Mod m Z) -> R Source #

norm_Inf :: Vector (Mod m Z) -> R Source #

KnownNat m => Normed (Vector (Mod m I)) Source # 

Methods

norm_0 :: Vector (Mod m I) -> R Source #

norm_1 :: Vector (Mod m I) -> R Source #

norm_2 :: Vector (Mod m I) -> R Source #

norm_Inf :: Vector (Mod m I) -> R Source #

Normed (Matrix C) Source # 
Normed (Matrix R) Source # 

norm_Frob :: (Normed (Vector t), Element t) => Matrix t -> R Source #

Frobenius norm (Schatten p-norm with p=2)

norm_nuclear :: Field t => Matrix t -> R Source #

Sum of singular values (Schatten p-norm with p=1)

Nullspace and range

orth :: Field t => Matrix t -> Matrix t Source #

return an orthonormal basis of the range space of a matrix. See also orthSVD.

nullspace :: Field t => Matrix t -> Matrix t Source #

return an orthonormal basis of the null space of a matrix. See also nullspaceSVD.

null1 :: Matrix R -> Vector R Source #

solution of overconstrained homogeneous linear system

null1sym :: Herm R -> Vector R Source #

solution of overconstrained homogeneous symmetric linear system

Singular value decomposition

svd :: Field t => Matrix t -> (Matrix t, Vector Double, Matrix t) Source #

Full singular value decomposition.

a = (5><3)
 [  1.0,  2.0,  3.0
 ,  4.0,  5.0,  6.0
 ,  7.0,  8.0,  9.0
 , 10.0, 11.0, 12.0
 , 13.0, 14.0, 15.0 ] :: Matrix Double
>>> let (u,s,v) = svd a
>>> disp 3 u
5x5
-0.101   0.768   0.614   0.028  -0.149
-0.249   0.488  -0.503   0.172   0.646
-0.396   0.208  -0.405  -0.660  -0.449
-0.543  -0.072  -0.140   0.693  -0.447
-0.690  -0.352   0.433  -0.233   0.398
>>> s
fromList [35.18264833189422,1.4769076999800903,1.089145439970417e-15]
>>> disp 3 v
3x3
-0.519  -0.751   0.408
-0.576  -0.046  -0.816
-0.632   0.659   0.408
>>> let d = diagRect 0 s 5 3
>>> disp 3 d
5x3
35.183  0.000  0.000
 0.000  1.477  0.000
 0.000  0.000  0.000
 0.000  0.000  0.000
>>> disp 3 $ u <> d <> tr v
5x3
 1.000   2.000   3.000
 4.000   5.000   6.000
 7.000   8.000   9.000
10.000  11.000  12.000
13.000  14.000  15.000

thinSVD :: Field t => Matrix t -> (Matrix t, Vector Double, Matrix t) Source #

A version of svd which returns only the min (rows m) (cols m) singular vectors of m.

If (u,s,v) = thinSVD m then m == u <> diag s <> tr v.

a = (5><3)
 [  1.0,  2.0,  3.0
 ,  4.0,  5.0,  6.0
 ,  7.0,  8.0,  9.0
 , 10.0, 11.0, 12.0
 , 13.0, 14.0, 15.0 ] :: Matrix Double
>>> let (u,s,v) = thinSVD a
>>> disp 3 u
5x3
-0.101   0.768   0.614
-0.249   0.488  -0.503
-0.396   0.208  -0.405
-0.543  -0.072  -0.140
-0.690  -0.352   0.433
>>> s
fromList [35.18264833189422,1.4769076999800903,1.089145439970417e-15]
>>> disp 3 v
3x3
-0.519  -0.751   0.408
-0.576  -0.046  -0.816
-0.632   0.659   0.408
>>> disp 3 $ u <> diag s <> tr v
5x3
 1.000   2.000   3.000
 4.000   5.000   6.000
 7.000   8.000   9.000
10.000  11.000  12.000
13.000  14.000  15.000

compactSVD :: Field t => Matrix t -> (Matrix t, Vector Double, Matrix t) Source #

Similar to thinSVD, returning only the nonzero singular values and the corresponding singular vectors.

a = (5><3)
 [  1.0,  2.0,  3.0
 ,  4.0,  5.0,  6.0
 ,  7.0,  8.0,  9.0
 , 10.0, 11.0, 12.0
 , 13.0, 14.0, 15.0 ] :: Matrix Double
>>> let (u,s,v) = compactSVD a
>>> disp 3 u
5x2
-0.101   0.768
-0.249   0.488
-0.396   0.208
-0.543  -0.072
-0.690  -0.352
>>> s
fromList [35.18264833189422,1.4769076999800903]
>>> disp 3 u
5x2
-0.101   0.768
-0.249   0.488
-0.396   0.208
-0.543  -0.072
-0.690  -0.352
>>> disp 3 $ u <> diag s <> tr v
5x3
 1.000   2.000   3.000
 4.000   5.000   6.000
 7.000   8.000   9.000
10.000  11.000  12.000
13.000  14.000  15.000

compactSVDTol :: Field t => Double -> Matrix t -> (Matrix t, Vector Double, Matrix t) Source #

compactSVDTol r is similar to compactSVD (for which r=1), but uses tolerance tol=r*g*eps*(max rows cols) to distinguish nonzero singular values, where g is the greatest singular value. If g<r*eps, then only one singular value is returned.

singularValues :: Field t => Matrix t -> Vector Double Source #

Singular values only.

leftSV :: Field t => Matrix t -> (Matrix t, Vector Double) Source #

Singular values and all left singular vectors (as columns).

rightSV :: Field t => Matrix t -> (Vector Double, Matrix t) Source #

Singular values and all right singular vectors (as columns).

Eigendecomposition

eig :: Field t => Matrix t -> (Vector (Complex Double), Matrix (Complex Double)) Source #

Eigenvalues (not ordered) and eigenvectors (as columns) of a general square matrix.

