search

gmdsi / GMDSI_notebooks

python-based predictive groundwater modeling workflow examples
GNU General Public License v3.0
49 stars 31 forks source link

inclusion of opt tools #24

Closed jtwhite79 closed 1 year ago

jtwhite79 commented 2 years ago

@rhugman I've been pondering how the opt tools are included in these notebooks and Im wondering about setting up a separate set of notebooks dedicated just to constrained optimization concepts. Currently in the class the Randy,@mnfienen and I teach, they are kinda tacked-on at the end but now with mou, and soon sqp, there will be a lot of potential content to be covered and digested. So Im wondering about leaving them out of the current effort and instead building up more notebooks in the uq/da sphere. Was thinking about a hypothesis testing notebook using da as the engine and maybe some "demystifying" or "intro to" notebooks that cover the underlying concepts related to ensemble methods (like manually calculate a jco from ensemble results)? Thoughts on this? @mnfienen - thoughts on this? Can we list out some other possible topics to cover - here is the draft website just as a reminder...

https://gmdsi.org/education/decision-support-modelling-with-pest-and-pyemu/

rhugman commented 2 years ago

@jtwhite79 - OK. I do think that there is a need for notebooks on optimisation concepts. However, if you think that it is too much for the current effort, fair enough (it's already looking like it will take more time than planned for...). What's your idea? Leave them for a hypothetical "second round" of tutorial-making?

Re intro to ensembles and hypothesis testing- yes, 100%. Something like an intro to da might be worthwhile as well given the novelty?

On a related side note, do you think it is feasible to condense Part1 down to "intro to's..."? I have not thought this through.

(btw - the site and outline can change entirely if desired; it is a placeholder)

jtwhite79 commented 2 years ago

I agree re ouu - i think it's the next thing to do after uq/da. It just needs its own intro notebooks I think. We can still bring over the existing opt notebook and put it out there. I just think ouu deserves more content. I'll work on a hypothesis testing notebook (the truth setup stuff is close...)