COMPUTATIONAL
PROTOTYPING
an interview with Jacob
K. White
(this is an expanded version of the interview appearing in the printed
issue of RLE at MIT)
2003 May Issue 3
RLE: What is "computational
prototyping?"
White: The notion of prototyping
is pervasive in engineering design; one investigates the viability
of a new idea by constructing a single implementation, a prototype.
Whether the problem is designing an integrated circuit, a micromachined
sensor, a new drug, or an off-shore structure, the cost and time
required to construct prototypes is high enough to discourage comprehensive
design exploration. We refer to the process of substituting computer
models for physical prototypes as computational prototyping. The
promise of using computational prototyping is that the ease of testing
alternative designs will allow designers to examine more radical,
and possibly much more efficient, design alternatives. The challenge
of computational prototyping is in developing accurate modeling
algorithms and techniques which are both flexible and fast enough
to allow designers to examine a wide range of design alternatives.
RLE: How did your early research
in computer aided design of integrated circuits lead to your current
interests?
White:
For my PhD, I worked on numerical techniques for improving the efficiency
of circuit simulation. I was fascinated by the problem of making
numerical techniques effective in a given application, and inspired
by the way in which practice so directly followed mathematical theory
in the case of circuit simulation. By investigating numerical techniques
for other applications, I am hoping to expose students in a variety
of engineering and science disciplines to the beauty of marrying
mathematical theory with engineering practice.
RLE: What are the special challenges
of developing computer aided design tools for mixed signal systems
that seek to combine analog and digital processing?
White: When designing digital
systems alone, one can improve simulation efficiency by exploiting
the existing abstract representations for digital design. For example,
one can simulate the behavior of a network of logic gates without
simulating each individual transistor in each logic gate, and therefore
one can afford to simulate a very complicated sequence of events.
Simulating analog systems usually involves simulating the behavior
of each individual transistor, either because there is no abstract
representation for the circuit of interest, or the abstract representation
does not model important second-order effects. Mixed signal simulation
is difficult because it is necessary to simulate very complicated
sequences of events to test the digital processing, but it is presently
too expensive to simulate the analog system's response to those
complicated sequences.
RLE: How do you see the techniques
that your group is developing being applied to next-generation efforts
to integrate micro-machined devices?
White: The most pressing problem
for designers is finding simulation tools that make it possible
to quickly examine design alternatives. To provide such simulation
speed for designs which combine digital processing, analog circuitry,
and micromachined devices, it is necessary to generate abstract
representations of the analog circuitry and micromachined devices.
Since there is such a wide range of analog circuits and micromachined
devices, we are following a strategy that numerically generates
the abstract representation, by extracting it from either a circuit
description or from a three-dimensional geometry.
RLE: Having just recently completed
your first year as one of RLE's two Associate Directors, what aspect
of your new role has surprised you the most?
White: The rapidly growing
economy of the last decade greatly expanded the economic resources
available at MIT, but I think we have all been surprised by how
much recent economic and political events have changed the situation.
Funding of all kinds is scarcer and what is available is harder
to get. I have been impressed, though perhaps not surprised, by
how RLE investigators have adapted to this difficult situation.
RLE investigators have increased their efforts to secure or expand
funding sources, and many have volunteered space so as to more economically
provide for junior faculty. This heroic effort is taking its toll
on RLE investigators, and is reducing time available for research.
I am surprised at how difficult it is to determine effective strategies
for RLE to help researchers find and raise those funds.
|