The Future of Optical Fibre

An anonymous reader writes "An Australian researcher has come up with a novel way of developing optical fibres. Steven Manos, a researcher at the Optical Fibre Technology Centre in Sydney, Australia has developed a method of using genetic algorithims for discovering optimal designs of optical fibres. An article on his work had this to say "The problem with designing optical fibres is starting with a specific set of criteria and then coming up with a design to fit this. The computer program developed by Manos, which is run on supercomputers, does this by mimicking the process of evolution. The computer program combines two patterns to create a third fibre 'offspring', which Manos described as "similar but a bit different". This process is repeated thousands of times with the 10 designs best suited for the particular application chosen to 'breed' again." Another case of "When in doubt, use brute force"?"

Comments

Brute force? Not exactly

[haluness]

I'd rather not think of the method as brute force. Ok, its not like a design from first principles, but its still way to search the parameter space without having to test all coimbinations of parameters

Re:Brute force? Not exactly

[neilmoore67]

I'd rather not think of the method as brute force.

Well said. Brute force would be enumerating every possible optical fibre and then testing them.

This method is more subtle and converges to a close-to-optimal solution with less computer power having to be applied.

Re:Brute force? Not exactly

[russellh]

Ok, its not like a design from first principles, but its still way to search the parameter space without having to test all coimbinations of parameters

So-called first principles are explanation, not design tools. In other words - guess what? - nature is still surprising even if it can be "explained" by what we already know. We can explain stuff. It's the construction that we don't understand so well.

Re:Brute force? Not exactly

[N Monkey]

I'd rather not think of the method as brute force.
I'll agree with that. Brute force searching would go though all the parameters a la ..

for(parameter1 = min limit ...)
for(parameter2 = min limit ...)
for(parameter3... )
etc....
Evaluate(parameter1, param2, ....)

Genetic algorithms try to limit the search space by starting with "probably good" sets of parameter values and trying to generate other "probably good but hopefully better" parameter combinations.

It won't necessarily find the absolute best set of parameters but it might find some reasonable ones.

Re:Brute force? Not exactly

[ObsessiveMathsFreak]

I think as this method becomes more popular it will displace the older method of finding the most mathematically perfect solution and designing from that.

In other word instead of fudging designs while you wait ten years for a mathematician to find the equations for the perfect wing, you just get a computer program to 'evolve' one for you. I'll bet this is what boeing and airbus already do.

Sadly this will leave most applied physics mathematicians out of a job. Danm computers!! Taking our jobs and our women!!

Re:Brute force? Not exactly

[essreenim]

I agree. Brute force on its own is inefficient and predominantly useless.
Even at tasks it can be applied to - like key cracking etc, it is still practiacally uselless without a bit of intelligence build in.
Even the best prime generators dont doa brute iteration through all integers > 0 to infinity - that would be pointless. You need to know where to look as well, or your waisting your time.
I hope as computers continue to advance we dont forget this and simply rely on computing power. Because no matter how far computing powere progresses, we still need to use our own brains with it, not have them replaced alltogether, if we did that we wouldn't get anywhere ...

PDFs from Manos

[antic]

There are some interesting PDFs of papers co-written by Steven Manos available including these two:

• Novel fibre Bragg grating design using multiobjective evolutionary algorithms
• Optical Fibre Design with Evolutionary Strategies: Computational implementation and results

I'm not going to pretend that I know exactly what's going on, but the first of those two is worth looking at if you have even a passing interest. The second looks to be a little more towards the "deep end".

Re:Question?

[PingKing]

It is specifically referring to the fabrication of fibre itself.

Optical Fibre Technology Centre:
http://www.oftc.usyd.edu.au/?section=fibre

Much Better

[irokie]

This is a much better example of the application of Genetic Algorithms than the story that was on slashdot the other day (can't find a link, the one about Formula One racing).

in this case they have a very specific set of criteria.

it didn't however mention in the article how they're testing the designs (did it?)...
and are they actually manufacturing any of the designs that have come from thiss yet?

Only on slashdot...

[Mateito]

.. would breeding be regarded as "brute forcing". :)

What the...

[eddy]

Another case of "When in doubt, use brute force"?

Evolutionary search isn't "brute force", you id... At least not for meaningful definitions of 'brute force'

Brute force would be starting at one end of design space and evaluating each design in turn.

GA's are not brute force

[jdrugo]

..as they don't search the state space exhaustively. Going through all possible combinations of parameters would be brute force, but in this case, as the parameters are real-valued, this is even impossible (if ignoring the possibility of quantisation)

Evolutionary Algorithms provide informed search as they perform competition among the individuals (each representing one possible solution) in the population. Their performance is way above exhaustive search techniques (which _are_ brute force) but below classical search techniques. In this case, however, such classical techniques cannot be applied as the problem space is not well-defined.

