Re: An running issue in eABF-replica calculations

From: Jérôme Hénin (heninj_at_gmail.com)
Date: Wed Mar 17 2021 - 08:52:57 CDT

On Tue, 9 Mar 2021 at 00:26, René Hafner TUK <hamburge_at_physik.uni-kl.de>
wrote:

> Thanks a lot Jérôme!
>
> To understand this correctly:
>
> * This means that when using MW-ABF the walkers are benefitting from each
> other (as gradient counts are shared, well reflected in the files)
>
> Yes.

> * but in case of MW-eABF sharing of eABF data is not (and was never?)
> shared (as it was not reflected in the files before and therefore also not
> internally?!) or do/did the walkers still benefit from each other?
>
> MW-eABF is not that different from ABF, it's just that the ABF quantities
(mainly the biasing force itself) pertain to the extended coordinate. Those
are shared exactly as in standard mwABF, because they are represented the
same way in the implementation.
The data for the *unbiased free energy estimators* is a separate question:
it is not shared in the current state of the code, and in a sense it does
not need to be shared during the simulation, as these estimators are really
a matter of post-processing. Anything that is done on the fly is just for
convenience.

    * Then latest literature would be a bit misleading when stating usage
> of MW-eABF.
>
> Is there something else than the QM/MM tutorial that mentions it?

Best,
Jerome

> On 3/8/2021 11:41 PM, Jérôme Hénin wrote:
>
> Dear Hua Hao and René,
>
> I was not aware of that use of eABF in multiple-walker form in the
> tutorial. That is not a documented/supported feature, and there is a little
> work to properly combine the data.
> Anyway, I have just pushed a commit that partly reverts the previous one.
> What remains is that ABF files are written after sharing, so the plain ABF
> data should be the same for all replicas.
>
> Best regards,
> Jérôme
>
> On Mon, 1 Mar 2021 at 16:28, hua hao <haohua0116_at_gmail.com> wrote:
>
>> Thanks, Abhishek.
>>
>> I try to do it, but the problem is still present.
>>
>> Abhishek Acharya <abhi117acharya_at_gmail.com>于2021年3月1日 周一上午8:50写道:
>>
>>> Hi,
>>>
>>> Maybe you can try something like this.
>>> mpirun -n 28 namd2 +replicas 14 eABF.tcl +stdout output_eABF/%d/job00.log
>>>
>>> Abhi
>>>
>>>
>>> <https://urldefense.com/v3/__https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail__;!!DZ3fjg!vtVzXY1Bh5MPmObB6F2Kwfd0GmfBZ6lSqQpfcqGBhlh4BSHjQYhNoJJNi11_LlyCCw$> Virus-free.
>>> https://urldefense.com/v3/__http://www.avast.com__;!!DZ3fjg!sevlfjQ6JGJuVU9XpERk00JW3ppf3VaHSmqjskAT9ZlEVUI_HsrREErAxJBGQwPcvg$
>>> <https://urldefense.com/v3/__https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail__;!!DZ3fjg!vtVzXY1Bh5MPmObB6F2Kwfd0GmfBZ6lSqQpfcqGBhlh4BSHjQYhNoJJNi11_LlyCCw$>
>>>
>>> On Mon, Jan 4, 2021 at 6:02 PM hua hao <haohua0116_at_gmail.com> wrote:
>>>
>>>> Hi NAMD developers and users,
>>>>
>>>> I am learning Tutorial_QMMM_String_eABF recently, which was introduced
>>>> to be an advanced tutorial for QMMM in NAMD page (
>>>> http://www.ks.uiuc.edu/Research/qmmm/).
>>>>
>>>> When running the eABF job by "mpirun -n 28 namd2 +replicas 14 eABF.tcl
>>>> +stdout output_eABF/%1d/job00.%1d.log", it is strange that some necessary
>>>> files with the suffix .grad, .count, .pmf, .zcount, and .zgrad were only
>>>> present in the first replica (i..e 0), but not in other replicas (i.e.
>>>> 1-13). NAMD-2.14 and NAMD-Gib-2020-12-07 were used to perform the above
>>>> calculations, and these packages were both complied using openmpi-3.1.4
>>>> build on Intel-icc. We also tried the pre-compiled Linux-x86_64-netlrts
>>>> and Linux-x86_64-verbs, the error was still present.
>>>>
>>>> Normally, these files should appear in each replica.
>>>>
>>>> Could you help me with this issue? Thank you very much.
>>>>
>>>> *Hua Hao*
>>>>
>>>> --
> --
> Dipl.-Phys. René Hafner
> TU Kaiserslautern
> Germany
>
>

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:11 CST