Replica Temperatures for Uniform Exchange and Efficient Roundtrip Times in Explicit Solvent Parallel Tempering Simulations

Abstract

The efficiency of parallel tempering simulations is greatly influenced by the distribution of replica temperatures. In explicit solvent biomolecular simulations, where the total energy is dominated by the solvent, specific heat is usually assumed to be constant. From this, it follows that a geometric distribution of temperatures is optimal. We observe that for commonly used water models (TIP3P, SPC/E) under constant volume conditions and in the range of temperatures normally used, the specific heat is not a constant, consistent with experimental observations. Using this fact, we derive an improved temperature distribution which substantially reduces the round-trip times, especially when working with a small number of replicas.