Optimization for parameter fitting of an EVP model

Next month, the 16th International Conference of IACMAG was supposed to be held. Sadly, the COVID-19 situation has forced the organisation to postpone the conference to May 2021. I was supposed to present my paper there on the application of local and global optimization algorithms for parameter fitting of 1D EVP modelling. I have posted a preprint on researchgate and made part of the code available here.

As part of my work to make consolidation modelling more practical and objective, I have benchmarked a range of optimization algorithms. Once a good (and fast!) model is set up, this can be done quite concisely thanks to the lmfit package. To ensure optimal performance, it is really important to get some insight about the shape of your (expected) loss function such as the simple example below.