rogues 0.3.0

Rogues is a Python and numpy/scipy port of Nicholas Higham’s m*lab test matrices.

These matrices are a collection of interesting matrices that appear in m*lab’s ‘gallery’ collection. This collection was originally defined and implemented by Prof. Nicholas Higham of Manchester University and is more fully discussed in “The Test Matrix Toolbox for Matlab (Version 3.0)”, N.J. Higham, Numerical Analysis Report No. 276, September 1995 and can be found at

http://www.maths.manchester.ac.uk/~nareports

By ‘interesting’ we mean that these matrices either present some challenges to numerical algorithms or have some a set of interesting properties. The documentation of the individual functions contains much more info, as well as references.

Also included are a set of matrix utility functions that are needed for generating some of members of the collection as well as a few functions from Prof. Higham’s matrixcomp package. One of the more interesting routines here is mdsmax, a direct search optimization algorithm.

The rogues package depends on numpy and scipy, both of which must be installed. Additionally, there are a few routines that deal with plotting, and these use matplotlib. While ipython is not strictly necessary, it is a very convenient environment for numpy / scipy / matplotlib. Finally, the unit tests utilize the nose package and the numpy wrappers around nose. To run the tests from inside ipython, for example, type the following:

!nosetests rogues

The included matrix generation functions are:

• cauchy Cauchy matrix

• chebspec Chebyshev spectral differentiation matrix

• chebvand Vandermonde-like matrix for the Chebyshev polynomials

• chow Chow matrix - a singular Toeplitz lower Hessenberg matrix

• clement Clement matrix - tridiagonal with zero diagonal entries

• comp Comparison matrices

• compan Companion matrix

• condex Counterexamples to matrix condition number estimators

• cycol Matrix whose columns repeat cyclically

• dingdong Dingdong matrix - a symmetric Hankel matrix

• dorr Dorr matrix - diagonally dominant, ill conditioned, tridiagonal.

• dramadah A (0,1) matrix whose inverse has large integer entries

• fiedler Fiedler matrix - symmetric

• forsythe Forsythe matrix - a perturbed Jordan block

• frank Frank matrix - ill conditioned eigenvalues.

• gearm Gear matrix

• gfpp Matrix giving maximal growth factor for GW with partial pivoting

• grcar Grcar matrix - a Toeplitz matrix with sensitive eigenvalues.

• hadamard Hadamard matrix

• hankel Hankel matrix

• hanowa A matrix whose eigenvalues lie on a vertical line in C

• hilb Hilbert matrix

• invhess Inverse of an upper Hessenberg matrix

• invol An involutory matrix

• ipjfact A Hankel matrix with factorial elements

• jordbloc Jordan block matrix

• kahan Kahan matrix - upper trapezoidal

• kms Kar-Murdock-Szego Toeplitz matrix

• krylov Krylov matrix

• lauchli Lauchli matrix - rectangular

• lehmer Lehmer matrix - symmetric positive definite

• lesp A tridiagonal matrix with real, sensitve eigenvalues

• lotkin Lotkin matrix

• minij Symmetric positive definite matrix min(i,j)

• moler Moler matrix symmetric positive definite

• neumann Singular matrix from the descrete Neumann problem (sparse)

• ohess Random, orthogonal upper Hessenberg matrix

• parter Parter matrix - a Toeplitz matrix with singular values near pi

• pascal Pascal matrix

• pdtoep Symmetric positive definite Toeplitz matrix

• pei Pei matrix

• pentoep Tentadiagonal Toeplitz matrix (sparse)

• poisson Block tridiagonal matrix from Poisson’s equation (sparse)

• prolate Prolate matrix - symmetric, ill-conditioned Toeplitz matrix

• qmult Pre-multiply by random orthogonal matrix

• rando Random matrix with elements -1, 0, or 1

• randsvd Random matrix with pre-assigned singular values

• redheff A (0,1) matrix of Redheffer associated with the Riemann hypothesis

• riemann A matrix associated with the Riemann hypothesis

• smoke Smoke matrix - complex, with a ‘smoke ring’ pseudospectrum

• triw Upper triangular matrix discussed by Wilkinson and others

• wathen Wathen matrix - a finite element matrix (sparse, random entries)

• wilk Various specific matrices devised /discussed by Wilkenson

• wilkinson Wilkinson matrix of size n, where n must be odd

Some of generally useful matrix utility functions:

• augment Agumented system matrix

• bandred Band reduction by two-sided unitary transformations

• cgs Classical Gram-Schmidt QR factorization

• cond Matrix condition number in 1,2,Frobenius, or infinity norm

• condeig Condition numbers for eigenvalues of a matrix

• cpltaxes Determine suitable axis for plot of complex vector

• dual Dual vector with respect to Holder p-norm

• ge Gaussian elimination without pivoting

• hankel Given first row, returns a Toeplitz type matrix

• house Householder matrix

• mdsmax Multidimensional search method for direct search optimization

• mgs Modified Gram-Schmidt QR factorization

• pow2 Vector whose i-th element is 2 ** x[i], where x[] is input

• ps Dot plot of a pseudospectrum

• repmat Simple re-implementation of m*lab’s repmat function

• rq Rayleigh quotient

• skewpart Skew-symmetric (skew-Hermitian) part

• sparsify Randomly sets matrix elements to zero

• sub Principal submatrix

• symmpart Symmetric (Hermitian) part

• toeplitz Returns toeplitz matrix given first row of the matrix

• treshape Reshape vector to or from (unit) triangular matrix

• tridiag Sparse tridiagonl matrix given the diagonals

• vand Vandermonde matrix

• vecperm Vector permutation matrix

More information is available on any of these functions by typing “help <funcname>”

Release 0.2.0 Notes The unit tests are now included in the distribution. They work with nosetests and if you have installed the source (not the zip’d egg file) you can do the following

!nosetests rogues

Sadly, this does not work with zipped egg files.

Fixed several small bugs in the use of np.max. Added matrix visualization routing ‘see’

Release 0.3.0 Notes Added distribute_setup.py to fix installation problems. Ported to Python 3. Tested on Ubuntu 12.04 with Python2.7.3, numpy 1.6.1, scipy 0.10.0, ipython 0.12. Also tested with Python 3.2.3, numpy 1.6.2, scipy 0.11.0, ipython 0.13.1. Tested on Windows 7 with Python 2.7.3, numpy 1.6.2, scipy 0.11.0, ipython 0.13.1