(Compatible with Mathematica version 5 to 7)
J. William Helton, Maurício C. de Oliveira and Mark Stankus
Math Dept., UCSD
Robert L. Miller
General Atomic Corp.
La Jolla, California 92093
Copyright by Helton and Miller on June 1991, Feb 1994
Copyright by Helton, Miller and Stankus on March 1996
Copyright by Helton, Miller and Stankus on September 2001
Copyright by Helton, Miller, Stankus and de Oliveira on January 2010
All Rights Reserved.
If you would like to try the NCAlgebra package or want updates go to the NCAlgebra web
or contact email@example.com or MathSource@wri.com.
The basic NCAlgebra program was written by the authors and David Hurst, Daniel Lamm,
Orlando Merino, Robert Obarr, Henry Pfister, Stan Yoshinobu, Phillipe Bergman, Dell
Kronewitter, and Eric Rowell. Various additions include contributions by Dave Glickstein, Juan
Camino, Jeff Ovall, Tony Mendes, and Tony Shaheen. The Mathematica C++ hybrid was
developed with Kurt Schneider, Victor Shih and Mike Moore. Control demos are coauthored with
Mike Walker. Simplification commands are based on theory developed by Bill Helton, John Wavrik
and Mark Stankus. The beginnings of the program come from eran@slac. This program
was written with support from the AFOSR, the NSF, the Lab for Mathematics and
Statistics at UCSD, the UCSD Faculty Mentor Program and the US Department of
0.1 ReleaseNotes NCAlgebra
ReleaseNotes NCAlgebra 4.0 January 2010
Code has been cleaned up and improved. Basic functionality is supported without changes.
ReleaseNotes NCAlgebra 3.7 Sept. 2002
Convexity We changed some defaults and fixed some bugs in NCConvexityRegion and
NCMatrixOfQuadratic. Now it runs quickly on fairly big problems.
Path Setting Paths were set to make NCAlgebra more path independent.
CSP Our support for Control System Professional was updated to accomodate the second (2002)
version of CSP.
ReleaseNotes NCAlgebra 3.5 Sept. 2001
Basic Changes You no longer have to set every variable to be noncommutative. We have a
command NCGuts which has an option called NCSetNC. When set to True, all letters are
automatically noncommutative unless SetCommutative makes them commutative.
A further option of NCGuts allows one to use “**” to multiply matrices with noncommutative
entries – the more cumbersome MatMult command is no longer needed. While this option seems
dangerous to Bill, it makes many computations prettier and easier to type. If you don’t trust the
answer, then don’t use the option.
Commands For Matricies With Noncommuting Entries We now have an LDU
decomposition for matricies with noncommuting entries. Also, there is a command for computing
the inverse of such matrices (however this only works under strong assumptions).
NCMatrixOfQuadratic gives a vector matrix factorization of a symmetric quadratic
A Second Derivative Command NCHessian computes the Hessian of a function with
noncommuting variables and coefficents.
Computing The Region Where A Noncommutative Function is Convex
NCConvexityRegion is a command used to determine the region of formal noncommutative
inequalities where a given noncommutative function is convex.
- NCGuts: NCGuts holds set of options geared for simplifying transposing, finding the inverse,
and multiplying matrices conaining noncommuting variables.
NCStrongProduct1 - > False is the first option of NCGuts. When True, ** serves
to multiply matrices with noncommutative entries as well as maintaining its original
function. In addition, tp[ ] and tpMat are the same. The default setting is True.
NCStrongProduct2 - > False is the second option of NCGuts. When set to True, if
M is a matrix with noncommutative entries, inv[M] returns a formula expression for
the inverse of M. NCStrongProduct2 forces NCStrongProduct1.
NCSetNC - > False is the last option of NCGuts. When set to True, all letters
are automatically noncommutative unless SetCommutative makes them commutative.
This replaces the need for repeated calls to SetNonCommutative.
Commands For Matricies With Noncommuting Entries
- NCLDUDecomposition: Given a square matrix M with noncommutative entries, this
command finds the LDU decomposition of M. It returns a list of four elements, namely
L,D,U, and P such that PXPT = LDU. The first element is the lower triangular
matrix L, the second element is the diagonal matrix D, the third element is the upper
triangular matrix U, and the fourth is the permutation matrix P (the identity is
returned if no permutation is needed). As an option, it may also return a list of the
permutations used at each step of the LDU factorization as a fifth element.
- NCAllPermutationLDU: NCAllPermutationLDU returns the LDU decomposition of a
matrix after all possible column permutations are applied. The code cycles through all
possible permutations and calls NCLDUDecomposition for each one. As an option, the
permutations used for each LDU decomposition can also be returned.
- NCMatrixOfQuadratic: NCMatrixOfQuadratic gives a vector matrix factorization of a
symmetric quadratic noncommutative function. A three element list is the output. The
first element is the left border vector, the second element is a symmetric coefficent
matrix, and the third is the right border vector. The border vectors contain the variables
in the given quadratic function and their transposes.
- NCIndependenceCheck: NCIndependenceCheck verifies whether or not a given set of
polynomials are independent or not. It analyzes each list of polynomials separately.
There are three possible types of outputs for each list. Two of them correspond to
NCIndependenceCheck successfully determining whether or not the list of polynomials
is independent. The third type of output corresponds to an unsuccessful attempt at
determining dependence or independence.
- NCBorderVectorGather: NCBorderVectorGather can be used to gather the polynomial
coefficents preceeding the elements given in a list of variables whenever they occur.
- NCPermutationMatrix: NCPermutationMatrix returns the permutation matrix associated
with the list of the first n integers. It gives the identity matrix with its columns
- NCMatrixToPermutation: NCMatrixToPermutation returns the permutation associated
with the permutation matrix, aMatrix. It is the inverse of NCPermutationMatrix.
- NCInverse: NCInverse gives a symbolic inverse of a matrix with noncommutative entries.
A Second Derivative Command
- NCHessian: NCHessian computes the Hessian of a function with noncommuting variables
and coefficents. This is a second directional derivative which can be thought of as
the second order term in the noncommutative Taylor expansion. Output will be a
symmetric quadratic function with respect to the directions of differentiation.
Computing The Region Where A Noncommutative Function is Convex
- NCConvexityRegion: This command is used to determine the region of formal
noncommutative inequalities where a given noncommutative function is convex.
NCConvexityRegion preforms three main operations. Given a noncommutative
function F, the Hessian of F is computed with NCHessian. Then, using
NCMatrixOfQuadratic, the Hessian is factored into vector matrix vector form. Finally,
NCAllPermutationLDU finds the LDU decomposition of the symmetric coefficent
matrix. The diagonal elements in the diagonal matrix in the LDU decomposition is
ReleaseNotes NCAlgebra 3.0
NCAlgebra 3.0 has several added functions.
- LDU decomposition for block matices, to include a block Cholesky decompsition.
- Formulas for inverses of block matrices.
- A command which differentiates functions of the form
- Support for the Mathematica toolbox Control System Professional. It gives CSP the
ability to handle non-commuting objects.
- A function which represents elements of an algebra as n×n matrices with commuting
symbolic entries, or with inegers.
- Online Help - While we have not set up help browsers at this time in the Mma style, one
can get searchable online help by viewing NCBIGDOCUMENT.html with Netscape,
etc. When you are in an NCAlgebra session just keep a web browser open with
NCBIGDOCUMENT.html loaded in. The powerfulsearch features of these browsers
allow you to look up things in the document.
An X in commands, e.g. NCXetc., always means that this command is experimental and we
reserve the right to change it.