Output error for many user defined prms

aacorrao

Hi,

I am refining many (200 - 400) user defined parameters, with an example of one prm below:

prm anatase_5015_lor 0.01250`_1.33147_LIMIT_MIN_0 min 0 max 10

While out_prm_vals_on_convergence provides the refined prms and values, I would like the errors. From browsing the tech ref and forum I have seen solutions such as individually outputting these prms (P. Metz), but this is a bit clunky for me with so many prms.

An attempted workaround was to define the error as an individual parameter:

prm anatase_5015_lor_err = Get_Prm_Error(anatase_5015_lor);: 1.33147`_0.00000

Using the dependents filter as shown below does output the parameters, but the values are all 0.

out_prm_vals_on_convergence "anatase_out.csv"
out_prm_vals_dependents_filter "*anatase"

I can parse the out file line by line based on keywords to extract the errors, but wanted to first check that there is not either a simpler solution using TOPAS syntax or something that I am missing / doing wrong in my out lines.

Thanks,
Adam

rowlesmr

Hi Adam

I've only ever output individual prms. That way I can guarantee that each column of values corresponds to a particular parameter.

Matthew

alancoelho

The Error function works in Version 7. For example:

do_errors
out aac.txt
   Out(sc1)
   Out_String(" ")
   Out(Error(sc1))
	
str
   scale sc1  0.00841898398`_3.265e-05
   ...

will produce in the AAC.TXT file the following:

0.00841898398 3.2652143e-05

aacorrao

Thanks for your reply. In Version 7 is there a macro to output error for all prms similar to out_prm_vals_on_convergence or must they be called individually as in your example?

-Adam

alancoelho

I'm afraid there's no function to output all errors at once; its something I have noted for the future.

johnsoevans

I agree that this would be really useful.

It's obvious/patronising, but you can often column edit the parameter names from the correlation matrix quickly. You can use '{{{ ..... '}}} jEdit folds to hide them away in the .inp file if you like.

rowlesmr

Also note that the reported errors are scaled by gof.