If (s,v) = eig m then m <> v == v <> diag s

a = (3><3)
 [ 3, 0, -2
 , 4, 5, -1
 , 3, 1,  0 ] :: Matrix Double
>>> let (l, v) = eig a
>>> putStr . dispcf 3 . asRow $ l
1x3
1.925+1.523i  1.925-1.523i  4.151
>>> putStr . dispcf 3 $ v
3x3
-0.455+0.365i  -0.455-0.365i   0.181
        0.603          0.603  -0.978
 0.033+0.543i   0.033-0.543i  -0.104
>>> putStr . dispcf 3 $ complex a <> v
3x3
-1.432+0.010i  -1.432-0.010i   0.753
 1.160+0.918i   1.160-0.918i  -4.059
-0.763+1.096i  -0.763-1.096i  -0.433
>>> putStr . dispcf 3 $ v <> diag l
3x3
-1.432+0.010i  -1.432-0.010i   0.753
 1.160+0.918i   1.160-0.918i  -4.059
-0.763+1.096i  -0.763-1.096i  -0.433

eigSH :: Field t => Herm t -> (Vector Double, Matrix t) Source #

Eigenvalues and eigenvectors (as columns) of a complex hermitian or real symmetric matrix, in descending order.

If (s,v) = eigSH m then m == v <> diag s <> tr v

a = (3><3)
 [ 1.0, 2.0, 3.0
 , 2.0, 4.0, 5.0
 , 3.0, 5.0, 6.0 ]
>>> let (l, v) = eigSH a
>>> l
fromList [11.344814282762075,0.17091518882717918,-0.5157294715892575]
>>> disp 3 $ v <> diag l <> tr v
3x3
1.000  2.000  3.000
2.000  4.000  5.000
3.000  5.000  6.000

eigenvalues :: Field t => Matrix t -> Vector (Complex Double) Source #

Eigenvalues (not ordered) of a general square matrix.

eigenvaluesSH :: Field t => Herm t -> Vector Double Source #

Eigenvalues (in descending order) of a complex hermitian or real symmetric matrix.

geigSH Source #

Arguments

:: Field t 
=> Herm t

A

-> Herm t

B

-> (Vector Double, Matrix t) 

Generalized symmetric positive definite eigensystem Av = lBv, for A and B symmetric, B positive definite.

QR

qr :: Field t => Matrix t -> (Matrix t, Matrix t) Source #

QR factorization.

If (q,r) = qr m then m == q <> r, where q is unitary and r is upper triangular. Note: the current implementation is very slow for large matrices. thinQR is much faster.

thinQR :: Field t => Matrix t -> (Matrix t, Matrix t) Source #

A version of qr which returns only the min (rows m) (cols m) columns of q and rows of r.

rq :: Field t => Matrix t -> (Matrix t, Matrix t) Source #

RQ factorization.

If (r,q) = rq m then m == r <> q, where q is unitary and r is upper triangular. Note: the current implementation is very slow for large matrices. thinRQ is much faster.

thinRQ :: Field t => Matrix t -> (Matrix t, Matrix t) Source #

A version of rq which returns only the min (rows m) (cols m) columns of r and rows of q.

qrRaw :: Field t => Matrix t -> QR t Source #

Compute the QR decomposition of a matrix in compact form.

qrgr :: Field t => Int -> QR t -> Matrix t Source #

generate a matrix with k orthogonal columns from the compact QR decomposition obtained by qrRaw.

Cholesky

chol :: Field t => Herm t -> Matrix t Source #

Cholesky factorization of a positive definite hermitian or symmetric matrix.

If c = chol m then c is upper triangular and m == tr c <> c.

mbChol :: Field t => Herm t -> Maybe (Matrix t) Source #

Similar to chol, but instead of an error (e.g., caused by a matrix not positive definite) it returns Nothing.

LU

lu :: Field t => Matrix t -> (Matrix t, Matrix t, Matrix t, t) Source #

Explicit LU factorization of a general matrix.

If (l,u,p,s) = lu m then m == p <> l <> u, where l is lower triangular, u is upper triangular, p is a permutation matrix and s is the signature of the permutation.

luFact :: Numeric t => LU t -> (Matrix t, Matrix t, Matrix t, t) Source #

Compute the explicit LU decomposition from the compact one obtained by luPacked.

Hessenberg

hess :: Field t => Matrix t -> (Matrix t, Matrix t) Source #

Hessenberg factorization.

If (p,h) = hess m then m == p <> h <> tr p, where p is unitary and h is in upper Hessenberg form (it has zero entries below the first subdiagonal).

Schur

schur :: Field t => Matrix t -> (Matrix t, Matrix t) Source #

Schur factorization.

If (u,s) = schur m then m == u <> s <> tr u, where u is unitary and s is a Shur matrix. A complex Schur matrix is upper triangular. A real Schur matrix is upper triangular in 2x2 blocks.

"Anything that the Jordan decomposition can do, the Schur decomposition can do better!" (Van Loan)

Matrix functions

expm :: Field t => Matrix t -> Matrix t Source #

Matrix exponential. It uses a direct translation of Algorithm 11.3.1 in Golub & Van Loan, based on a scaled Pade approximation.

sqrtm :: Field t => Matrix t -> Matrix t Source #

Matrix square root. Currently it uses a simple iterative algorithm described in Wikipedia. It only works with invertible matrices that have a real solution.

m = (2><2) [4,9
           ,0,4] :: Matrix Double
>>> sqrtm m
(2><2)
 [ 2.0, 2.25
 , 0.0,  2.0 ]

For diagonalizable matrices you can try matFunc sqrt:

>>> matFunc sqrt ((2><2) [1,0,0,-1])
(2><2)
 [ 1.0 :+ 0.0, 0.0 :+ 0.0
 , 0.0 :+ 0.0, 0.0 :+ 1.0 ]

matFunc :: (Complex Double -> Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double) Source #

Generic matrix functions for diagonalizable matrices. For instance:

logm = matFunc log

Correlation and convolution

corr Source #

Arguments

:: (Container Vector t, Product t) 
=> Vector t

kernel

-> Vector t

source

-> Vector t 

correlation

>>> corr (fromList[1,2,3]) (fromList [1..10])
fromList [14.0,20.0,26.0,32.0,38.0,44.0,50.0,56.0]

conv :: (Container Vector t, Product t, Num t) => Vector t -> Vector t -> Vector t Source #

convolution (corr with reversed kernel and padded input, equivalent to polynomial product)

