Do you know what are the best methods in MATLAB for optimizing the geometry of Nano-photonics, please?

I am interested in doing optimization of geometry for nanophotonic devices using MATLAB, I have seen several artificial intelligence techniques that are existing in MATLAB ToolBox. I am looking for some methods that I can use in MATLAB for optimizing the shape of nano-photonic devices.

Can you advise me, please?

[Walter Roberson]

Are you measuring "best" according to execution time? So if one method required 20000 lines of code but was 0.28 seconds faster than the one that required only 50 lines of code, you would consider the faster one to be "best" ?

Or are you measuring "best" according to how "understandable" the code is?

Or are you measuring "best" according to how long it takes to develop the code?

Or are you measuring "best" based on time-to-completion, including time required to develop the code and time required to execute the code?

Or are you measuring "best" according to lowest energy requirement, or lowest heat dissipation, or smallest planar geometry, or ... ?

[John D'Errico]

The best method for anything is the one you understand, the one you know how to use, the one you can use at all. Then you learn other methods, expand your capabilities.

I'd just say to start to learn MATLAB, and don't worry too much about the optimal solution for your specific universe of problems. Optimal does not exist anyway in this sense. As has already been pointed out, there are too many dimensions of what is good.

[Walter Roberson]

When people ask about the "best" way to do anything, I hesitate to offer advice. What if my meaning of "best" is different than their meaning? What if there is an analysis tool commonly known to people who work in that field but which I hae not happened to have reason to encounter, so my proposals are only "decent enough" but results could be improved with only a few hundred hours of specialized reading? What if this is an example of a task that belongs to a class of tasks that there is no general solution for, but it turns out that there is an obscure research paper showing that using some known but obscure techniques, this particular case can be solved? What if it is an example of a situation like Fermat's Last Theorem that is considered very very very difficult but turns out to be within reach of an extrodinary genius now alive, if only they had 7 years to dedicate to it? What if it is an example of a situation where that extrodinary genius is not yet alive and will not be alive for a few thousand years, so my proposals are far from as good as the small furry creatures from Alpha Centuri will eventually come up with? If there turns out to be a "better" solution possible if only the entire intellect of humanity turned to doing nothing other than solving the problem (or growing food to make it possible for everyone else to eat while solving the problem) -- wouldn't I have blundered badly by only offering you "as good as I can do" instead of the best ??

When people ask for "best" it is safer for the volunteers to sit back and say nothing.

[Manal Shlebik]

Hi everyone:

Thank you for your responce,

I am measuring the smallest planar geometry for the nanodevice.

[Walter Roberson]

Well, sure I could spend a number of months reading all the published papers on the topic to try to figure out the best way -- but what about the advances since then that are available through the pre-print servers? And what about the advances that are too new to have made it to the pre-print servers?

Something is either the best or it is not the best. If I had a program that could execute for 30 seconds and give you a result that was 99.99% as good as the theoretical possible, then that is not suitable for your purpose, and you would rather have the program that took 400 years to get 99.999% as good as the theoretical possible, or have the program that took "only" a billion years to get to 99.9999% as good as the theoretical possible.

I mostly feel as if when someone asks for the "best" that there is nothing I can do for them. (The rest of the time I might happen to know a reason why it is mathematically impossible to come up with a "best" solution.)

Answer 1

[Dinesh]

Hi Manal,

I understand that you are trying to optimize the geometry of Nano-photonics. You can use the following methods for the optimization. All these algorithms are already a part of optimization toolbox.

1. Genetic Algorithm : A heuristic optimization method based on natural selection and evolution.
2. swarm intelligence: Particle Swarm and Ant colony optimization techniques
3. Simulated Annealing: A stochastic optimization method that mimcs how annealing process is done

Hope this helps,

Thank you!

[Walter Roberson]

GA and Particle Swarm and Ant Colony are examples of a field sometimes called Evolutionary Optimization. There are File Exchange contributions to implement other Evolutionary Optimization strategies. For example, perhaps it would help in this case to use Shark Smell Optimization, or Honey Badger Optimization, or Chaos Bird Swarm Optimization. (Those all exist.)

[Manal Shlebik]

Dear Dinesh

Many thanks for your reply. Your answer is helpful and your help is appreciated.

All the best