search

kmaterna / Strain_2D

Strain rate modeling from GNSS velocity fields
MIT License
44 stars 19 forks source link

New methods, better descriptions #26

Closed jlmaurer closed 1 year ago

jlmaurer commented 1 year ago

Starting a draft PR with new methods and bettter descriptions.

  1. Placeholder for adding a Haines and Holt method
  2. Possibly a placeholder for adding a VELMAP-like method
  3. Updating geostats descriptions and hopefully improving the method using Bayes' Theorem

Fixes #10

jlmaurer commented 1 year ago

@kmaterna I'm getting a funny error related to the I2 calculation when running delaunay (see the tutorial notebook in this PR). Any ideas on whats going on with that?

kmaterna commented 1 year ago

Ah, working on it. Right away I can see that the last of my commits in the branch you're working shows me un-commenting one of the output lines that is probably causing the error.
It think it's a stupid bug. I wrote a single function meant to perform operations on 2D arrays, 1D arrays, and scalars, but probably didn't fully test it. I'll let you know when it's fixed.

kmaterna commented 1 year ago

@jlmaurer fixed with two commits just now. Found three simple typos that had just clearly never been on a testing path (e.g., e2 instead of e2[i], and map_sacle instead of map_scale). Pushed two commits to my main branch if that's okay.

jlmaurer commented 1 year ago

@kmaterna I added some todos to this, mainly because I wanted to very quickly see the weighted residual misfit norm, and seems like that's hard to do at the moment. I also updated the tutorial; I would like to add running a couple more methods and then the comparison script, but wanted to get some of the other issues out there first.

kmaterna commented 1 year ago

@jlmaurer I'll check this out soon when I get a chance and merge. Btw back on main, I added the misfit when write_metrics == 1 (both printed and written), and it was a beautiful idea.