>>> conv (fromList[1,1]) (fromList [-1,1])
fromList [-1.0,0.0,1.0]

corrMin :: (Container Vector t, RealElement t, Product t) => Vector t -> Vector t -> Vector t Source #

similar to corr, using min instead of (*)

corr2 :: Product a => Matrix a -> Matrix a -> Matrix a Source #

2D correlation (without padding)

>>> disp 5 $ corr2 (konst 1 (3,3)) (ident 10 :: Matrix Double)
8x8
3  2  1  0  0  0  0  0
2  3  2  1  0  0  0  0
1  2  3  2  1  0  0  0
0  1  2  3  2  1  0  0
0  0  1  2  3  2  1  0
0  0  0  1  2  3  2  1
0  0  0  0  1  2  3  2
0  0  0  0  0  1  2  3

conv2 Source #

Arguments

:: (Num (Matrix a), Product a, Container Vector a) 
=> Matrix a

kernel

-> Matrix a 
-> Matrix a 

2D convolution

>>> disp 5 $ conv2 (konst 1 (3,3)) (ident 10 :: Matrix Double)
12x12
1  1  1  0  0  0  0  0  0  0  0  0
1  2  2  1  0  0  0  0  0  0  0  0
1  2  3  2  1  0  0  0  0  0  0  0
0  1  2  3  2  1  0  0  0  0  0  0
0  0  1  2  3  2  1  0  0  0  0  0
0  0  0  1  2  3  2  1  0  0  0  0
0  0  0  0  1  2  3  2  1  0  0  0
0  0  0  0  0  1  2  3  2  1  0  0
0  0  0  0  0  0  1  2  3  2  1  0
0  0  0  0  0  0  0  1  2  3  2  1
0  0  0  0  0  0  0  0  1  2  2  1
0  0  0  0  0  0  0  0  0  1  1  1

Random arrays

type Seed = Int Source #

data RandDist Source #

Constructors

Uniform

uniform distribution in [0,1)

Gaussian

normal distribution with mean zero and standard deviation one

randomVector Source #

Arguments

:: Seed 
-> RandDist

distribution

-> Int

vector size

-> Vector Double 

Obtains a vector of pseudorandom elements (use randomIO to get a random seed).

rand :: Int -> Int -> IO (Matrix Double) Source #

pseudorandom matrix with uniform elements between 0 and 1

randn :: Int -> Int -> IO (Matrix Double) Source #

pseudorandom matrix with normal elements

>>> disp 3 =<< randn 3 5
3x5
0.386  -1.141   0.491  -0.510   1.512
0.069  -0.919   1.022  -0.181   0.745
0.313  -0.670  -0.097  -1.575  -0.583

gaussianSample Source #

Arguments

:: Seed 
-> Int

number of rows

-> Vector Double

mean vector

-> Herm Double

covariance matrix

-> Matrix Double

result

Obtains a matrix whose rows are pseudorandom samples from a multivariate Gaussian distribution.

uniformSample Source #

Arguments

:: Seed 
-> Int

number of rows

-> [(Double, Double)]

ranges for each column

-> Matrix Double

result

Obtains a matrix whose rows are pseudorandom samples from a multivariate uniform distribution.

Misc

meanCov :: Matrix Double -> (Vector Double, Herm Double) Source #

Compute mean vector and covariance matrix of the rows of a matrix.

>>> meanCov $ gaussianSample 666 1000 (fromList[4,5]) (diagl[2,3])
(fromList [4.010341078059521,5.0197204699640405],
(2><2)
 [     1.9862461923890056, -1.0127225830525157e-2
 , -1.0127225830525157e-2,     3.0373954915729318 ])

rowOuters :: Matrix Double -> Matrix Double -> Matrix Double Source #

outer products of rows

>>> a
(3><2)
 [   1.0,   2.0
 ,  10.0,  20.0
 , 100.0, 200.0 ]
>>> b
(3><3)
 [ 1.0, 2.0, 3.0
 , 4.0, 5.0, 6.0
 , 7.0, 8.0, 9.0 ]
>>> rowOuters a (b ||| 1)
(3><8)
 [   1.0,   2.0,   3.0,   1.0,    2.0,    4.0,    6.0,   2.0
 ,  40.0,  50.0,  60.0,  10.0,   80.0,  100.0,  120.0,  20.0
 , 700.0, 800.0, 900.0, 100.0, 1400.0, 1600.0, 1800.0, 200.0 ]

pairwiseD2 :: Matrix Double -> Matrix Double -> Matrix Double Source #

Matrix of pairwise squared distances of row vectors (using the matrix product trick in blog.smola.org)

normalize :: (Normed (Vector t), Num (Vector t), Field t) => Vector t -> Vector t Source #

Obtains a vector in the same direction with 2-norm=1

peps :: RealFloat x => x Source #

1 + 0.5*peps == 1, 1 + 0.6*peps /= 1

relativeError :: Num a => (a -> Double) -> a -> a -> Double Source #

magnit :: (Element t, Normed (Vector t)) => R -> t -> Bool Source #

Check if the absolute value or complex magnitude is greater than a given threshold

>>> magnit 1E-6 (1E-12 :: R)
False
>>> magnit 1E-6 (3+iC :: C)
True
>>> magnit 0 (3 :: I ./. 5)
True

haussholder :: Field a => a -> Vector a -> Matrix a Source #

udot :: Product e => Vector e -> Vector e -> e Source #

unconjugated dot product

nullspaceSVD Source #

Arguments

:: Field t 
=> Either Double Int

Left "numeric" zero (eg. 1*eps), or Right "theoretical" matrix rank.

-> Matrix t

input matrix m

-> (Vector Double, Matrix t)

rightSV of m

-> Matrix t

nullspace

The nullspace of a matrix from its precomputed SVD decomposition.

orthSVD Source #

Arguments

:: Field t 
=> Either Double Int

Left "numeric" zero (eg. 1*eps), or Right "theoretical" matrix rank.

