Q&A: The MathWorks' Andy Grace on simulation

Andy Grace is vice president of engineering for design tools at The MathWorks. He is responsible for The MathWorks' family of model-based simulation, code-generation, and testing tools. He has more than 15 years of experience in software development and has been involved with a number of The MathWorks' product developments, including Simulink, Real-Time Workshop, Control System Design Toolbox, Optimization Toolbox, and Stateflow. He received his doctorate in computer-aided control-system design from the University of Wales in 1989.

What led you to becoming an engineer?

I loved making little circuits that would flash lights or make sounds when I was a kid. Then, when my school bought a Z80-based computer back in 1980, I became hooked on programming. In graduate school, I got hooked on an early version of Matlab, wrote an add-on product, and that's how I ended up at The MathWorks.

What has been the impact of more powerful PCs?

I think it goes without saying that more powerful computers have been tremendous enablers for engineers. Simulation, numerical optimization, and graphical programming are examples of technologies we work on that are much more widespread now that powerful machines are prevalent. The only bad impact I can think of is that, with these powerful numerical techniques, there is less reliance on theoretical or closed-form approaches—and my PC will be obsolete in six months' time!

Where do users of products such as yours have the greatest misunderstanding?

If you've used Matlab and Simulink before, you know the real benefit is that they give you a playground in which to learn, explore, and innovate. When you enter new markets or address users who've never been exposed to these products, they often want canned functions, such as GUIs, and don't always fully appreciate the power of having two very powerful languages—one textual, Matlab, and one graphical, Simulink—in which to explore ideas.

How has the industry changed since the 1980s and 1990s? Who is now leading the R&D and product-development efforts?

The MathWorks addresses all industries in which you find engineers and scientists, so it's hard to make a general statement. One common theme I've witnessed over the last 10 years is that companies have become more focused on the process. Quality-process-improvement initiatives, such as Six Sigma [a quality initiative that strives for 3.4 defects per million opportunities], are prevalent in many engineering organizations. We have benefited from this approach, because organizations are trying to tie together disparate parts of their organizations, such as R&D and production, and are looking for technologies such as model-based design to reduce cost and improve quality.

What is model-based design?

"Model-based design" is a term we've coined to refer to the process of using models throughout a design process. It starts with the conceptual phase, in which you might try out an idea using a graphical model and simulation. You elaborate the model adding more detail at each design iteration—for example, adding data types as you near the implementation phase for an embedded system. You test at each design iteration using any combination of simulation, rapid-prototyping, hardware-in-the-loop, and model-coverage information. You automate the implementation through code generation, which eliminates hand-coding errors.

Enormous benefit comes by tying together all these steps, so all groups are using the same models and talking the same language. For example, this approach means that you can reuse plant or environment models, such as hardware-in-the-loop testing, that you've developed in the conceptual phase for testing of the implementation. And you eliminate a lot of waste by having an unambiguous executable specification, such as a simulatable model that is, using code generation, also the implementation.

Any comments on the state of engineering and science education?

I believe that engineering education has improved because students and professors have greater access to the tools that they will use in industry. They are much more readily able to put into practice what they've learned in the classroom. With more than 3500 universities worldwide having adopted Matlab and Simulink, we are continuing to see broader use of our products in diverse technical curricula. Students are now entering the workforce with a greater understanding of how to solve today's complex R&D and engineering challenges. I am someone who learns best by discovery, so I have always found tools such as Matlab to be great learning vehicles.

Editor's note: The above is an expanded version of the article that appears in the print edition. The PDF below reproduces the printed version.