Hello,
is it possible to get calculated acceptance ratio in output file? if not, then how I can calculate it from general output file with replicas, for example the following one:
LAMMPS (5 Jun 2019)
Running on 4 partitions of processors
Step T0 T1 T2 T3
0 0 1 2 3
2000 0 1 3 2
4000 1 0 2 3
6000 2 0 1 3
8000 2 1 0 3
10000 1 2 0 3
Thanks,
Elena.
Hello,
is it possible to get calculated acceptance ratio in output file? if not, then how I can calculate it from general output file with replicas, for example the following one:
you know the number of attempted swaps and by comparing adjacent lines of the output you can determine how many swaps are successful. this can be easily programmed in a little post-processing script.
axel.
yes, you can infer the tempering acceptatnce ratio from the output that you included.
Note that at each swap step one of 2 kinds of swap pairing are done. For 4 replicacas
it would either be 0 <-> 1 and 2 <-> 3. Or it would be just 1 <-> 2, with replicas 0 and 4 idle.
This is odd/even pairing described on the temper doc page based on seed1.
So in your case:
0 0 1 2 3 # initial states
2000 0 1 3 2 # 0,1 and 2,3 attempts, only 2,3 was successful
4000 1 0 2 3 # 0,1 and 2,3 attemps, both were successful
6000 2 0 1 3 # 1,2 attempt, successful
8000 2 1 0 3 # 01, and 2,3 atttempts, only 0,1 successful
10000 1 2 0 3 # 1,2 attempt, successful
I guess if nothing is successful, it is not possible to tell whether that was the odd or even pairing.
We should probably add a couple columns of output to indicate odd vs even, and the count of
accepted.
Steve