-> Matrix t

input matrix m

-> (Matrix t, Vector Double)

leftSV of m

-> Matrix t

orth

The range space a matrix from its precomputed SVD decomposition.

ranksv Source #

Arguments

:: Double

numeric zero (e.g. 1*eps)

-> Int

maximum dimension of the matrix

-> [Double]

singular values

-> Int

rank of m

Numeric rank of a matrix from its singular values.

iC :: C Source #

imaginary unit

sym :: Field t => Matrix t -> Herm t Source #

Compute the complex Hermitian or real symmetric part of a square matrix ((x + tr x)/2).

mTm :: Numeric t => Matrix t -> Herm t Source #

Compute the contraction tr x <> x of a general matrix.

trustSym :: Matrix t -> Herm t Source #

At your own risk, declare that a matrix is complex Hermitian or real symmetric for usage in chol, eigSH, etc. Only a triangular part of the matrix will be used.

unSym :: Herm t -> Matrix t Source #

Extract the general matrix from a Herm structure, forgetting its symmetric or Hermitian property.

Auxiliary classes

class Storable a => Element a Source #

Supported matrix elements.

Minimal complete definition

constantD, extractR, setRect, sortI, sortV, compareV, selectV, remapM, rowOp, gemm, reorderV

Instances

Element Double Source # 
Element Float Source # 
Element CInt Source # 
Element Z Source # 
Element (Complex Double) Source # 
Element (Complex Float) Source # 
KnownNat m => Element (Mod m Z) Source # 

Methods

constantD :: Mod m Z -> Int -> Vector (Mod m Z)

extractR :: MatrixOrder -> Matrix (Mod m Z) -> CInt -> Vector CInt -> CInt -> Vector CInt -> IO (Matrix (Mod m Z))

setRect :: Int -> Int -> Matrix (Mod m Z) -> Matrix (Mod m Z) -> IO ()

sortI :: Vector (Mod m Z) -> Vector CInt

sortV :: Vector (Mod m Z) -> Vector (Mod m Z)

compareV :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector CInt

selectV :: Vector CInt -> Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

remapM :: Matrix CInt -> Matrix CInt -> Matrix (Mod m Z) -> Matrix (Mod m Z)

rowOp :: Int -> Mod m Z -> Int -> Int -> Int -> Int -> Matrix (Mod m Z) -> IO ()

gemm :: Vector (Mod m Z) -> Matrix (Mod m Z) -> Matrix (Mod m Z) -> Matrix (Mod m Z) -> IO ()

reorderV :: Vector CInt -> Vector CInt -> Vector (Mod m Z) -> Vector (Mod m Z)

KnownNat m => Element (Mod m I) Source # 

Methods

constantD :: Mod m I -> Int -> Vector (Mod m I)

extractR :: MatrixOrder -> Matrix (Mod m I) -> CInt -> Vector CInt -> CInt -> Vector CInt -> IO (Matrix (Mod m I))

setRect :: Int -> Int -> Matrix (Mod m I) -> Matrix (Mod m I) -> IO ()

sortI :: Vector (Mod m I) -> Vector CInt

sortV :: Vector (Mod m I) -> Vector (Mod m I)

compareV :: Vector (Mod m I) -> Vector (Mod m I) -> Vector CInt

selectV :: Vector CInt -> Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

remapM :: Matrix CInt -> Matrix CInt -> Matrix (Mod m I) -> Matrix (Mod m I)

rowOp :: Int -> Mod m I -> Int -> Int -> Int -> Int -> Matrix (Mod m I) -> IO ()

gemm :: Vector (Mod m I) -> Matrix (Mod m I) -> Matrix (Mod m I) -> Matrix (Mod m I) -> IO ()

reorderV :: Vector CInt -> Vector CInt -> Vector (Mod m I) -> Vector (Mod m I)

class Element e => Container c e Source #

Basic element-by-element functions for numeric containers

Minimal complete definition

conj', size', scalar', scale', addConstant, add', sub, mul, equal, cmap', konst', build', atIndex', minIndex', maxIndex', minElement', maxElement', sumElements', prodElements', step', ccompare', cselect', find', assoc', accum', scaleRecip, divide, arctan2', cmod', fromInt', toInt', fromZ', toZ'

Instances

Container Vector Double Source # 

Methods

conj' :: Vector Double -> Vector Double

size' :: Vector Double -> IndexOf Vector

scalar' :: Double -> Vector Double

scale' :: Double -> Vector Double -> Vector Double

addConstant :: Double -> Vector Double -> Vector Double

add' :: Vector Double -> Vector Double -> Vector Double

sub :: Vector Double -> Vector Double -> Vector Double

mul :: Vector Double -> Vector Double -> Vector Double

equal :: Vector Double -> Vector Double -> Bool

cmap' :: Element b => (Double -> b) -> Vector Double -> Vector b

konst' :: Double -> IndexOf Vector -> Vector Double

build' :: IndexOf Vector -> ArgOf Vector Double -> Vector Double

atIndex' :: Vector Double -> IndexOf Vector -> Double

minIndex' :: Vector Double -> IndexOf Vector

maxIndex' :: Vector Double -> IndexOf Vector

minElement' :: Vector Double -> Double

maxElement' :: Vector Double -> Double

sumElements' :: Vector Double -> Double

prodElements' :: Vector Double -> Double

step' :: Vector Double -> Vector Double

ccompare' :: Vector Double -> Vector Double -> Vector I

cselect' :: Vector I -> Vector Double -> Vector Double -> Vector Double -> Vector Double

find' :: (Double -> Bool) -> Vector Double -> [IndexOf Vector]

assoc' :: IndexOf Vector -> Double -> [(IndexOf Vector, Double)] -> Vector Double

accum' :: Vector Double -> (Double -> Double -> Double) -> [(IndexOf Vector, Double)] -> Vector Double

