Hi Snehasis,

your question is not clear because you need to further detail what you mean by artificial. So I will make a guess at what you intend. For the sake of argument, let us define natural heating to be properties of a system in thermodynamic equilibrium - which the others who have answered your question have explained can be simulated using various types of thermostats (e.g. Langevin, Tobias Tuckerman Martyna Klein (MTTK) , formulations) applied typically to all the degrees of freedom of your system. The physics and mathematics of thermostats and barostats are actually very deep - see recent papers by Alessandro Sergi and Mark Tuckerman for example. Artificial heating would then correspond to a heating protocol/algorithm which does not reproduce equilibrium properties, and can be very useful, for example if you want to model the transfer of heat across a system held at different temperatures, or some other non-equilibrium set-up. These can to some extent still be modeled using thermostats, applied for example to subsets of the degrees of freedom. A great simulation code for doing all this sort of thing is LAMMPS from Sandia National Labs which is open-source, and where the source code is readable by mere mortals. Other codes (gromacs, dlpoly, namd) may work well for this too but I am less familiar with them, and how transparent their source codes are.

Best Regards,

Donal

Dear Snehasis

If i understand your question well, you can increase the temperature gradually and automatically from T(start) to T(end) during you simulation time. Hope this can help you. Be free if there are more question about it.

As you are probably aware, for molecular simulation techniques it is standard practice to heat a simulated liquid to a desired temperature via various thermostating methods (Direct velocity rescaling, Langevin, Nose Hoover).

If you could expand a bit on your question maybe we can give you some more insight.

Kind regards, Jeff.

Hi Snehasis,

your question is not clear because you need to further detail what you mean by artificial. So I will make a guess at what you intend. For the sake of argument, let us define natural heating to be properties of a system in thermodynamic equilibrium - which the others who have answered your question have explained can be simulated using various types of thermostats (e.g. Langevin, Tobias Tuckerman Martyna Klein (MTTK) , formulations) applied typically to all the degrees of freedom of your system. The physics and mathematics of thermostats and barostats are actually very deep - see recent papers by Alessandro Sergi and Mark Tuckerman for example. Artificial heating would then correspond to a heating protocol/algorithm which does not reproduce equilibrium properties, and can be very useful, for example if you want to model the transfer of heat across a system held at different temperatures, or some other non-equilibrium set-up. These can to some extent still be modeled using thermostats, applied for example to subsets of the degrees of freedom. A great simulation code for doing all this sort of thing is LAMMPS from Sandia National Labs which is open-source, and where the source code is readable by mere mortals. Other codes (gromacs, dlpoly, namd) may work well for this too but I am less familiar with them, and how transparent their source codes are.

Best Regards,

Donal

Other than at a phase transition, how do you heat something and leave its temperature the same? The question has some auxiliary assumptions that need clarification.

Thanks all... I want to clarify my question :

I want to add heat in water slowly and want to study the temperature rise. In order to do so in computer simulation what method/technique can be used ?

This is not a phase transition but just heating a system.

Is it clear ?

Hi Snehasis,

Heating is due to the interaction with a heat bath which has specific geometry and properties. If you are interested in the temperature rise (real time evolution of temperature) of the water, these issues are important.

For example, you may consider that your water is in a container and that heating comes from the walls which are maintained at a given temperature T. In a simulation of gas dynamics for example, you may technically do this by imposing a thermalized distribution (at temperature T) for the energy (or speed) of the particles after each collision with the walls.

You may then observe the time evolution of temperature accross the whole container. Note that this temperature will be spatially inhomogeneous during the transient period between the initial thermal state (if you choose an equilibrium initial condition at temperature T_0) and the final equilibrium state at temperature T.

The final equilibrium state is obviously always the same independently of the choice of heat bath. However, the evolution of temperature and the relaxation time scale will strongly depend on the choice of heat bath and system (water, dimensions,...).

Hope this helps,

Best,

Gregory

If there are no equilibration issues (for example such issues arise for glasses), then the standard approach is to add or subtract energy from the momentum coordinates, and watch what happens. Normally one does this by multiplying all momentum values by a number slightly larger than one, and waiting until the average values of the mean-square kinetic energies settles down. For normal systems, it will settle down at a slightly lower value, because some kinetic energy will redistribution into potential energy. For reasonable small-molecule systems, the energy is reasonably expected to equilibrate between all coordinates in readily accessible computational time scales. For water, there is a major simulational complication: The protons (hydrogen atom) in hydrogen bonds (this is every H atom in liquid water, more or less) hop back and forth between their two oxygen atoms. that's a substantial contribution to the specific heat. I do not recall a discussion of how to simulate this effect.

I think you have to be careful in asking what the true physical mechanism of heating in your system is. Nonetheless these references which consider transient heating for molecular dynamics might be of some use to you.

All the best, Jeff.

http://pubs.rsc.org/en/Content/ArticleLanding/2010/CP/b918607g#!divAbstract

http://pubs.rsc.org/en/Content/ArticleLanding/2009/SM/b817666c#!divAbstract

Hi Snehasis again,

just in case that problem is not resolved, you can slowly heat the system by literally slowing increasing the temperature of the thermostat in your md. You can play with this rate too - to uncover/reveal meta-stabilities. For instance, use a very fast ramp rate - say 100 degrees over a few ns. You can monitor various quantities to see how the system is responding, for example energy fluctuations which are related easily enough to the specific heat capacity. If you are also using a barostat, you can look an the total volume. Then, double the simulation time and repeat. If there are no meta-stabilities, the curves should remain close to each other. However close to a phase transition you will find a significant deviation. Now focus/zoom on to those intervals in temperature, and repeat the process, redefining the initial and final temperatures.

It will prove a lot of fun, and you can extract a lot of physics out of it. This is in spirit similar to what experimentalists call time dependent calorimetry.

Best Regards,

Donal

Many many thanks for your reply. It will really help me a lot !