Hello Adrian,
Actually, there is one experimental situation that does resemble this issue - that of perturbing an equilibrium system by the sudden introduction of energy into it.
This can be done with a laser induced temperature jumps or volume change induced pressure jumps, for example. This is different than trying to maintain a steady state with a sustained
Temperature gradient. In the above examples, typically used to study chemically reacting systems, the assumption is that the perturbation is small so that the system will relax to the equilibrium state at the new temperature or pressure. This will happen in time scales determined by the transport processes and the chemical reactions that occur in the system. Since the perturbation can be considered weak, then linear response theory is the appropriate theoretical background, and that enables one to compute correlation functions as ensemble averages over an equilibrium distribution. The tricky part is to know how long the simulation needs to be so that the assumptions of linear response theory are appropriate. Since the total relaxation time involves mass transport, one would think that the simulation would have to be quite long. At least the simulation time has to be much longer than the time interval over which the system is perturbed.
The easy situation to interpret is when the entire system absorbs energy all at once - that is the standard weak perturbation jump experiment. Using a dilute system and a highly focused laser beam, one could introduce an energy perturbation on selected molecules in a system (but still quite a few of them at a time), and then watch the system relax. The simulation you are all thinking about is one in which we would have an imaginary laser capable of pinpointing just one molecule or portion of it and heating that part of the system up for a finite time interval (perhaps the entire simulation interval). While I've heard of statistical mechanics developments for steady state systems this type of simulation appears to require new theoretical ground for its interpretation.
I think it's interesting to think about it. Finding a way to achieve it in the simulation environment is a first step towards its investigation.
Cheers, Sergio
-----Original Message-----
From: Adrian Roitberg [mailto:roitberg.qtp.ufl.edu]
Sent: Wednesday, June 22, 2011 11:54 AM
To: AMBER Mailing List
Subject: Re: [AMBER] System temperature
Dear Sergio.
I thought you had already known that this would create a non-equilibrium
system and was asking how to do it anyways.
Since that is not the case, you got the issue correctly described now.
There is no way to have two thermostats at different temperatures in the
same system and keep in in equilibrium. As far as I know there is no
statistical mechanical description for such an ensemble, unless, as you
mention, the two systems are decoupled. I imagine one could try to set
up a weak coupling regime between subsystems and treat the coupling as a
perturbation, but I have no idea what it will look like.
Adrian
On 6/22/11 8:47 PM, Sergio R Aragon wrote:
> Hello Fernando, Carlos and others,
>
> I've been thinking about the idea of having one part of a system a different temperature than another part of the system in a simulation. David Case is correct that just changing the masses of the atoms will not work, for the temperature is a measure of the average kinetic energy, and that depends on both mass and velocities. There is another fundamental problem with the idea, however, because presumably you will not eliminate all coupling of the ligand to the rest of the system. This implies that if you generate a separate thermostat for the ligand part and keep its temperature higher, then energy will be transferred, via the interaction couplings with the rest of the system, to other parts of the molecule and eventually to the solvent. Thus the solvent thermostat will have to be removing energy that gets input into the ligand via its own thermostat. This is not an equilibrium system, but a steady state one in which there is a gradient in temperature in the system. T
he sustained temperature gradient will cause other effects such as mass transport, and it may take a while of simulation before the steady state gets achieved. Then, what kind of an ensemble is suitable for interpreting the data? Some serious thought about the fundamentals appears to be warranted before proceeding with this idea.
>
> Cheers,
>
> Sergio Aragon
> SFSU
>
>
> -----Original Message-----
> From: Carlos Simmerling [mailto:carlos.simmerling.gmail.com]
> Sent: Wednesday, June 22, 2011 10:58 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] System temperature
>
> yes this should be possible, buy setting the # of copies to 1 for the ligand.
>
> 2011/6/22 Fernando Martín García<fmgarcia.cbm.uam.es>:
>> I think it's a better aprox. for our lab. Nevertheless, I've been
>> reading the manual and i have a question. Can LES program give me an
>> approach to my problem, this is, give to the ligand a temperature
>> different to the system one?
>>
>> Fernando
>>
>> On Fri, 17 Jun 2011 19:55:46 -0600, Jason Swails
>> <jason.swails.gmail.com> wrote:
>>> Another idea, though it may require careful thinking about, is to
>>> 'trick'
>>> the system into heating different atoms differently simply by
>>> changing the
>>> masses of certain atoms (for instance, make them heavier if you want
>>> to
>>> lower the temperature of that part, etc.). None of the forces should
>>> depend
>>> on the masses of adjacent particles (the amber force field doesn't
>>> take
>>> gravitational effects into consideration ;)). Of course you'd have
>>> to
>>> calculate by how much to increase (or decrease) the masses to reach
>>> the
>>> desired "target" temperature (or effective temperature).