scaleRecip :: Double -> Vector Double -> Vector Double

divide :: Vector Double -> Vector Double -> Vector Double

arctan2' :: Vector Double -> Vector Double -> Vector Double

cmod' :: Double -> Vector Double -> Vector Double

fromInt' :: Vector I -> Vector Double

toInt' :: Vector Double -> Vector I

fromZ' :: Vector Z -> Vector Double

toZ' :: Vector Double -> Vector Z

Container Vector Float Source # 

Methods

conj' :: Vector Float -> Vector Float

size' :: Vector Float -> IndexOf Vector

scalar' :: Float -> Vector Float

scale' :: Float -> Vector Float -> Vector Float

addConstant :: Float -> Vector Float -> Vector Float

add' :: Vector Float -> Vector Float -> Vector Float

sub :: Vector Float -> Vector Float -> Vector Float

mul :: Vector Float -> Vector Float -> Vector Float

equal :: Vector Float -> Vector Float -> Bool

cmap' :: Element b => (Float -> b) -> Vector Float -> Vector b

konst' :: Float -> IndexOf Vector -> Vector Float

build' :: IndexOf Vector -> ArgOf Vector Float -> Vector Float

atIndex' :: Vector Float -> IndexOf Vector -> Float

minIndex' :: Vector Float -> IndexOf Vector

maxIndex' :: Vector Float -> IndexOf Vector

minElement' :: Vector Float -> Float

maxElement' :: Vector Float -> Float

sumElements' :: Vector Float -> Float

prodElements' :: Vector Float -> Float

step' :: Vector Float -> Vector Float

ccompare' :: Vector Float -> Vector Float -> Vector I

cselect' :: Vector I -> Vector Float -> Vector Float -> Vector Float -> Vector Float

find' :: (Float -> Bool) -> Vector Float -> [IndexOf Vector]

assoc' :: IndexOf Vector -> Float -> [(IndexOf Vector, Float)] -> Vector Float

accum' :: Vector Float -> (Float -> Float -> Float) -> [(IndexOf Vector, Float)] -> Vector Float

scaleRecip :: Float -> Vector Float -> Vector Float

divide :: Vector Float -> Vector Float -> Vector Float

arctan2' :: Vector Float -> Vector Float -> Vector Float

cmod' :: Float -> Vector Float -> Vector Float

fromInt' :: Vector I -> Vector Float

toInt' :: Vector Float -> Vector I

fromZ' :: Vector Z -> Vector Float

toZ' :: Vector Float -> Vector Z

Container Vector Z Source # 
Container Vector I Source # 
(Num a, Element a, Container Vector a) => Container Matrix a Source # 

Methods

conj' :: Matrix a -> Matrix a

size' :: Matrix a -> IndexOf Matrix

scalar' :: a -> Matrix a

scale' :: a -> Matrix a -> Matrix a

addConstant :: a -> Matrix a -> Matrix a

add' :: Matrix a -> Matrix a -> Matrix a

sub :: Matrix a -> Matrix a -> Matrix a

mul :: Matrix a -> Matrix a -> Matrix a

equal :: Matrix a -> Matrix a -> Bool

cmap' :: Element b => (a -> b) -> Matrix a -> Matrix b

konst' :: a -> IndexOf Matrix -> Matrix a

build' :: IndexOf Matrix -> ArgOf Matrix a -> Matrix a

atIndex' :: Matrix a -> IndexOf Matrix -> a

minIndex' :: Matrix a -> IndexOf Matrix

maxIndex' :: Matrix a -> IndexOf Matrix

minElement' :: Matrix a -> a

maxElement' :: Matrix a -> a

sumElements' :: Matrix a -> a

prodElements' :: Matrix a -> a

step' :: Matrix a -> Matrix a

ccompare' :: Matrix a -> Matrix a -> Matrix I

cselect' :: Matrix I -> Matrix a -> Matrix a -> Matrix a -> Matrix a

find' :: (a -> Bool) -> Matrix a -> [IndexOf Matrix]

assoc' :: IndexOf Matrix -> a -> [(IndexOf Matrix, a)] -> Matrix a

accum' :: Matrix a -> (a -> a -> a) -> [(IndexOf Matrix, a)] -> Matrix a

scaleRecip :: a -> Matrix a -> Matrix a

divide :: Matrix a -> Matrix a -> Matrix a

arctan2' :: Matrix a -> Matrix a -> Matrix a

cmod' :: a -> Matrix a -> Matrix a

fromInt' :: Matrix I -> Matrix a

toInt' :: Matrix a -> Matrix I

fromZ' :: Matrix Z -> Matrix a

toZ' :: Matrix a -> Matrix Z

Container Vector (Complex Double) Source # 

Methods

conj' :: Vector (Complex Double) -> Vector (Complex Double)

size' :: Vector (Complex Double) -> IndexOf Vector

scalar' :: Complex Double -> Vector (Complex Double)

scale' :: Complex Double -> Vector (Complex Double) -> Vector (Complex Double)

addConstant :: Complex Double -> Vector (Complex Double) -> Vector (Complex Double)

add' :: Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double)

sub :: Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double)

mul :: Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double)

equal :: Vector (Complex Double) -> Vector (Complex Double) -> Bool

cmap' :: Element b => (Complex Double -> b) -> Vector (Complex Double) -> Vector b

konst' :: Complex Double -> IndexOf Vector -> Vector (Complex Double)

build' :: IndexOf Vector -> ArgOf Vector (Complex Double) -> Vector (Complex Double)

atIndex' :: Vector (Complex Double) -> IndexOf Vector -> Complex Double

minIndex' :: Vector (Complex Double) -> IndexOf Vector

maxIndex' :: Vector (Complex Double) -> IndexOf Vector

minElement' :: Vector (Complex Double) -> Complex Double

maxElement' :: Vector (Complex Double) -> Complex Double

sumElements' :: Vector (Complex Double) -> Complex Double

prodElements' :: Vector (Complex Double) -> Complex Double

step' :: Vector (Complex Double) -> Vector (Complex Double)

ccompare' :: Vector (Complex Double) -> Vector (Complex Double) -> Vector I

cselect' :: Vector I -> Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double)

find' :: (Complex Double -> Bool) -> Vector (Complex Double) -> [IndexOf Vector]

assoc' :: IndexOf Vector -> Complex Double -> [(IndexOf Vector, Complex Double)] -> Vector (Complex Double)

accum' :: Vector (Complex Double) -> (Complex Double -> Complex Double -> Complex Double) -> [(IndexOf Vector, Complex Double)] -> Vector (Complex Double)

scaleRecip :: Complex Double -> Vector (Complex Double) -> Vector (Complex Double)

divide :: Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double)

arctan2' :: Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double)

cmod' :: Complex Double -> Vector (Complex Double) -> Vector (Complex Double)

fromInt' :: Vector I -> Vector (Complex Double)

toInt' :: Vector (Complex Double) -> Vector I

fromZ' :: Vector Z -> Vector (Complex Double)

toZ' :: Vector (Complex Double) -> Vector Z

Container Vector (Complex Float) Source # 

Methods

conj' :: Vector (Complex Float) -> Vector (Complex Float)

