Hi,
RELION proceeds to a finer angular sampling when the resolution does not improve
AND hidden variables (e.g. angles) do not change much. See "no gain for X iter" messages
in the log. If the current angular sampling is fine enough (finer than 75% of the
minimum angular sampling required for the current resolution), RELION
declares convergence instead of going to even finer samplings.
It is normal for refinement jobs to fluctuate a bit and the resolution
to wobble one or two shells. This often helps overcoming local minima.
However, in some rare cases the refinement keeps wobbling and performs
many futile iterations. In this case, you can stop the job and continue
it with "--force_converge". However, there is always a risk that you
terminate too early.
I would keep it going if this is the final refinement aimed for publication.
Otherwise (e.g. before CtfRefine, just testing masks, etc), I would stop it
since risking a few shells does not really matter.
Best regards,
Takanori Nakane
On 2021/03/11 7:55, Dieter Blaas wrote:
> Hi,
>
> I am wondering whether everything is in order: During a refinement run (box 512 px) I am having the same Healpix order (9), the same image size
> (456) and the same resolution (2.71 A) for 9 iterations (it022 till it030). Is there a problem with convergence or is everything fine? It is still
> running and has not yet reported that it would now combine the two half maps!
>
> Thanks, best, Dieter
>
> ------------------------------------------------------------------------
> Dieter Blaas,
> Max Perutz Laboratories
> Medical University of Vienna,
> Inst. Med. Biochem., Vienna Biocenter (VBC),
> Dr. Bohr Gasse 9/3,
> A-1030 Vienna, Austria,
> Tel: 0043 1 4277 61630,
> Mobile: 0043 699 1942 1659
> e-mail: [log in to unmask]
> ------------------------------------------------------------------------
>
> ########################################################################
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCPEM&A=1
>
> This message was issued to members of www.jiscmail.ac.uk/CCPEM, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at
> https://www.jiscmail.ac.uk/policyandsecurity/
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCPEM&A=1
This message was issued to members of www.jiscmail.ac.uk/CCPEM, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
|