RANDOM SEED: Difference between revisions
No edit summary |
No edit summary |
||
| Line 11: | Line 11: | ||
The initial value of {{TAG|RANDOM_SEED}} and the value after each MD step are written to the {{FILE|REPORT}} file. | The initial value of {{TAG|RANDOM_SEED}} and the value after each MD step are written to the {{FILE|REPORT}} file. | ||
{{NB|tip|The random seed is used to improve reproducibility of exact numbers during every step. If the time-averaged results depend on the random seed, the ensemble is not big enough, i.e., the molecular-dynamics simulation time is too short.}} | {{NB|tip|The random seed is used to improve reproducibility of exact numbers during every step. If the time-averaged results depend on the random seed, the ensemble is not big enough, i.e., the molecular-dynamics simulation time is too short.}} | ||
Revision as of 09:19, 8 April 2022
RANDOM_SEED = [integer][integer][integer]
Default: RANDOM_SEED = based on the system clock
Description: RANDOM_SEED specifies the seed of the random-number generator (compile VASP with -Dtbdyn).
The random-number generator (RNG) generates a sequence of random numbers for molecular-dynamics simulations for example to initialize atomic velocities and for the Andersen thermostat (MDALGO=1 | 11). The seed for the RNG determines the trajectory a molecular-dynamics simulation takes. Set the RANDOM_SEED tag to reproduce a previously performed calculation.
A typical input for the RANDOM_SEED looks like this:
RANDOM_SEED = 248489752 0 0
The initial value of RANDOM_SEED and the value after each MD step are written to the REPORT file.
| Tip: The random seed is used to improve reproducibility of exact numbers during every step. If the time-averaged results depend on the random seed, the ensemble is not big enough, i.e., the molecular-dynamics simulation time is too short. |