hi! I am trying to replicate sw2003 in dynare .when i run the bayesian estimation ,i find that using different mode_compute have different results.mode_compute=6 and 7 is ok. but others is not very satisfying , matrix is not positive definite. why is that happen data.mat (5.2 KB)
sw_euro.mod (5.2 KB)
Mode-finding is hard and it is not uncommon for mode-finders to get stuck at local optima.
1 Like