next up previous contents index
Next: Geometry Optimization Up: Controlling adiabaticity for CP-dynamics Previous: Choosing Time Step and   Contents   Index

Additional Considerations and Potential Problems

Another situation that can happen is that you start from an (idealized) structure and the initial wavefunction optimization converges to the wrong state. Now you run CP dynamics and (initially) symmetries are preserved and nothing bad happens, then you switch to massive NH chains and suddenly the electronic system can leave the local minimum it was trapped and you get a large increase of fictitious energy. This is most likely to happen, when the initial atomic guess is quite different from the final structure for a few elements (e.g. transition metals) and in that case it is in many cases better to start the wavefunction optimization from a random wavefunction and/or use CP dynamics with annealing and all atom positions fixed.

One more trick is to lower the highest ionic frequencies by turning the protons, that are not crucial to the process you want to investigate, into deuterium. Their bonded vibrations may be of pretty high frequency and thus slowing them down by increasing their mass may help improving the stability w.r.t. timestep and fictitious mass.

Summarizing, it might happen that you will not be able to assure perfect adiabaticity. Nevertheless you always have to make a choice between numerical accuracy, algorithmical correctness and statistical convergence. It is very important to quantify the amount of error that you are going to introduce, but then you have to weight it against the systematic error of the method (DFT, pseudopotentials, CP-dynamics, ergodicity) and make a balanced choice. For instance the typical error for DFT calculations can be easily in the 1-2 kcal/mol regime, e.g. due to the exchange-correlation functional used. If the energy drift rate has the order of 0.05 kcal/mol per picosecond and what you can afford is 10 ps simulation, the total error introduced by non-adiabaticity is 0.5 kcal/mol during the whole run. Since this seems tolerable although not perfect, you can consider using a thermostat on the electrons during you production run.


next up previous contents index
Next: Geometry Optimization Up: Controlling adiabaticity for CP-dynamics Previous: Choosing Time Step and   Contents   Index
Costas Bekas 2008-09-04