[gmx-users] crazy temperatures

[chris.neale at utoronto.ca]

I disagree.

What one is generally trying to obtain with elevated temperatures is  
enhanced sampling, not temperature-dependent properties. I believe  
that even TIP4P-EW is not very good at getting the properties of water  
correct at 600 K, temperatures that are commonly used during replica  
exchange simulations (not to mention that nobody has any idea how  
accurate protein forcefields are at temperatures other than the one at  
which they were parameterized).

So I think that doing simulations at massively elevated temperatures  
can possibly be useful.

That said, while doing simulated annealing, I have found previously  
using charmm that once you get to about 3,000 K you will get chiral  
inversions that can not resolve at lower temperature. This is because  
our improper dihedral terms only maintain the given chirality, rather  
than favouring one over the other.

To address your question directly, I believe that chiral inversions  
will be a big problem for you at 20,000 K. Obviously you also have  
simulation stability issues, but one presumes that you could resolve  
those by using a small enough timestep.

Chris.

-- original message --

At that temperature most matter is going to be a plasma, not many  
bonds to be simulated and a lot of free electrons.

Warren Gallin

On 2012-03-28, at 4:43 PM, Mark Abraham wrote:

[Hide Quoted Text]
On 29/03/2012 9:39 AM, Asaf Farhi wrote:
Dear GMCS users

Hi. Does anyone know if MD at 20000K is feasible?
Please start new email threads rather than hijacking old ones.

I doubt anybody knows the answer to your question. Force fields are  
parameterized to reproduce data at around 300K. I can't imagine any  
possible use for simulating an MM force field at a temperature hotter  
than the sun.

Mark
--