GitLab

interrupted by non finite result

simulations: draw random shocks so that, given a seed, the sequence is the same, independently of the number of periods

for current needs)

-pass along errors returned by cycle reduction algorithms;
-logarithmic reduction fails on error
-replace expensive and random condest() by call to linsolve()
-uses linsolve() to compute hx

dynamic systems

options_.gpu (no interface yet, use options_.gpu=1). Added test
./tests/first_order/fs2000_cr.mod

dyn_first_order_solver.m in global_initialization.m This is necessary
if noclearall option is used.

and singular coefficient matrix for current variables

Fixed random bug of MC identification test, occurring when there is no solution at the very first iteration of the MC loop.
(cherry picked from commit 906bf8e0)

fixing bug in evaluating rank condition for BK when using check (introduced in commit 4d51f38b)

Fixed bug (karaba:/var/tmp/dynare-master-check-WMwywKtbcz.log). dr.state_var is not defined in all cases...

Added a new routine to solve quadratic matrix equation (based on a Newton algorithm with line search).

i.e. don't fail when they happen and instead return 0 or Inf

Fixed bug (reported in karaba:/var/tmp/dynare-master-check-o0FCvBRHnf.log). dr.state_var was defined in the wrong place.

Added walsh.mod in the testsuite to compare the solutions obtained with the cyclic reduction algorithm and the generalized schur algorithm.

Comparison of the solutions returned by the cyclic reduction and logarithmic reduction algorithm. Note
that in the case considered here the logarithmic reduction algorithm is faster.

This algorithm is a slower alternative to the cyclic reduction algorithm (useful for testing purpose).