io.improbable.keanu.tensor.dbl

Class KeanuLapack

java.lang.Object

io.improbable.keanu.tensor.dbl.KeanuLapack

public class KeanuLapack
extends java.lang.Object

https://software.intel.com/en-us/mkl-developer-reference-c-matrix-factorization-lapack-computational-routines

This is a wrapper around a few LAPACK functions used directly in Keanu

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	KeanuLapack.Triangular

Constructor Summary

Constructors

Constructor and Description
KeanuLapack()

Method Summary

Modifier and Type	Method and Description
static int	dgetrf(int m, int n, double[] buffer, int[] ipiv) https://software.intel.com/en-us/mkl-developer-reference-c-getrf
static int	dgetri(int N, double[] buffer, int[] ipiv) https://software.intel.com/en-us/mkl-developer-reference-c-getri
static int	dpotrf(KeanuLapack.Triangular upperLower, int N, double[] buffer) https://software.intel.com/en-us/mkl-developer-reference-c-potrf

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

KeanuLapack

public KeanuLapack()

Method Detail

dpotrf

public static int dpotrf(KeanuLapack.Triangular upperLower,
                         int N,
                         double[] buffer)

https://software.intel.com/en-us/mkl-developer-reference-c-potrf

Computes the Cholesky factorization of a symmetric (Hermitian) positive-definite matrix.

Parameters:

upperLower - either upper triangular or lower triangular specifies format of result

N - length of one dimension in the square matrix to operate on

buffer - matrix to take cholesky decoposition of and buffer to store result

Returns:

0 if successful, -i if error at i index, or positive if matrix is not positive-definite

dgetrf

public static int dgetrf(int m,
                         int n,
                         double[] buffer,
                         int[] ipiv)

https://software.intel.com/en-us/mkl-developer-reference-c-getrf

Computes the LU factorization of a general m-by-n matrix.

Parameters:

m - row dimension of matrix

n - column dimensino of matrix

buffer - buffer that represents the matrix for LU factorization

ipiv - The pivot indices

Returns:

0 if successful, -i if error at i index, or i if matrix is singular and ii index is zero

dgetri

public static int dgetri(int N,
                         double[] buffer,
                         int[] ipiv)

https://software.intel.com/en-us/mkl-developer-reference-c-getri

Computes the inverse of an LU-factored general matrix.

Parameters:

N - length of one dimension in the square matrix to operate on

buffer - buffer of data that represents square matrix

ipiv - The pivot indices

Returns:

0 if successful, -i if error at i index, or i if matrix is singular and ii index is zero