### MathWork's math tool requires advanced programming skill

There are two kinds of advanced math programs for PCs. One kind has an

interactive, graphical interface with a programming language hidden underneath. The other

kind is mostly just a programming language.

MatLab falls into the second category. It's easy for mathematicians, engineers and

programmers to learn, but hard for nontechnical users. Although you can solve problems

interactively at the command line, MatLab's strength is its powerful, C-like language.

MatLab is short for MatrixLab. From the first page of the documentation, the matrix

clearly is the problem-solving paradigm.

A matrix containing a single number is called a scalar. A matrix with a row or column

of numbers is a vector. Nearly every mathematical problem can be represented by matrices

of some sort.

If you work in electrical engineering or systems simulation, MatLab is a good tool for

you. Most MatLab users routinely handle large programs. For midsized problems, MatLab has

a compiler, but users more often will program, make trial runs within MatLab, then export

the C code they've written to a more sophisticated compiler and perhaps a more powerful

computer.

Though it's largely a programming environment, MatLab isn't stuck in the 1970s. Its 3-D

graphics output tools now equal any competitor's.

One of the hardest tasks for most engineers is creating models to test their designs.

That often involves solving differential equations over and over. The powerful SimuLink

add-on makes this easy for either linear or nonlinear systems sampled continuously,

periodically or both.

Unlike the basic MatLab, SimuLink is graphical. Its drag-and-drop methods combine

symbols in the working window.

Other add-ons and toolboxes hide MatLab's complexity by overlaying special tools and

interfaces for financial and communications calculations, fuzzy logic, signal analysis,

linear matrices and neural networks.

This is the premier math package for solving mechanical, electronic or other

engineering problems on PCs, Unix systems and Apple Macintoshes. Although the basic

package has no drag-and-drop capability, its target audience grew up with Fortran and

looks on C as big improvement.

Given that orientation, MatLab's primitive interface could be a plus for the

experienced engineer, because it cuts the learning curve to almost nothing.

I found I could begin programming in MatLab almost immediately after learning a few

language basics.

Don't buy MatLab if you are uncomfortable with programming. But if your work is with

systems simulation and you write lots of C-type programs, MatLab will not only be easy to

learn, it will speed up your program development dramatically.

If you solve an occasional symbolic math problem, there is no need to buy a separate

symbolic analysis package.

The $595 Symbolic Toolbox add-on is actually the Maple V analysis engine from Waterloo

Maple Inc. of Waterloo, Ontario.

John McCormick, a free-lance writer and computer consultant, has been working with

computers since the early 1960s.