[RMCProfile-users] [EXTERNAL] Run time while using RMCProfile

[Zhang, Yuanpeng]

Hi Sumit,

For sure if you have problems setting up linux version running on cluster, just let me know.

Concerning the way to tell whether we have a good fit, I would stay with the obvious one you mentioned, i.e. the chi^2. But I do have something to say about it. Currently, we have already implemented a way to automatically adjust weights for various datasets involved in the fit, on the fly. You can track the chi^2 data for all datasets in the file, for example, stem_name-00.log to see how chi^2 is performing for all datasets simultaneously. You can find more information about this in the manual (if you go to the appendix of latest version manual, you will find it). Any concerns about this, let me know.

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<mailto:zhangy3 at ornl.gov> | Web: https://www.iris2020.net<https://www.iris2020.net/>

From: Sumit Khadka <sumitkhadka08 at gmail.com>
Date: Friday, September 18, 2020 at 10:46 AM
To: "Zhang, Yuanpeng" <zhangy3 at ornl.gov>
Cc: "rmcprofile-users at shadow.nd.rl.ac.uk" <rmcprofile-users at shadow.nd.rl.ac.uk>, "rmcprofile-users at rmcprofile.org" <rmcprofile-users at rmcprofile.org>
Subject: Re: [EXTERNAL] [RMCProfile-users] Run time while using RMCProfile

Hi Dr Zhang,

Thank you. I will definitely look into the linux version and email you if I have any further questions on that.
Definitely waiting for the school, hope they do some online version if possible.
So, in terms of recognizing a good fit, what different parameters can we inspect to get an idea of a good fit, the obvious one being the chi^2 values.

Regards,

Sumit Khadka
University of Houston, Physics



On Fri, Sep 18, 2020 at 8:01 AM Zhang, Yuanpeng <zhangy3 at ornl.gov<mailto:zhangy3 at ornl.gov>> wrote:
Hi Sumit,

First, I think the running time to produce a reasonable RMC fit really depends on system and machine. So I think the best way to tell whether a good fit can be obtained in a reasonable time (say, within an hour) is to have a try on a machine and inspect the result frequently to get a feel about how much time is really needed.

Second, we do have Linux version of the program so you can grab it from sourceforge page: https://sourceforge.net/projects/rmcprofile/files/Version_6.7.8/. To run it on cluster, we may need a bit steps for which we don't have instructions yet (will produce, you actually remind me of that. Thank you!). But for now, if you are interested in running it on cluster and have problems figuring out how-to, please let me know!

For the school, usually ORNL will hold a summer school on total scattering every year which involves RMCProfile. But nowadays because of the special situation, I have no idea about what will happen. We will send out news about it if we do have updates on the school.

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<mailto:zhangy3 at ornl.gov> | Web: https://www.iris2020.net <https://www.iris2020.net/>

On 9/18/20, 3:57 AM, "rmcprofile-users on behalf of Sumit Khadka via rmcprofile-users" <rmcprofile-users-bounces at shadow.nd.rl.ac.uk<mailto:rmcprofile-users-bounces at shadow.nd.rl.ac.uk> on behalf of rmcprofile-users at shadow.nd.rl.ac.uk<mailto:rmcprofile-users at shadow.nd.rl.ac.uk>> wrote:

    Hi developers & RMCProfile Users,

    I have been trying to use RMCProfile for PDF data on disordered 2D
    materials. One obvious problem while using RMCProfile is the run time
    needed to produce any significant result, like in the tutorial of
    refinement of SF6 it has been mentioned that about 20 hrs is required to
    produce a useful model or fit. Are there ways that this run time could be
    decreased or at least have the program run in a cluster or something?

    Also, are you planning on organizing a school for RMCProfile anytime soon?

    Regards,

    Sumit Khadka
    University of Houston, Physics
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <https://lists.rmcprofile.org/pipermail/rmcprofile-users/attachments/20200917/11a5f0e1/attachment.htm>
    _______________________________________________
    rmcprofile-users mailing list
    rmcprofile-users at rmcprofile.org<mailto:rmcprofile-users at rmcprofile.org>
    http://lists.rmcprofile.org/mailman/listinfo/rmcprofile-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.rmcprofile.org/pipermail/rmcprofile-users/attachments/20200918/476c96d8/attachment-0001.htm>