Simulation takes a long time to finish

[pm1357]

Hi,
I am trying to simulate a more complex optical system with multiple diffractive surfaces and while the simulation works, it takes a long time to compute. Since I don't need the graphics for my problem, only data about illumination in the detector at the end I was wondering whether there is any way to speed up the simulation (running the software locally, other solutions?). I would also appreciate suggestions about other software I could use (anything similar to Ray Optics, but maybe without such extensive graphics, preferably in Python)

[ricktu288]

This project supports calling the core simulator directly from Python without graphics rendering.
See the Integration Tools for more information.

[pm1357]

Hello, I have another question regarding the Integration tools option. I managed to get the simulation to work (via python), however if I try to max out the Ray Density option (in the JSON file, I found that 20 is the highest value in web app), it will not work (it will just crash and give me an error. Do you maybe happen to know why this is happening?

I was also wondering what the 'image mode density' option in JSON file means - I haven't found anything about it in the documentation.

[ricktu288]

This project is not designed to handle a huge amount of rays. The proper way to do that is to use GPU, but that requires rewriting the core completely. Also see #138. If you still want to use this project to simulate a large amount of rays, you may instead randomly sample rays and take the average of the detector readings. For the Beam tool there is a "random" option.

The "ray density" does not corresponds to a single property in the JSON. When "mode" is either "rays" or "extended", the property "rayModeDensity" is used, otherwise ("images or "observer") the "imageModeDensity" is used, and is documented in https://phydemo.app/ray-optics/docs/Scene.html
This behavior is to allow new users of this app (who may not be familiar with the "ray density" slider) to switch between different mode (views) while maintaining fairly good visualizations.

[pm1357]

Thank you for a fast reply and the information!

[pm1357]

Hello, I have another question regarding the integration tools, I would like to rewrite the core ("rayOptics.js") a bit (I would like to run it on a computer cluster and would like to parallelize it somehow, haven't figured out yet how to do that or if it is possible with your core) and I was wondering whether there is any documentation written about this core (is it just a compacted version of some other code that is documented or something else completely...), thank you for your time and help in advance!

[ricktu288]

The rayOptics.js is built from the source https://github.com/ricktu288/ray-optics/tree/master/src/core via Webpack. See the core/index.js part of https://phydemo.app/ray-optics/docs/index.html for documentation.

Note that it is not easy to make it parallelize, but as a workaround, if your bottleneck is to generate a huge amount of rays, you may just randomly sample rays on each computational node, and then combine the final detector readings, so that you don't need to modify anything in the core.