Estimagic 0.0.22 has huge performance improvements in the reparametrizations. This lowers the runtime of reparametrize_from_internal from 0.4 seconds or almost 80 % on the CHS example model to 2 miliseconds (less than 2 percent).
Adjustments needed
[x] implement the start_params_helpers that used to be part of estimagic in skillmodels
Further improvements
[x] Fix the numba deprecation warning in the translog function
closes #44 closes #33
Why update
Estimagic
0.0.22
has huge performance improvements in the reparametrizations. This lowers the runtime ofreparametrize_from_internal
from 0.4 seconds or almost 80 % on the CHS example model to 2 miliseconds (less than 2 percent).Adjustments needed
Further improvements
parse_params
fasterrobust_cholesky
Speed improvements
Before estimagic update
After estimagic update, old parse_params
After estimagic update with faster parse_params