No, Taco, No

[neoshroom]

Another case of "When in doubt, use brute force"?"

No, Taco, No.

From the 'brute force' entry in Wikipedia:

In computer science, Brute Force, sometimes called the Naive Method, is a term used to refer to the simplest, most intuitive, most spontaneous, and usually most inefficient methods of accomplishing a task.

This is exactly what a genetic algorhthem is not. If you have a million numbers brute force would be to go from the first to the last in order. Using a genetic algorhythem provides a shortcut though Design Space wherein you need to try far fewer combinations in order to come to a successful result.

C'mon Taco, of all people, you should know this!

An Australian Genetic algorithm?

[Timesprout]

With all the weirdo animals the Australian continent has produced I guess this program will produce some highly interesting results. I cant wait for the announcement that a pattern resembling a Duck Billed Platypus is ideal for streaming Digital TV.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Slow, haunting music plays... incessantly

[unithom]

I am... Torgo..; I .. polish the. .. fiber while the Master... is away... There is no way... out of.. here, the fiber will.. go dark ... soon, there is no way... out of .. here... etc, etc.

Brute force? No way?

[carldot67]

Genetic algorithms are computational shortcuts that are used to very quickly find minima in complex multiparameter functions.

Suppose you wanted to find the lowest value of f(x)=sin(x) where x is from 0-360. (OK we all know its at x=270 but hear me out) - you can do it a couple of ways:

1. calculate sin(x) for all 360 possible values of "x" or
2. calculate sin(x) for (say) 20 values of "x".

Statistics says approach 2 will give you a couple of promising results, for only 1/18th of the effort. Now "breed" another 20 from the 6 values of x for which sin(x) were lowest, say 190, 210, 212, 260, 278, 290. This "next generation" gives sin(x) values whiach are closer to zero. Take the best 6 again. After three generations you are *close* to finding the values for "x" that give you sin(x)=0.

So systematic examination takes 360 tries and the genetic shortcut takes 60 tries - about 17% of the computational effort.

Now imagine a function a bit more complex; some mad multivariate affair like the wave equation. Each variable becomes a "gene" in the above "breeding program". All the time we are looking for parents and offspring that *tend* towards the answer we are looking for. (We also chuck in some unrelated parents too, since inbreeding can be bad - a tip stolen from Monte Carlo techniques [which see]).
The computational savings from GA, GP and MC techniques are potentially huge (as in orders of magnitude) so long as you dont care that:

a) The answer is not 100% exact
b) Some alternative minima are missed

Re:Brute force? No way?

[saigon_from_europe]

But only problem, just as parent said, is that

b) Some alternative minima are missed

A friend of mine got a job to work on genetic algorithms, in an academic institute. Being an engineer, he asked to be moved to another position 3 months after.

His explanation was very short: in GA you look for problems and you try to prove that they could be solved by this method.

Actually, all methods used in engineering were invented to solve some problem; not vice versa. Ok, maybe they not all of them, i canno say for sure since we learned about methods, not about their history. Most succesful methods moved from one area to another one, but basicaly it was problem before solution, not solution before problem.

The interesting point is that this is second story about GA in only couple of days; there must be some conspiracy theory to explain this.

Re:Brute force? No way?

[ortholattice]

...so long as you dont care that... b) Some alternative minima are missed

The problem of local minima is often significant. A good analogy is real genetics - each species has evolved into a "local minima" for likelihood of extinction. If the wings on a given type of butterfly become slightly larger or smaller there will typically be a survivability penalty of some kind, and wing size has stabilized at the optimum for that species. But look at the difference in possible local minima: in one case it results in a whale, and in another it results in toenail fungus. Neither could survive if suddenly given some of the characteristics of the other. A beautiful orchid is a local minimum, and so is pond slime. Your genetic algorithm could decide that pond slime is the optimal product, and the difference can impact your ornamental plant business due to subjective things like beauty that can't be quantified.

Re:First cars...

[agilen]

Well, genetic algorithms are optimization algorithms. Any problem that is non-deterministic, as long as it can be defined with a "genotype", can probably be optimized with a genetic algorithm.

One thing they get used for in academia is designing robots. Its very hard to teach a robot to do something like walking, and the optimal solution depends on so many factors that its hard for humans to hard-code the behavior. But set up the proper simulated environment on a computer, and have a genetic algorithm whose fitness function depends on the robots walking across a room, and you'll see some pretty amazing things...