size' :: Vector (Complex Float) -> IndexOf Vector

scalar' :: Complex Float -> Vector (Complex Float)

scale' :: Complex Float -> Vector (Complex Float) -> Vector (Complex Float)

addConstant :: Complex Float -> Vector (Complex Float) -> Vector (Complex Float)

add' :: Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float)

sub :: Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float)

mul :: Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float)

equal :: Vector (Complex Float) -> Vector (Complex Float) -> Bool

cmap' :: Element b => (Complex Float -> b) -> Vector (Complex Float) -> Vector b

konst' :: Complex Float -> IndexOf Vector -> Vector (Complex Float)

build' :: IndexOf Vector -> ArgOf Vector (Complex Float) -> Vector (Complex Float)

atIndex' :: Vector (Complex Float) -> IndexOf Vector -> Complex Float

minIndex' :: Vector (Complex Float) -> IndexOf Vector

maxIndex' :: Vector (Complex Float) -> IndexOf Vector

minElement' :: Vector (Complex Float) -> Complex Float

maxElement' :: Vector (Complex Float) -> Complex Float

sumElements' :: Vector (Complex Float) -> Complex Float

prodElements' :: Vector (Complex Float) -> Complex Float

step' :: Vector (Complex Float) -> Vector (Complex Float)

ccompare' :: Vector (Complex Float) -> Vector (Complex Float) -> Vector I

cselect' :: Vector I -> Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float)

find' :: (Complex Float -> Bool) -> Vector (Complex Float) -> [IndexOf Vector]

assoc' :: IndexOf Vector -> Complex Float -> [(IndexOf Vector, Complex Float)] -> Vector (Complex Float)

accum' :: Vector (Complex Float) -> (Complex Float -> Complex Float -> Complex Float) -> [(IndexOf Vector, Complex Float)] -> Vector (Complex Float)

scaleRecip :: Complex Float -> Vector (Complex Float) -> Vector (Complex Float)

divide :: Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float)

arctan2' :: Vector (Complex Float) -> Vector (Complex Float) -> Vector (Complex Float)

cmod' :: Complex Float -> Vector (Complex Float) -> Vector (Complex Float)

fromInt' :: Vector I -> Vector (Complex Float)

toInt' :: Vector (Complex Float) -> Vector I

fromZ' :: Vector Z -> Vector (Complex Float)

toZ' :: Vector (Complex Float) -> Vector Z

KnownNat m => Container Vector (Mod m Z) Source # 

Methods

conj' :: Vector (Mod m Z) -> Vector (Mod m Z)

size' :: Vector (Mod m Z) -> IndexOf Vector

scalar' :: Mod m Z -> Vector (Mod m Z)

scale' :: Mod m Z -> Vector (Mod m Z) -> Vector (Mod m Z)

addConstant :: Mod m Z -> Vector (Mod m Z) -> Vector (Mod m Z)

add' :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

sub :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

mul :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

equal :: Vector (Mod m Z) -> Vector (Mod m Z) -> Bool

cmap' :: Element b => (Mod m Z -> b) -> Vector (Mod m Z) -> Vector b

konst' :: Mod m Z -> IndexOf Vector -> Vector (Mod m Z)

build' :: IndexOf Vector -> ArgOf Vector (Mod m Z) -> Vector (Mod m Z)

atIndex' :: Vector (Mod m Z) -> IndexOf Vector -> Mod m Z

minIndex' :: Vector (Mod m Z) -> IndexOf Vector

maxIndex' :: Vector (Mod m Z) -> IndexOf Vector

minElement' :: Vector (Mod m Z) -> Mod m Z

maxElement' :: Vector (Mod m Z) -> Mod m Z

sumElements' :: Vector (Mod m Z) -> Mod m Z

prodElements' :: Vector (Mod m Z) -> Mod m Z

step' :: Vector (Mod m Z) -> Vector (Mod m Z)

ccompare' :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector I

cselect' :: Vector I -> Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

find' :: (Mod m Z -> Bool) -> Vector (Mod m Z) -> [IndexOf Vector]

assoc' :: IndexOf Vector -> Mod m Z -> [(IndexOf Vector, Mod m Z)] -> Vector (Mod m Z)

accum' :: Vector (Mod m Z) -> (Mod m Z -> Mod m Z -> Mod m Z) -> [(IndexOf Vector, Mod m Z)] -> Vector (Mod m Z)

scaleRecip :: Mod m Z -> Vector (Mod m Z) -> Vector (Mod m Z)

divide :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

arctan2' :: Vector (Mod m Z) -> Vector (Mod m Z) -> Vector (Mod m Z)

cmod' :: Mod m Z -> Vector (Mod m Z) -> Vector (Mod m Z)

fromInt' :: Vector I -> Vector (Mod m Z)

toInt' :: Vector (Mod m Z) -> Vector I

fromZ' :: Vector Z -> Vector (Mod m Z)

toZ' :: Vector (Mod m Z) -> Vector Z

KnownNat m => Container Vector (Mod m I) Source # 

Methods

conj' :: Vector (Mod m I) -> Vector (Mod m I)

size' :: Vector (Mod m I) -> IndexOf Vector

scalar' :: Mod m I -> Vector (Mod m I)

scale' :: Mod m I -> Vector (Mod m I) -> Vector (Mod m I)

addConstant :: Mod m I -> Vector (Mod m I) -> Vector (Mod m I)

add' :: Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

sub :: Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

mul :: Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

equal :: Vector (Mod m I) -> Vector (Mod m I) -> Bool

cmap' :: Element b => (Mod m I -> b) -> Vector (Mod m I) -> Vector b

konst' :: Mod m I -> IndexOf Vector -> Vector (Mod m I)

build' :: IndexOf Vector -> ArgOf Vector (Mod m I) -> Vector (Mod m I)

atIndex' :: Vector (Mod m I) -> IndexOf Vector -> Mod m I

minIndex' :: Vector (Mod m I) -> IndexOf Vector

maxIndex' :: Vector (Mod m I) -> IndexOf Vector

minElement' :: Vector (Mod m I) -> Mod m I

maxElement' :: Vector (Mod m I) -> Mod m I

sumElements' :: Vector (Mod m I) -> Mod m I

prodElements' :: Vector (Mod m I) -> Mod m I

step' :: Vector (Mod m I) -> Vector (Mod m I)

