Dear Dynare Users,
have someone be able to use the “use_dll” option?
I have followed the steps described here: dynare.org/DynareWiki/Config … ompilation
I have installed Microsoft Windows SDK. But when I run “mex -setup” I get the following error message:
“Der Befehl “C:\Program” ist entweder falsch geschrieben oder
konnte nicht gefunden werden.”
I think it has to do with the fact that in the path there is a blank. But I do not know how I can solve this problem.
Check whether you installed the correct version, particularly considering your Matlab version (64 vs. 32 bit): de.mathworks.com/support/compilers/R2015a/index.html
Note also that for older versions of Matlab, the detection might not work.
Okay. I have Matlab R2011a (64bit). Does it mean use_dll will not work? Or what should I do?
(In general: I want to use the dll option, hoping that it will speed up the computation. My model has 71 equation. Can I expect that it will run faster? I solve with order 3. Or should I use dynare++? But when I use dynare++ on a very simple model, it is also very slow for higher order. What is the advantage of using dynare++?)
Again many thanks in advance!!
(I also have another question: I’m doing the following steps:
- run the dynare mod file with a simulation length of 10000 and order 3.
- Calculate the kurtosis of one simulated variable say output.
- As a check I then use:
[oo_.dr,info,M_,options_,oo_] = resol(0,M_,options_,oo_);
y_ = simult_(oo_.dr.ys,oo_.dr,shocks,options_.order);
shocks is equal to oo_.exo_simul.
- Form y_ is again calculate the kurtosis of the output.
BUT: the values of the both kurtosis are NOT exactly the same. Why??)
See de.mathworks.com/support/compilers/R2011a/win64.html. It seems you need to install Microsoft Visual C++ 2010 Express as well (and in the correct order)
Regarding your second question: Most probably because of the burnin in options_.drop.