A key point. Cooperation between the engineers and the computer can be
even more productive. Years ago, when I worked at Bausch & Lomb (in
electronics, not optical engineering) I was told that many of their
breakthrough products (Cinemascope, stay-in-focus zoom lenses, etc)
were the result of engineers being able to use the computer to develop
designs which would otherwise have taken inordinate amounts of time.
Engineers were suddenly free to explore ideas which would previously
not have been feasible to research.
This is why I'm interested in more and more powerful computing platforms.
It's all very well being able to watch Lara's almost inhumanly larges
breasts rendered in realtime, but it's not exactly useful.
It's often trivial to describe the model for a problem such that it's
something along the lines of:
make something which meets these criteria, as well as is possible.
For most applications this is producing a function which tests how
'good' a given guess is with some optional contraints. The obvious
progression is to shove it into a computer which selects guesses which
'optimise' this function. There are of course piles of other
'optimisation' problems, and algorithms which attempt to solve them,
about selecting good or near optimal solutions for different kinds of
problems.
Such a function is often trivial to produce compared to solving the
problem containing it. The algorithms which solve these problems are
often expensive to compute without a large amount of mathematical
analysis to make it more computable.
One thing that is clear: the faster computers get and the faster the
algorithms, the number of problems which can be explored increases. The
faster, the more variables can be handle, better solutions found or even
(*gasp*) whole new problems which could never be solved well suddenly
become solvable.
It's a shame that most of the computing power in the world is actually
spent on almost useless triviality. We're not short of processing power,
but we're definately short on co-ordinating it effectively!
Ian Woods