ccompare' :: Vector (Mod m I) -> Vector (Mod m I) -> Vector I

cselect' :: Vector I -> Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

find' :: (Mod m I -> Bool) -> Vector (Mod m I) -> [IndexOf Vector]

assoc' :: IndexOf Vector -> Mod m I -> [(IndexOf Vector, Mod m I)] -> Vector (Mod m I)

accum' :: Vector (Mod m I) -> (Mod m I -> Mod m I -> Mod m I) -> [(IndexOf Vector, Mod m I)] -> Vector (Mod m I)

scaleRecip :: Mod m I -> Vector (Mod m I) -> Vector (Mod m I)

divide :: Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

arctan2' :: Vector (Mod m I) -> Vector (Mod m I) -> Vector (Mod m I)

cmod' :: Mod m I -> Vector (Mod m I) -> Vector (Mod m I)

fromInt' :: Vector I -> Vector (Mod m I)

toInt' :: Vector (Mod m I) -> Vector I

fromZ' :: Vector Z -> Vector (Mod m I)

toZ' :: Vector (Mod m I) -> Vector Z

class (Num e, Element e) => Product e Source #

Matrix product and related functions

Minimal complete definition

multiply, absSum, norm1, norm2, normInf

Instances

Product Double Source # 
Product Float Source # 
Product Z Source # 
Product I Source # 
Product (Complex Double) Source # 
Product (Complex Float) Source # 
KnownNat m => Product (Mod m Z) Source # 

Methods

multiply :: Matrix (Mod m Z) -> Matrix (Mod m Z) -> Matrix (Mod m Z)

absSum :: Vector (Mod m Z) -> RealOf (Mod m Z)

norm1 :: Vector (Mod m Z) -> RealOf (Mod m Z)

norm2 :: Vector (Mod m Z) -> RealOf (Mod m Z)

normInf :: Vector (Mod m Z) -> RealOf (Mod m Z)

KnownNat m => Product (Mod m I) Source # 

Methods

multiply :: Matrix (Mod m I) -> Matrix (Mod m I) -> Matrix (Mod m I)

absSum :: Vector (Mod m I) -> RealOf (Mod m I)

norm1 :: Vector (Mod m I) -> RealOf (Mod m I)

norm2 :: Vector (Mod m I) -> RealOf (Mod m I)

normInf :: Vector (Mod m I) -> RealOf (Mod m I)

class LSDiv c Source #

Minimal complete definition

linSolve

Instances

LSDiv Vector Source # 

Methods

linSolve :: Field t => Matrix t -> Vector t -> Vector t

LSDiv Matrix Source # 

Methods

linSolve :: Field t => Matrix t -> Matrix t -> Matrix t

data Herm t Source #

A matrix that, by construction, it is known to be complex Hermitian or real symmetric.

It can be created using sym, mTm, or trustSym, and the matrix can be extracted using unSym.

Instances

Field t => Linear t Herm Source # 

Methods

scale :: t -> Herm t -> Herm t Source #

(Element t, Show t) => Show (Herm t) Source # 

Methods

showsPrec :: Int -> Herm t -> ShowS #

show :: Herm t -> String #

showList :: [Herm t] -> ShowS #

(NFData t, Numeric t) => NFData (Herm t) Source # 

Methods

rnf :: Herm t -> () #

Field t => Additive (Herm t) Source # 

Methods

add :: Herm t -> Herm t -> Herm t Source #

class Complexable c Source #

Structures that may contain complex numbers

Minimal complete definition

toComplex', fromComplex', comp', single', double'

Instances

Complexable Vector Source # 

Methods

toComplex' :: RealElement e => (Vector e, Vector e) -> Vector (Complex e)

fromComplex' :: RealElement e => Vector (Complex e) -> (Vector e, Vector e)

comp' :: RealElement e => Vector e -> Vector (Complex e)

single' :: Precision a b => Vector b -> Vector a

double' :: Precision a b => Vector a -> Vector b

Complexable Matrix Source # 

Methods

toComplex' :: RealElement e => (Matrix e, Matrix e) -> Matrix (Complex e)

fromComplex' :: RealElement e => Matrix (Complex e) -> (Matrix e, Matrix e)

comp' :: RealElement e => Matrix e -> Matrix (Complex e)

single' :: Precision a b => Matrix b -> Matrix a

double' :: Precision a b => Matrix a -> Matrix b

class (Element t, Element (Complex t), RealFloat t) => RealElement t Source #

Supported real types

type family RealOf x Source #

Instances

type RealOf Double Source # 
type RealOf Float Source # 
type RealOf Z Source # 
type RealOf Z = Z
type RealOf I Source # 
type RealOf I = I
type RealOf (Complex Double) Source # 
type RealOf (Complex Float) Source # 
type RealOf (Mod n Z) Source # 
type RealOf (Mod n Z) = Z
type RealOf (Mod n I) Source # 
type RealOf (Mod n I) = I

type family SingleOf x Source #

type family DoubleOf x Source #

type family IndexOf (c :: * -> *) Source #

Instances

class (Numeric t, Convert t, Normed Matrix t, Normed Vector t, Floating t, Linear t Vector, Linear t Matrix, Additive (Vector t), Additive (Matrix t), RealOf t ~ Double) => Field t Source #

Generic linear algebra functions for double precision real and complex matrices.

(Single precision data can be converted using single and double).

