A question about PTV(peak-to-valley) default merit function

[AhuRanger1027]

Dear Optiland Community,

I am very new to optical design and is currently learning Optiland as my first optical design tool. Since most of the courses use Zemax, I learn the course via reproducing the tasks and examples of Zemax in Optiland.
It worked out well in the first couple of episodes. I worked through most tasks, and thanks to the nicely-written codes of Optiland, I can see the optical design process in a crystal-clear manner.

All went well, until I encountered PTV(peak-to-valley ratio).
In Zemax, PTV (peak-to-valley-ratio) is one of the two types of default merit function (the other being RMS). I tried to implement it in Optiland myself, It then raises two questions:
1 I could not find on the internet the algorithm used for PTV. ChatGPT tells me that the algorithms are propertiary and the details are not known to the public. Is it truly that the algorithms widely used in Zemax are actually black boxes to the users?
2 Is there a similar function or work-around in Optiland or in the realm of optical design?

I apologize for these questions not being appropriate Optiland questions.
Thank you very much for your kind help!

[HarrisonKramer]

Hi @AhuRanger1027,

Welcome to the world of optical design.

First off, yes, Zemax is closed-source, so you will not be able to actually inspect the internal workings of the code directly. However, you might find answers in its docs, provided you have access to this.

Peak-to-valley in optimization simply means that the optimizer minimizes the maximum extent of the operands, or calculated metrics. As far as I know, this is much less common for lens design. This type of optimization is not available in Optiland, but it could be added. That would entail modifying the OptimizationProblem class to calculate the PTV function and then you'd need to update OptimizerGeneric to handle an argument for the optimization type, e.g., "rss" or "ptv".

Hope that helps.

Kramer

[AhuRanger1027]

Dear Kramer,

Thank you very much for your kind reply! It is more than helpful!

Thank you for solving my frustration about Zemax, and I am still surprised by this fact. I sincerely wish Optiland to become the main-stream optical design tool, so that everyone can see clearly what happens in their optimization 😄 .

PTV is a little awkward since it is not a directly differentiable quantity. In Zemax, it is likely that an indirect (but differentiable) operand is used to represent PTV, since the Ansys website indicates " This merit function type attempts to minimize the PTV extent of the errors. However, the numerical value of the merit function is not the actual peak-to-valley error."[*]. I don't know if "max - min" could be directly optimized via Optiland, but yes, I should give it a try and see what happens!
PTV is indeed a very rarely used operand, since many optical engineers I asked did even know its existence 🥲 .

Thank you again for your kind help!
Best Wishes,

Zhe

*: https://ansyshelp.ansys.com/public/account/secured?returnurl=/Views/Secured/Zemax/v251/en/OpticStudio_User_Guide/OpticStudio_Help/topics/Optimization_Function_Types.html

[AhuRanger1027]

Dear Kramer,

PTV works!
I have just used "max - min" as the return value of the operand, and let Optiland handles the rest. The result was more than reasonable and was comparable with the result from the course examples.
Thank you again for your help and your excellent work on this fantastic optical design tool!

Best Wishes,

Zhe