[RMCProfile-users] Unusual BROADENING_CORRECTION :: bug?

[Zhang, Yuanpeng]

Hi Summer, Paul and Community,

I was playing with the issue reported by Summer and Paul concerning the Qbroad-like correction in RMC.

I did find a bug with my implementation and hopefully I get it right this time (see the link below for new version of the package to try out). I want to note down a bit here about the issue.

https://sourceforge.net/projects/rmcprofile/files/Version_6.7.8/

In RMC calculation, all atomic distances are thrown into their own bins, e.g. with bin size 0.02 angstrom. Within a certain bin, a simple number counting will be conducted and finally we will have a histogram across all bins. Therefore, each r-point in RMC-calculated PDF pattern is actually representing a certain bin but not a single point. Saying that, we do not distinguish atomic distances falling into a certain bin and that further means any broadening effect with width smaller than the bin size being used won't make sense in RMC. So with this round of bug fixing, when the input width for the Gaussian broadening function is detected to be smaller than the bin size, we will force it to be identical to the bin size.

As for the r- and Q- limit issues mentioned by Summer and Paul, I am not actually seeing any problems with the limit values. Maybe when you did your fitting you enabled 'FITTED_SCALE', which makes you feel the limit value isn't right? If you still feel uncomfortable with those limit values, just leave me a message and we can talk offline.

Cheers
Yuanpeng

-- 
 
Dr. Yuanpeng Zhang
Computational Instrument Scientist
Spallation Neutron Source
 
Oak Ridge National Laboratory
1 bethel valley road, Oak Ridge, TN 37830, US
Mobile: +1(865)-335-4818
Email: zhangy3 at ornl.gov | Web: https://www.iris2020.net <https://www.iris2020.net/>

On 10/28/20, 11:03 AM, "rmcprofile-users on behalf of SUMMER, ARON (PGR) via rmcprofile-users" <rmcprofile-users-bounces at shadow.nd.rl.ac.uk on behalf of rmcprofile-users at shadow.nd.rl.ac.uk> wrote:

    Hi,

    Just reporting what I believe to be a bug with the "BROADENING_CORRECTION ::" function.

    I've recently upgraded to RMCProfile v 6.7.8 in the hopes that the BROADENING_CORRECTION could solve some slight peak broadness mismatches I was experiencing. While including this term I noticed some unusual behaviour of the "baseline" of the RMC fit when using values below 0.01. In the G(r) fit the low r region of the baseline had a negative curve to it which abruptly truncates at my minimum atom-atom distances, this curve is present in the initial system and after having been run for a period of time (although hidden under broad peaks, the cut-off is still very visible in the G(r).

    I have checked the same system without the BROADENING_CORRECTION keyword (to check if it's something else in 6.7.8) and with a value above 0.01 and this feature is only present when BROADENING_CORRECTION is set to below 0.01 in my experience.

    Racked my mind on why a peak broadening function would change an absolute baseline and came up blank, so figured it was probably a bug in the code.

    Please let me know if you want me to send across my setup files.

    Kind regards,
    Aron Summer
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <https://lists.rmcprofile.org/pipermail/rmcprofile-users/attachments/20201028/7b7bf4e5/attachment.htm>
    _______________________________________________
    rmcprofile-users mailing list
    rmcprofile-users at rmcprofile.org
    http://lists.rmcprofile.org/mailman/listinfo/rmcprofile-users