💬 Using fixed spectal vector as initial guess for optimization

[ankit7540]

Thank you for the project. Very useful !

By specifying the number of kinetic decays we specify how many spectra vectors the input data matrix is deomposed into. SVD provides the initial spectral vectors. This works fine.

I wonder if one can supply one or more of the spectral vectors before optimization. If such vector(s) can be supplied, can they be fixed during optimization.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[jsnel]

Our apologies for overlooking this issue, we definitely need to do some work on maintaining our issues tracking, we'll try to do better in 2026!

---

To try to answer your question, if I understand it correctly, yes I think what you are asking is possible. There are 2 ways in which (known) spectral shapes can dictate or guide your fitting.

First, there is such a thing as a spectral model where the input is a (known) spectral shape (or a parametrization thereof) and then the kinetic traces are estimated as the conditionally linear parameters. If you have high confidence in your spectral shapes (which you may, given that you want to 'fix' them) this might be an interesting angle to explore.

There's also something like 'guidance spectra', this is where you postulate a kinetic model with free form spectra, but you penalize the delta between the estimated spectra and some known and/or target shape by adding a delta to the residuals.
A simple illustrative example of this you can find in our "examples" repository, which you can find here, specifically these lines in the model.

For a more advanced use case, we used this in one of the case studies explained in the (open access) launch paper:
Pyglotaran: a lego-like Python framework for global and target analysis of time-resolved spectra
(search for the word "guidance" there)
And note that you can find the code for that in the supplementary information found here.

If you could clarify a bit more what you exactly wish to achieve (or even provide a sample notebook) we can point you in the right direction.

Thank you for your patience!

[ankit7540]

Hello,
Thank you for the response. This is an open case and highly relevant for my analysis.
Give me a few days to prepare a notebook for this. :-)