? MJDGGES requires two square real matrices of the same

[lombard]

Hi,

While trying to solve a model I got this (new to me) error message with Dynare 4.0.3.

??? MJDGGES requires two square real matrices of the same dimension.

Error in ==> dr1 at 354
[ss,tt,w,sdim,dr.eigval,info1] = mjdgges(e,d,options_.qz_criterium);

Error in ==> resol at 101
[dr,info,M_,options_,oo_] = dr1(dr,check_flag,M_,options_,oo_);

Error in ==> stoch_simul at 40
[oo_.dr, info] = resol(oo_.steady_state,0);

Error in ==> FF_portfolio at 566
info = stoch_simul(var_list_);

Error in ==> dynare at 102
evalin(‘base’,fname) ;

Solving the same model with a different code I got that the solution is unstable.
Is this a bug?

Best

Gianni

[stepan-a]

Hi Gianni, Did you try to solve your model with the snapshot version? We recently added
a trap when the jacobian matrix is complex…

Best,
Stéphane.

[lombard]

Hi Stephane,

I just did a fresh download of Dynare using svn but the trap does not seem to be there.

Gianni

[stepan-a]

In dr1.m you should have (lines 209 to 217):

    if ~isreal(jacobia_)
        if max(max(abs(imag(jacobia_)))) < 1e-15
            jacobia_ = real(jacobia_);
        else
            info(1) = 6;
            info(2) = sum(sum(imag(jacobia_).^2));
            return
        end
    end

If the jacobian matrix is complex dr1 returns info(1)=6. Can you post (or send) your mod file if you do not fall in the trap ?

Stéphane.

[lombard]

Hi,

I guess that the command

svn co https://www.dynare.org/svn/dynare

did not get me the recent snapshot.

Where do I get it using svn (if at all)?

Thanks

Gianni

[stepan-a]

Hi, The snapshot version is in the trunk subdirectory. You may follow these steps :

svn co https://www.dynare.org/svn/dynare/trunk
cd ./dynare/trunk
./configure
make

The third step is new. The configuration script checks that your PC has all the needed libraries and writes the MakeFiles used to build the preprocessor and the documentation (you will find more information here).

Best,
Stéphane.