Re: multiple replicas metadynamics

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Tue Jan 10 2017 - 15:40:43 CST

On Tue, Jan 10, 2017 at 3:57 PM, Mihaela Drenscko <quo.physics_at_gmail.com>
wrote:

> Hi,
>
> I was running NAMD 2.12 on my desktop, with different outputName for each
> replica.
>

Good. What are the results of this test?

>
> So if the contents are the same in each pmf replica files (I checked),
> that means, at the end of simulations, I will not have one pm file made of
> combining all replicas, but 10 identical pmf files . So any of these
> identical files can be taken to represent "the file *outputName*.pmf
> contains the combined PMF from all replicas", where *outputName* can be
> any of the 1, 2, 3, ... 10?
>
> I must be absolutely sure this is correct, because in the website
> description
>
> http://www.ks.uiuc.edu/Research/namd/2.10/ug/node58.html
>
> the singular case "file" is used, not "files", as in 10 identical pmf
> files from 10 different replicas.
>

Of course it's singular, because outputName has only one value at a time
and outputName.pmf is the result of *one* run of NAMD. The entire user's
guide describes the input and output of one NAMD run at a time. If it
wasn't phrased like that, you'd also need to make plural pretty much
everything else.

>
> Thank you,
> Mihaela
>

So far I replied mostly to benefit other readers of the mailing list, since
you started with the report of a bug, which is generally useful (although
it was fixed 2 years ago already).

But you are now arguing your way out of running a simple test to make sure
you understand the manual. What's worse, it's not the first time for you
(see e.g. your recent messages to the LAMMPS mailing list). Since you
consider your own time too valuable to do your part, I'll act accordingly.
Good bye and good luck.

