Mathematics, MATLAB Version 7.2 (R2006a) Release Notes

Mathematics, MATLAB Version 7.2 (R2006a)

New features and changes introduced in this version are

• New Library CHOLMOD for Sparse Cholesky Factorization

• New Solver for State-Dependent DDEs

• Upgrade to BLAS Libraries

• New Function for Integer Division

• New Input to gallery Function

• Improved Algorithm for expm

• More Efficient condest for Sparse Matrices

• accumarray Accepts Cell Vector Input

New Library CHOLMOD for Sparse Cholesky Factorization

For sparse matrices, MATLAB now uses CHOLMOD version 1.0 to compute the Cholesky factor. CHOLMOD is a set of routines offering improved performance in factorizing sparse symmetric positive definite matrices. See the function reference pages for chol, spparms, and mldivide for more information on how CHOLMOD is used by MATLAB.

New Solver for State-Dependent DDEs

In this release, MATLAB provides a second solver function, ddesd, in addition to the existing dde23 function, for delay differential equations (DDEs). This new solver is for use on equations that have general delays. You supply a function in the input argument list that returns the vector of delays to be used by the solver. See the function reference page for ddesd, and DDEs in the MATLAB Mathematics documentation for more information.

Upgrade to BLAS Libraries

MATLAB now uses new versions of the Basic Linear Algebra Subroutine (BLAS) libraries. For Intel processors on Windows and Linux platforms, MATLAB supports the Math Kernel Library (MKL) version 8.0.1. For AMD processors on Linux platforms, MATLAB uses the AMD Core Math Library (ACML) version 2.7.

New Function for Integer Division

The new idivide function provides division similar to A./B on integers except that fractional quotients are rounded to integers according to a specified rounding mode.

New Input to gallery Function

The gallery function has a new, optional input argument called classname. The classname input is a quoted string that must be either 'single' or 'double'. When you specify a classname argument in the call to gallery, MATLAB produces a matrix of that class.

Improved Algorithm for expm

The expm function now uses an improved algorithm to compute a matrix exponential. This algorithm often requires fewer matrix multiplications.

More Efficient condest for Sparse Matrices

The condest function handles sparse matrices more efficiently when estimating a 1-norm condition number.

accumarray Accepts Cell Vector Input

The accumarray function now accepts a cell vector as the subs input. This vector can have one or more elements, each element a vector of positive integers. All the vectors must have the same length. In this case, subs is treated as if the vectors formed columns of an index matrix.

Desktop Tools and Development Environment, MATLAB Version 7.2 (R2006a)

Data Analysis, MATLAB Version 7.2 (R2006a)