View previous topic :: View next topic 
Author 
Message 
Alessandro Di Marco
Joined: 01 Apr 2015 Posts: 15 Affiliation: University of Rome  Tor Vergata

Posted: December 15 2015 


Hi Antony!
I have a little problem when I try to work with the defaults .py.
When I plot the for example the root.pdf or root_3D.pdf from the corrispondent .py I have no problem but when I do the same thing for root_2D I get empty plots while for the triangular plot I get only the marginalization plots along the diagonal, in the rest blank again.
Any suggestion?
Could be a python libraryproblems or some setting in input files!?
Thank you in advance for the response!
Have a nice day,
The Best,
Alessandro 

Back to top 


Antony Lewis
Joined: 23 Sep 2004 Posts: 1351 Affiliation: University of Sussex

Posted: December 15 2015 


Can you view them OK in the getdist GUI? Are these scripts from fortran or python getdist? 

Back to top 


Alessandro Di Marco
Joined: 01 Apr 2015 Posts: 15 Affiliation: University of Rome  Tor Vergata

Posted: December 16 2015 


Hi Antony,
I have produced these scripts by fortran getdist and I think that the problem is connected
with the following: in test_2D_ns_omegach2 file and in other file like test_2D  var1_var2, I get all NaN. Is it right or reasonable?
Thank you! 

Back to top 


Antony Lewis
Joined: 23 Sep 2004 Posts: 1351 Affiliation: University of Sussex

Posted: December 16 2015 


Not really, do you get the same with the python version? (fortran version is on the way out). If so, assuming there are no NaN in your chain files, perhaps you could send the chain files to try to reproduce it. 

Back to top 


Alessandro Di Marco
Joined: 01 Apr 2015 Posts: 15 Affiliation: University of Rome  Tor Vergata

Posted: December 18 2015 


Hi Antony,
I have tried to use the python version of getdist on distparam.ini file and I get the following traceback:
chains/test_1.txt
WARNING:root:outlier fraction 0.101075268817
using 465 rows, 11 parameters; mean weight 1.47096774194, tot weight 684.0
Equiv number of single samples (sum w)/max(w): 76.0
Effective number of weighted samples (sum w)^{2}/sum(w^{2}): 299
Best fit sample log(Like) = 212.365800
Ln(mean 1/like) = 215.476421
mean(Ln(like)) = 213.392138
Ln(mean like) = 213.070706
WARNING:root:fine_bins not large enough to well sample smoothing scale  kd
Calculating plot data...
Traceback (most recent call last):
File "python/GetDist.py", line 326, in <module>
main(args)
File "python/GetDist.py", line 226, in main
plot_3D.append(filterParList(line, 3))
File "python/GetDist.py", line 150, in filterParList
raise Exception('%iD plot has not wrong number of parameters: %s' % (num, pars))
Exception: 3D plot has not wrong number of parameters: ['tau']
Can this be of some help?
Thank you in advance! 

Back to top 


Alessandro Di Marco
Joined: 01 Apr 2015 Posts: 15 Affiliation: University of Rome  Tor Vergata

Posted: December 18 2015 


Hi Antony!
Hi think that I understand the problem related to the empyt bidimensional plot.
In distparams.ini file of the last version of CosmoMc there are not the following
command lines:
num_bins = 100
num_bins_2D=40
smooth_scale_2D =2
So I think that without this options, when I interrupt a given simulation I get 2d blanck as results.
I also understand that the traceback (in python getdist) is only a problem concerning which parameters I fix and which not.
Thank you for your help and patience!
Alessandro 

Back to top 