>>>
>>> Keep in mind, whether you take the T approach as Ross suggested (but
>>> bear in
>>> mind that you may have to adjust thermostat behavior in addition to
>>> setvel
>>> that in dynlib.f, since that will only affect the initial
>>> velocities).
>>>
>>> I may be corrected in this advice if it turns out that I'm missing
>>> something
>>> critical.
>>>
>>> HTH,
>>> Jason
>>>
>>> 2011/6/16 Fernando Martín García<fmgarcia.cbm.uam.es>
>>>
>>>> Thank you for the answer, but i have a question: is that means i
>>>> have
>>>> to create a tempi and tempi_atomselect in my input files to specify
>>>> the
>>>> temperatures?
>>>>
>>>> Thanks
>>>>
>>>> Fer
>>>>
>>>> On Wed, 15 Jun 2011 16:00:04 -0700, "Ross Walker"
>>>> <ross.rosswalker.co.uk> wrote:
>>>>> Hi Fernando,
>>>>>
>>>>> Unfortunately there is no 'automated' way to do this, however, it
>>>>> would not be too difficult to code something in.
>>>>>
>>>>> See the function setvel in dynlib.f. This is called in the case
>>>> when
>>>>> you are not doing a restart. You can call the atommask function
>>>> from
>>>>> within here to process a mask of your choice then modify the atoms
>>>>> that come back from the mask within setvel depending on what you
>>>> want
>>>>> their velocities (and ultimately temperature) to be.
>>>>>
>>>>> All the best
>>>>> Ross
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Fernando Martín García [mailto:fmgarcia.cbm.uam.es]
>>>>>> Sent: Wednesday, June 15, 2011 7:31 AM
>>>>>> To: amber.ambermd.org
>>>>>> Subject: [AMBER] System temperature
>>>>>>
>>>>>> Dear Amber users:
>>>>>>
>>>>>> I have a question about the assignment of temperature of the
>>>>>> system:
>>>>>> is
>>>>>> there any possibility to assign different temperatures to
>>>> different
>>>>>> parts of the system? I mean, for example, give a high
>>>> temperature
>>>>>> to a
>>>>>> ligand and a "room temperature" to the protein, using two
>>>> different
>>>>>> mask
>>>>>> for the system.
>>>>>>
>>>>>> Thanks
>>>>>>
>>>>>> --
>>>>>> ==============================================
>>>>>> FERNANDO MARTIN GARCIA
>>>>>> GRUPO DE MODELADO MOLECULAR - LAB 312.1
>>>>>> CENTRO DE BIOLOGíA MOLECULAR SEVERO OCHOA
>>>>>> C/ NICOLáS CABRERA, 1.
>>>>>> CAMPUS UAM. CANTOBLANCO, 28049 MADRID. SPAIN.
>>>>>> TEL: (+34) 91-196-4662 FAX: (+34) 91-196-4420
>>>>>> ==============================================
>>>>>>
>>>>>> _______________________________________________
>>>>>> AMBER mailing list
>>>>>> AMBER.ambermd.org
>>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> AMBER mailing list
>>>>> AMBER.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>
>>>> --
>>>> ==============================================
>>>> FERNANDO MARTIN GARCIA
>>>> GRUPO DE MODELADO MOLECULAR - LAB 312.1
>>>> CENTRO DE BIOLOGíA MOLECULAR SEVERO OCHOA
>>>> C/ NICOLáS CABRERA, 1.
>>>> CAMPUS UAM. CANTOBLANCO, 28049 MADRID. SPAIN.
>>>> TEL: (+34) 91-196-4662 FAX: (+34) 91-196-4420
>>>> ==============================================
>>>>
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> AMBER.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>
>>
>> --
>> ==============================================
>> FERNANDO MARTIN GARCIA
>> GRUPO DE MODELADO MOLECULAR - LAB 312.1
>> CENTRO DE BIOLOGíA MOLECULAR SEVERO OCHOA
>> C/ NICOLáS CABRERA, 1.
>> CAMPUS UAM. CANTOBLANCO, 28049 MADRID. SPAIN.
>> TEL: (+34) 91-196-4662 FAX: (+34) 91-196-4420
>> ==============================================
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Dr. Adrian E. Roitberg
Full Professor
Quantum Theory Project, Department of Chemistry
University of Florida
on Sabbatical in Barcelona until August 2011.
Email roitberg.ufl.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 22 2011 - 12:30:05 PDT