Minimal complete definition

svd', thinSVD', sv', luPacked', luSolve', mbLinearSolve', linearSolve', cholSolve', triSolve', triDiagSolve', ldlPacked', ldlSolve', linearSolveSVD', linearSolveLS', eig', eigSH'', eigOnly, eigOnlySH, cholSH', mbCholSH', qr', qrgr', hess', schur'

Instances

Field Double Source # 
Field (Complex Double) Source # 

Methods

svd' :: Matrix (Complex Double) -> (Matrix (Complex Double), Vector Double, Matrix (Complex Double))

thinSVD' :: Matrix (Complex Double) -> (Matrix (Complex Double), Vector Double, Matrix (Complex Double))

sv' :: Matrix (Complex Double) -> Vector Double

luPacked' :: Matrix (Complex Double) -> (Matrix (Complex Double), [Int])

luSolve' :: (Matrix (Complex Double), [Int]) -> Matrix (Complex Double) -> Matrix (Complex Double)

mbLinearSolve' :: Matrix (Complex Double) -> Matrix (Complex Double) -> Maybe (Matrix (Complex Double))

linearSolve' :: Matrix (Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double)

cholSolve' :: Matrix (Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double)

triSolve' :: UpLo -> Matrix (Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double)

triDiagSolve' :: Vector (Complex Double) -> Vector (Complex Double) -> Vector (Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double)

ldlPacked' :: Matrix (Complex Double) -> (Matrix (Complex Double), [Int])

ldlSolve' :: (Matrix (Complex Double), [Int]) -> Matrix (Complex Double) -> Matrix (Complex Double)

linearSolveSVD' :: Matrix (Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double)

linearSolveLS' :: Matrix (Complex Double) -> Matrix (Complex Double) -> Matrix (Complex Double)

eig' :: Matrix (Complex Double) -> (Vector (Complex Double), Matrix (Complex Double))

eigSH'' :: Matrix (Complex Double) -> (Vector Double, Matrix (Complex Double))

eigOnly :: Matrix (Complex Double) -> Vector (Complex Double)

eigOnlySH :: Matrix (Complex Double) -> Vector Double

cholSH' :: Matrix (Complex Double) -> Matrix (Complex Double)

mbCholSH' :: Matrix (Complex Double) -> Maybe (Matrix (Complex Double))

qr' :: Matrix (Complex Double) -> (Matrix (Complex Double), Vector (Complex Double))

qrgr' :: Int -> (Matrix (Complex Double), Vector (Complex Double)) -> Matrix (Complex Double)

hess' :: Matrix (Complex Double) -> (Matrix (Complex Double), Matrix (Complex Double))

schur' :: Matrix (Complex Double) -> (Matrix (Complex Double), Matrix (Complex Double))

class Linear t c where Source #

Minimal complete definition

scale

Methods

scale :: t -> c t -> c t Source #

Instances

Container Matrix t => Linear t Matrix Source # 

Methods

scale :: t -> Matrix t -> Matrix t Source #

Container Vector t => Linear t Vector Source # 

Methods

scale :: t -> Vector t -> Vector t Source #

Field t => Linear t Herm Source # 

Methods

scale :: t -> Herm t -> Herm t Source #

class Additive c where Source #

Minimal complete definition

add

Methods

add :: c -> c -> c Source #

Instances

Container Vector t => Additive (Vector t) Source # 

Methods

add :: Vector t -> Vector t -> Vector t Source #

Container Matrix t => Additive (Matrix t) Source # 

Methods

add :: Matrix t -> Matrix t -> Matrix t Source #

Field t => Additive (Herm t) Source # 

Methods

add :: Herm t -> Herm t -> Herm t Source #

Additive (C n) Source # 

Methods

add :: C n -> C n -> C n Source #

Additive (R n) Source # 

Methods

add :: R n -> R n -> R n Source #

KnownNat n => Additive (Sym n) Source # 

Methods

add :: Sym n -> Sym n -> Sym n Source #

(KnownNat m, KnownNat n) => Additive (M m n) Source # 

Methods

add :: M m n -> M m n -> M m n Source #

(KnownNat m, KnownNat n) => Additive (L m n) Source # 

Methods

add :: L m n -> L m n -> L m n Source #

class Transposable m mt | m -> mt, mt -> m where Source #

Minimal complete definition

tr, tr'

Methods

tr :: m -> mt Source #

conjugate transpose

tr' :: m -> mt Source #

transpose

Instances

Transposable GMatrix GMatrix Source # 
(CTrans t, Container Vector t) => Transposable (Matrix t) (Matrix t) Source # 

Methods

tr :: Matrix t -> Matrix t Source #

tr' :: Matrix t -> Matrix t Source #

KnownNat n => Transposable (Her n) (Her n) Source # 

Methods

tr :: Her n -> Her n Source #

tr' :: Her n -> Her n Source #

KnownNat n => Transposable (Sym n) (Sym n) Source # 

Methods

tr :: Sym n -> Sym n Source #

tr' :: Sym n -> Sym n Source #

(KnownNat n, KnownNat m) => Transposable (M m n) (M n m) Source # 

Methods

tr :: M m n -> M n m Source #

tr' :: M m n -> M n m Source #

(KnownNat n, KnownNat m) => Transposable (L m n) (L n m) Source # 

Methods

tr :: L m n -> L n m Source #

tr' :: L m n -> L n m Source #

data LU t Source #

LU decomposition of a matrix in a compact format.

Constructors

LU (Matrix t) [Int] 

Instances

(Element t, Show t) => Show (LU t) Source # 

Methods

showsPrec :: Int -> LU t -> ShowS #

show :: LU t -> String #

showList :: [LU t] -> ShowS #

(NFData t, Numeric t) => NFData (LU t) Source # 

Methods

rnf :: LU t -> () #

data LDL t Source #

LDL decomposition of a complex Hermitian or real symmetric matrix in a compact format.

Constructors

LDL (Matrix t) [Int] 

Instances

(Element t, Show t) => Show (LDL t) Source # 

Methods

showsPrec :: Int -> LDL t -> ShowS #

show :: LDL t -> String #

showList :: [LDL t] -> ShowS #

(NFData t, Numeric t) => NFData (LDL t) Source # 

Methods

rnf :: LDL t -> () #

data QR t Source #

QR decomposition of a matrix in compact form. (The orthogonal matrix is not explicitly formed.)

Constructors

QR (Matrix t) (Vector t) 

Instances

(NFData t, Numeric t) => NFData (QR t) Source # 

Methods

rnf :: QR t -> () #

data CGState Source #

Constructors

CGState 

Fields

class Testable t where Source #

Minimal complete definition

checkT

Methods

checkT :: t -> (Bool, IO ()) Source #

ioCheckT :: t -> IO (Bool, IO ()) Source #

Instances

KnownNat m => Testable (Matrix (Mod m I)) Source # 

Methods

checkT :: Matrix (Mod m I) -> (Bool, IO ()) Source #

ioCheckT :: Matrix (Mod m I) -> IO (Bool, IO ()) Source #