When attempting to write an optimiser, potential contenders find that it is very
easy to get the waste down from say 15% to 10%, somewhat harder to get the waste
down to 6% and virtually impossible to get the waste down consistently to around
3% or even less. This is because when an optimiser takes a less than perfect route
in the calculation, it is usually replaced by the results of another less than perfect
route, so all evidence of the first imperfection is immediately lost. The author
may therefore continue under the delusion that the results are in the first division
when usually they are in a much lower league.
The precise way in which different optimisers work varies quite considerably, which
is why it is a fallacy to believe that any optimiser produces optimum results. In
reality, the results differ widely and Trimloss gets usually better answers by the
inclusion of human ingenuity at many stages in the calculations. This turns it from
a pure trial and error program, into an intelligent thinking program. This gives
a much better balance between processing time and results, enabling Trimloss more
time to explore different and often beneficial options. In technical terms, these
processes can be described as an advanced form of mathematics called decision theory.
Decision theory is now being exploited in many ways to make sense of the oceans
of data available in the modern computerised world. We have however, been utilising
this branch of mathematics in Trimloss, for the last 35 years.
What some optimisers also try to do is produce the lowest possible waste on the first
cutting diagram. They often do this by using up far too many of the smaller sizes
which are best left in reserve until the end of the run to help avoid very high waste
on the final few diagrams. Trimloss is clever enough, using decision theory, to
take all these points into account, so it sometimes deliberately chooses a very slightly
inferior early cutting diagram, to yield much better results towards the end of the
batch of orders. This always gives a lower overall waste.
One day in 1965, I was ordered to fly with this visiting top brass officer, Air Commodore
Coward. I was strapped in the right hand seat of an RAF T21 trainer, awaiting my
passenger. When he arrived from his own aircraft, a Dehavilland Devon, he apologised
for having to pull rank on me because he needed to sit in the right hand seat, because
of his gammy leg. Only recently have I learned that he lost his left leg to an Me
109 after his Spitfire guns had jammed in an encounter at the start of the Battle
of Britain. I knew nothing of this at the time and looking back, I now treasure
the fact that I did not ask him if he knew how to fly. He assumed he was to fly
the aircraft, even though I was told that I was to be at the controls. I need not
have worried because his flying was very good and we did not encounter an Me 109.
I now like to think that some of the esteem of this great man rubbed off on me and
helped me in no small way in the daunting task of writing the best glass optimiser
in the world.