getdist error

Use of Cobaya. camb, CLASS, cosmomc, compilers, etc.
Post Reply
Sheng Li
Posts: 57
Joined: May 26 2009
Affiliation: University of Sussex
Contact:

getdist error

Post by Sheng Li » November 13 2013

I encountered now this error when I was trying to get the dist for parameters:
reading chains/test_1.txt
reading chains/test_2.txt
reading chains/test_3.txt
reading chains/test_4.txt
Number of chains used = 4

MKL ERROR: Parameter 5 was incorrect on entry to SSYEV
Error in Matrix_Diagonalize
For this issue, I would like to think about the possible reason from the wmap9 data which was used in the cosmomc2012 version.
How I updated WMAP9

I just updated the Makefile under the source folder for cosmomc2012, by changing the pattern "wmap...7yr" to "wmap...9yr".
Similar Error in this Forum

I searched this forum, I found two cases referring to this kind of issue. But they gave me no hint to cure my issues.

Search results in this forum

http://cosmocoffee.info/viewtopic.php?t ... ntry+ssyev
http://cosmocoffee.info/viewtopic.php?t ... ntry+ssyev
Search on Google

From Google search, I noticed one solution is to update the compiler (ifort) then this issue disappeared.
But I don't know if it can help for my question here. Since I used the ifort 2011 version, and there was NO this error before, say about half year ago.
This is the new ERROR

This error just is new after I updated the WMAP7 to WMAP9, so I suspect there was something wrong in link or use of new data set.

WMAP9 Setting?

If this is the reason, can anyone advice me how to incorporate the WMAP9 data with the old version, say 2012, CosmoMC?
Thanks,

Sheng Li
Posts: 57
Joined: May 26 2009
Affiliation: University of Sussex
Contact:

[SOLVED] getdist error

Post by Sheng Li » November 19 2013

This issue is SOLVED!
The antidote is

Code: Select all

#if zero, columnnum calculated automatically as total number of columns 
columnnum  = 41
which is set to zero by default. If anyone would get the similar problem, you need to set this number explicitly for the output data which is output in form of matrix.

But I don't investigate what causes the failure I mentioned here.
The previous version of my code can automatically handle and get the correct value for getdist working.

But this issue has now been solved, here is the backtracking process.
I spotted the subroutine *SSYEV* which has *NINE* parameters, and that the *fifth* param is the \_DIMENSION_ of the symmetry matrix whose eigenvalue is going to be evaluated which must be determined.

Done.

Post Reply