#21: Add partial information ---------------------------+------------------------------------------------ Reporter: sebastien | Owner: george Type: feature | Status: accepted Priority: major | Milestone: 4.2 Component: Core M-files | Version: Resolution: | Keywords: ---------------------------+------------------------------------------------
Comment(by sebastien):
Replying to [comment:3 sebastien]:
Two minor issues to be solved (among others):
- the file matlab/disclyap_fast.m has no copyright: this should be
clarified
Done in a5b88195189eb694b2c8fd2cae65ed29818ee673
- the files matlab/partial_information/{qzswitch,qzdiv}.m are redundant
with the versions in matlab/qz. These files should be deleted once the PI code relies on mjdgges.
Files removed in 44b268ec6742f0500b6627c029f7653b0a572e13.
As of a5b88195189eb694b2c8fd2cae65ed29818ee673, the code does not give the same results under Octave and MATLAB.
Hi Sebastien
I have not yet looked into this issue but I am starting to - when do you need it (and whole Dynare 4.2 code) fixed by for the release? Is the next week ok or do you want it by end of this week (tomorrow?)
Basically, PI does not and cannot rely on, or use mjdgges.
Also, as far as the PI manual page is concerned, can you just add the core usage info from the PI wiki page to the manual on the lines of what you suggested? When I designed teh wiki I had the manual pagin mind.
PS: The current use of PI varobs was discussed and this is the way Michel and Joe wanted it.
Best regards George
On 11/10/2010 16:55, Trac Dynare wrote:
#21: Add partial information ---------------------------+------------------------------------------------ Reporter: sebastien | Owner: george Type: feature | Status: accepted Priority: major | Milestone: 4.2 Component: Core M-files | Version: Resolution: | Keywords: ---------------------------+------------------------------------------------
Comment(by sebastien):
Replying to [comment:3 sebastien]:
Two minor issues to be solved (among others):
- the file matlab/disclyap_fast.m has no copyright: this should be
clarified
Done in a5b88195189eb694b2c8fd2cae65ed29818ee673
- the files matlab/partial_information/{qzswitch,qzdiv}.m are redundant
with the versions in matlab/qz. These files should be deleted once the PI code relies on mjdgges.
Files removed in 44b268ec6742f0500b6627c029f7653b0a572e13.
As of a5b88195189eb694b2c8fd2cae65ed29818ee673, the code does not give the same results under Octave and MATLAB.
George Perendia george@perendia.orangehome.co.uk writes:
I have not yet looked into this issue but I am starting to - when do you need it (and whole Dynare 4.2 code) fixed by for the release? Is the next week ok or do you want it by end of this week (tomorrow?)
Obviously we can’t release PI in Dynare 4.2 before it is fixed under Octave. So as soon as possible is obviously the best.
Basically, PI does not and cannot rely on, or use mjdgges.
No problem.
Also, as far as the PI manual page is concerned, can you just add the core usage info from the PI wiki page to the manual on the lines of what you suggested? When I designed teh wiki I had the manual pagin mind.
Ok I will do that.
PS: The current use of PI varobs was discussed and this is the way Michel and Joe wanted it.
Ok.
Concerning the QZ discrepancy between MATLAB and Octave, are you using the qz() primitive ? Because this primitive does not behave the same way under Octave and MATLAB. The convention for Q is different under Octave (We have AA = Q*A*Z under Matlab and AA = Q'*A*Z under Octave). And Octave is only able to compute real QZ, while MATLAB can do both complex and real, and does complex by default. See the respective documentation for more details.
Best,
-
George Perendia -
Sébastien Villemot -
Trac Dynare