Genetic Algorithm + Hill climbing

[G4from128k]

For exploring real-valued phase spaces, one solution is to combine a GA with a classical hill-climber. A hill-climber evaluates the local gradient (the partial derivatives of fitness with respect to the independent variables) and then makes a directed adjustment of the solution in the direction of better performance. Hillclimbers can reach optima in floating-point spaces very quickly, but tend to get stuck on local solutions.

GAs are great for jumping out of local optima to find new realms of the solution space, but don't converge as quickly on the neighborhood optima. So the combination of a GA with more classical optimzation can work well.

Re:Genetic Algorithm + Hill climbing

[namidim]

I would argue that though GAs are better than simple hill climbing they are far from being "Great" at getting out of local optima. In fact, a lot of the theory surrounding GAs has to do with how to avoid exactly that since the basic GA of some mutation combined with splicing for reproduction tends to get stuck extremely easily being limited almost entirely to the values represented in the original population. Evolutionary algorithms might be a better choice to pair with hill climbers. Evolutionary algorithms use vector cross products and gausian probabalistic mutation as the main operations for reproduction and my uderstanding is that they tend to get much better coverage of the search space than GAs. For instance a big problem in standargd GAs is that you get stuck with various combinations of whatever values were in the original population (only mutation can put in new values and mutated "genes" tend not to stick around for very long in the later stages of the evolution) . In EAs where most of the reproduction is based on probabalistic calculations however, you get a whole gradiant of values that change all the time.

this guy is way too confident

[kjba]

No other algorithm can come up with a design for optical fibres that are cheap to make and transmit data at a high rate, Manos said.

How can anyone make a claim like this? Just the fact that one can't think of any other algorithm doesn't mean no such algorithm exists. For many problems that can be solved by genetic algorithms, other (problem-specific) algorithms exists (or may exists) that are way more efficient. The nice thing about genetic algorithms is that it is a standard tool that often works, not that it is an exceptionally smart way of doing things.

Re:this guy is way too confident

[geeber]

Well, actually lot's of algorithms exist for designing optical fiber, and they do it efficiently and very accurately. I use a number of in house proprietary programs for designing optical fibers all the time. And I can tell you we don't waste time messing around with GA's

So why don't you hear a great deal about such algorithms? Well, for one, they don't have cool names like "Genetic Algorithms". Also, they are highly prized and considered extremely valuable intellectual property for the companies that actually make optical fiber. We are not going to publicise all the details the most fundamental design tools of our business.

GA's are not the future of optical fiber. They are, however, excellent for generating academic papers, which in turn are highly useful for getting tenure.

Re:this guy is way too confident

[evangellydonut]

it's just a matter of time before academic papers are generated by GAs, and whoever think of it first will have a endless supply of funding at the expense of other guys :-P

Re:this guy is way too confident

[mankei]

I think you are being too critical to their research. Granted I don't like the idea of using genetic algorithm for optics research either, but what they design are not conventional fibers, but holey fibers, (a.k.a. photonic crystal fibers or microstructure fibers) which can have varying hole patterns and sizes in the cladding, making fiber design much more complicated. These holey fibers are unique because they can have much higher effective nonlinearity (smaller core size) and unique dispersion properties (e.g. anomalous group-velocity dispersion at 800nm), and I believe there is no existing technique or program that tells you how to design those hole patterns to get desired dispersion properties.

What's so great about meat?

[szquirrel]

This process is repeated thousands of times with the 10 designs best suited for the particular application chosen to 'breed' again." Another case of "When in doubt, use brute force"?

More like another case of computer science being fascinated by meat.

Remember when neural networks were the next big thing? Everyone was applying them to everything, whether or not it made sense to solve the problem that way. It's neural! Just like our brains! Our brains are smart, they will make our computers smart!

I'm sure genetic algorithms will eke out a useful place in the computer science toolkit, I just doubt it will be as broad as the current fashion of applying them to everything from optical fiber to race cars to compilers.

It had to be said.

[Anonymous Coward]

Manos... the hands of bitrate.

Genetic Algorithms are so cool

[shaka999]

I've written my open GA for doing circuit optimizations and it works very well. The thing I love most is that they are so simple to write. There are things you can do to speed up convergence but the basic algorithm is very straight forward.

The difficult thing is how to score individual trials. I don't know how many times I've checked things after a overnight run and found that my results aren't what I expected. Pretty much everytime this comes down to how I've scored a trial. Just remember you get what you ask for.

For a circuit example, suppose I ask for a certain power comsumption and speed but I overstate the speed goal. Because I'm so far off the speed goal the power will largely be ignored. There are easy ways to tweak this but the point is...again...you get what you ask for.