1. Could you explain the significance of your article to the non-specialist?
Our article presents and discusses the parallel tempering computer simulation technique.   In this technique, the investigator conducts simulations at a series of temperatures.  Swaps of configurations between these parallel simulations allow the low temperature system of interest to escape from local free energy minima where it might otherwise have been trapped.  We discuss various implementations of the method, how to optimize the technique to most efficiently sample phase space, and several important applications of the method where new science has been uncovered due to the use of parallel tempering.

2. What has motivated you to conduct this work?
The main motivation for doing this work was to encourage other researchers to examine and adopt the method in their own studies.  Parallel tempering is easy to implement, increases the efficiency of computer simulations in almost all the areas it can be applied, and works well on large clusters of CPUs that are becoming more and more widespread.

3. Where do you see this work developing in the future?
We expect that parallel tempering will become a staple method in the computer simulation of complex systems, due to its effectiveness and the ease with which the method can be combined with Monte Carlo or molecular dynamics programs.  We see a particular area of opportunity in the replacement of simulated annealing for X-ray crystal structure determination.  Other potential new areas of application include crystal polymorph prediction and computational rational drug design.

4. Are there any particular challenges facing future research in this area?
Future challenges in the area include how to optimally choose the order parameters in which to temper.  That is, rather than just temperature, one can run multiple simulations where the order parameters that vary between replicas can include the Hamiltonian, chemical potential, pressure, and many others.  Another challenge is how to swap only part of a system, to avoid a \sqrt N increase in the number of required replicas as a system becomes larger.  A practical issue is how to use the increase sampling efficiency that parallel tempering brings to improve current force fields, especially implicit solvent models.