Monte carlo temperature in ReplicaExchange0
Hello everyone
I am modeling a system with the ReplicaExchange0 macro and my simulation constantly falls into a local energy minima after a few replica exchange steps (<100). This reduces a lot the reproducibility of the simulation, as the final results greatly depend on the starting configuration of the rigid bodies. I realized that tweaking temperature parameters improves the situation, which allows the simulation to sample a much larger space and hopefully explore all minima. I would like to know if the values that I chose in the following parameters are in reasonable range.
monte_carlo_temperature = 100 replica_exchange_minimum_temperature = 100 replica_exchange_maximum_temperature = 250
Thank you and best regards
Genis Valentin
On 10/16/18 12:59 AM, Genis Valentin wrote: > I am modeling a system with the ReplicaExchange0 macro and my simulation > constantly falls into a local energy minima after a few replica exchange > steps (<100).
How many replicas are you using, and how many independent trajectories? This sounds to me like you need more of each.
> I would like to know if the values > that I chose in the following parameters are in reasonable range. > > monte_carlo_temperature = 100 > replica_exchange_minimum_temperature = 100 > replica_exchange_maximum_temperature = 250
These values seem very large to me. Typical values are more in the 1-10 range. I imagine that such large temperatures would allow far too many score-increasing moves to occur, essentially ignoring the scoring function.
Ben
participants (2)
-
Ben Webb -
Genis Valentin