Janos Gabler
Janos Gabler
Now that respy is much more flexible, we still want to be able to reproduce the classical Keane Wolpin models easily. Therefore I propose to rewrite the specification helpers. They...
This enhancement was suggested by [Johannes Schmieder](https://sites.google.com/site/johannesschmieder/?pli=1) ### Current Situation The default sampling method in multistart optimization is a non-randomized sobol sampling. This is unexpected for users and a problem...
## Current Situation We use [`fuzzywuzzy`](https://github.com/seatgeek/fuzzywuzzy) for proposals in error messages (e.g. when an invalid algorithm was requested). Unfortunately, it seems unmaintained. ## Alternatives - [thefuzz](https://github.com/seatgeek/thefuzz) by the same author...
### Problem In empirical applications we often have weakly identified parameters. This can have two sources: 1. The criterion function for the estimation problem is flat 2. Two or more...
### Desired Situation - We can calculate the second derivative of vector and scalar valued functions that take parameters as numpy arrays or DataFrames - The three standard hessian formulae...
### Terminology Assume we have a criterion function `f(params)` and its derivative (gradient or jacobian) `d`. To enforce constraints, the internal optimizer sees a reparametrized version `g(x) = f(r(params))`. By...
Pandas has a [nice way](https://github.com/pandas-dev/pandas/blob/5f96f2868f6dc2a96c7154003c0e217173d8c050/pandas/compat/_optional.py#L64) of handling imports of optional dependencies. We should check if this can make our code less repetitive.
Add type hints to estimagic. Look [here](https://sethmlarson.dev/blog/2021-10-18/tests-arent-enough-case-study-after-adding-types-to-urllib3) for tips.
## Problem description We have some local optimizers that can deal with some level of noise (pounders, pybobyqa and dfols) but they fail on very noisy problems such as our...