Giacomo

>
> On Tue, Jan 10, 2017 at 12:37 PM, Giacomo Fiorin <giacomo.fiorin_at_gmail.com
> > wrote:
>
>> Hi Mihaela, I'm a bit surprised that you are asking whether the value of
>> outputName should be the same between the replicas in NAMD 2.12 vs 2.10.
>> First of all, the user's guides for 2.12 and 2.10 are both available.
>> Secondly, you probably realize that setting outputName the same for all
>> replicas can be risky: if you run them in the same folder, they would
>> overwrite each other's files. If you run them in different folders, they
>> may have files with the same name but different full path (and the registry
>> file keeps the entire path).
>>
>> The short answer: you will have different outputName.pmf files, but the
>> contents will be the same in each file (you can easily check). The partial
>> PMF option is there for diagnostic purposes, and is disabled by default: if
>> the meaning of the option is not clear to you, why enable it?
>>
>> If you're still in doubt, just try running NAMD 2.12 on your desktop.
>> Since the communication for the metadynamics multiple-replicas option is
>> done through files, you can run many single-processor NAMD jobs and they
>> will communicate with each other through the registry file. You do not
>> need to run an expensive simulation, and actually the smaller the better:
>> you can definitely afford to run 4 replicas of deca-alanine on your desktop.
>>
>> Lastly, you do not need admin privileges to compile scientific software
>> in your cluster, and it is actually the job of the scientist who uses it to
>> do that. Of course, it is always possible to ask a colleague for a favor,
>> coerce or bribe the sysadmins into doing it, or hire somebody else with a
>> PhD for the job, but it is not required. There are just as many successful
>> labs that have a lab manager, as many that don't.
>>
>> Giacomo
>>
>> On Tue, Jan 10, 2017 at 6:57 AM, quo.physics_at_gmail.com <
>> quo.physics_at_gmail.com> wrote:
>>
>>> Hi,
>>> I will update, but it cannot run this lenghty 10 replicas simulations on
>>> a regular computer. And what is updated on a supercomputer is for other
>>> people to decide.
>>>
>>> Anyway, can you please tell me, with the new version, the outputName for
>>> each replica is different or is the same? If it is different, does this
>>> mean that named will ascribe a name for the combined PMT file, or do I have
>>> to do that myself? It is not indicated in the description in which way this
>>> procedure is accomplished.
>>>
>>> Thank you,
>>> Mihaela
>>> Sent from my Verizon 4G LTE Smartphone
>>>
>>> ------ Original message------
>>> *From: *Ajasja Ljubetič
>>> *Date: *Tue, Jan 10, 2017 12:55 AM
>>> *To: *namd-l;Mihaela Drenscko;
>>> *Cc: *
>>> *Subject:*Re: namd-l: multiple replicas metadynamics
>>>
>>> Wh! y not si mply update
>>> <https://www-s.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD> your
>>> NAMD version (2.12)? You will get bugfixes and better performance. The
>>> updates are free and the installation almost trivial.
>>>
>>> Best,
>>> Ajasja
>>>
>>> On 9 January 2017 at 23:20, Mihaela Drenscko <quo.physics_at_gmail.com>
>>> wrote:
>>>
>>>> Hi,
>>>> By the way, the namd version I was using so far was 2.10.So for the
>>>> multiple replica metadynamics, when naming all replicas outputs with one
>>>> common name, say r (instead of! giving individual names as 1, 2, 3...10): I
>>>> get these errors:
>>>> ERROR: Error on renaming file r.restart.coor to r.restart.coor.old: No
>>>> such file or directory
>>>> FATAL ERROR: Unable to open binary file r.restart.coor: File exists
>>>>
>>>> So if I give the same name to each replica, I get this error; if I give
>>>> different names, the 10 pmfs (for 10 replicas) don't combine to one pmf
>>>> file. What, then is the way out?
>>>>
>>>> Thank you,
>>>> Mihaela
>>>>
>>>>
>>>>
>>>> On Sun, Jan 8, 2017 at 6:37 PM, Giacomo Fiorin <
>>>> giacomo.fiorin_at_gmail.com> wrote:
>>>>
>>>>> < div dir="ltr">
>>>>> Hi Mihaela, this is a bug fixed a while ago.
>>>>>
>>>>> Please update to the latest NAMD.
>>>>>
>>>>> Giacomo
>>>>>
>>>>> On Fri, Jan 6, 2017 at 9:24 PM, Mihaela Drenscko <
>>>>> quo.physics_at_gmail.com> wrote:
>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> When I run namd with multiple replicas metadynamics, I don't get
>>>>>> files outputwq.colvars.name.replicaID.state and outputwq.colvars.name
>>>>>> .replicaId.hills, where outputw! q is the name of each outputName.
>>>>>> Instead, the paths of these file are written in a file called
>>>>>> metadynamics1.outputwq.files, text:
>>>>>> /.colvars.metadynamics1.1.state
>>>>>> /.colvars.metadynamics1.1.hills
>>>>>> These files should read outputwq.metadynamics1.1.state and
>>>>>> outputwq.metadynamics1.1.hills, according to:
>>>>>>
>>>>>> http://www.ks.uiuc.edu/Research/namd/cvs/ug/node58.html#sec:
>>>>>> colvarbias_meta
>>>>>>
>>>>>> So these files can't be read by other replicas and pmfs can't be
>>>>>> combine into one. All i get are partila pmf, from each replica.
>>>>>>
>>>>>> Any suggestions, what is wrong? Why .state and .hills files are not
>>>>>> written with outputName pr! efixes, without the need foe their paths to be
>>>>>> written in afil! e?
>>>>>> Thank you,
>>>>>> Mihaela
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Giacomo Fiorin
>>>>> Associate Professor of Research
>>>>> Institute for Computational Molecular Science (ICMS)
>>>>> College of Science and Technology, Temple University
>>>>> 1925 North 12th Street (035-07), Room 704D
>>>>> Philadelphia, PA 19122-1801
>>>>> Phone: +1-215-204-4213 <%28215%29%20204-4213>
>>>>>
>>>>> Scholar: http://goo.gl/Q3TBQU
>>>>> Personal: http://giacomofiorin.github.io/
>>>>> Lab page: https://icms.cst.temple.edu/members.html
>>>>>
>>>>> *"As computer programmers we have a responsibility to make sure thatwe
>>>>> run the computers instead of the computers running us."* - Steve
>>>>> Oualline
>>>>>
>>>>>
>>>>>
>>>
>>>
>>
>>
>> --
>> Giacomo Fiorin
>> Associate Professor of Research
>> Institute for Computational Molecular Science (ICMS)
>> College of Science and Technology, Temple University
>> 1925 North 12th Street (035-07), Room 704D
>> Philadelphia, PA 19122-1801
>> Phone: +1-215-204-4213 <%28215%29%20204-4213>
>>
>> Scholar: http://goo.gl/Q3TBQU
>> Personal: http://giacomofiorin.github.io/
>> Lab page: https://icms.cst.temple.edu/members.html
>>
>> *"As computer programmers we have a responsibility to make sure that we
>> run the computers instead of the computers running us."* - Steve
>> Oualline
>>
>>
>>
>

-- 
Giacomo Fiorin
Associate Professor of Research
Institute for Computational Molecular Science (ICMS)
College of Science and Technology, Temple University
1925 North 12th Street (035-07), Room 704D
Philadelphia, PA 19122-1801
Phone: +1-215-204-4213 <(215)%20204-4213>
Scholar: http://goo.gl/Q3TBQU
Personal: http://giacomofiorin.github.io/
Lab page: https://icms.cst.temple.edu/members.html
*"As computer programmers we have a responsibility to make sure that we run
the computers instead of the computers running us."*  - Steve Oualline

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:58 CST