Dear people,
I have optimized a dimer (each monomer contains about 50 to 60 atoms) structure in its equilibrium distances (I have done a rigid scan to find the equilibrium distance between monomers ). The optimization is finished successfully, but how can I approve that the structure is trapped in its global potential well not in the local one. Usually the frequency calculation confirm that, but this kind of calculation for such a big system is very time and resources consuming.
Regards,
Bahareh Bamdad
This can be a difficult problem if, for example, several 'steep' minima are 'close' for similar values of the variables to be optimized.
First, determine the range of variables acceptable for the global minimum.
Second, use a genetic algorithm (like simulated annealing) to precondition the optimization algorithm. If the objective function is computationally expensive this will take a lot of CPU.
Third, after the genetic algorithm has, hopefully successfully, found the range of variables that isolates the global minimum, a more rigorous (gradient, Lagrange multipliers ...) optimization can be applied using the isolated range as a starting point.
An example of these three steps is shown at
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4730672/
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms