D
DavidLee
Guest
Not exactly like evolution. A lot like rapid trial and error. Calling an algorithm genetic does not make it so.The Barbarian said:After running the genetic algorithm scenarios, the computer selects the best performer from a group of trials, much like natural selection in biological evolution, and combine characteristics from that engine with those of other high performers.
Sounds a lot like CAD doesn't it. Let's exchange the word "genes" with the word "parameters" and see if it fits... Yes, it does fit. Quite well in fact.The Barbarian said:The engines with these "genes" are simulated using the same CFD and visualization process as in past studies. Senecal then rates the engines on their fuel efficiency and the amount of soot and nitrate wastes they generate.
By far the greatest decrease in emissions came from the removal of sulphur from on-highway fuel.The Barbarian said:Convergent Thinking's findings have helped optimize engine design for increased efficiency and lower emissions - two things of great importance to engine manufacturers and environmentalists.
Diesel engine are already very efficient. Particulate matter makes them look bad.
That's not new. The reason for electronic injectors is so that the patterns and volumes of fuel can be varied quickly and efficiently.The Barbarian said:"We can now indicate to designers the variables that are most important or ones that might have been overlooked had the computer not identified them," says Senecal. The computational studies, for example, have highlighted the importance of injecting fuel in short bursts instead of a single stream, increasing the surface area of the fuel and leading to a cleaner and more efficient burning.
You have not shown any sort of natural selection. To ignore the evidence you posted, and even to say it proves your statement is at best wishful thinking.The Barbarian said:So natural selection works better than design for complex problems.
I believe the researcher or article publisher may not know, but the developer of the algorithm can usually explain it. After all, someone wrote the algorithm. You might even say someone intelligent designed it. Well, you wouldn't, but others might.The Barbarian said:Solutions in genetic algorithms aren't designed at all. Often the researcher doesn't even know why they work.
When you say "natural selection" the phrase you should use is "trial and error". CAD is not evolution or natural selection.The Barbarian said:That's why they are more efficient; they are free to vary in any way at all, using natural selection to identify the best solutions.
No. What you have done is present random facts and called them evidence of your conclusion.The Barbarian said:Darn, I presented evidence, again. They hate that.