Simulation of Co(II) dimer (large matrices)

[niconeuman]

Greetings,
I was trying to simulate a dinuclear low spin Co(II) system where there may be a small exchange and dipolar interaction, as well as large and resolved hyperfine splittings to the Co(II), using pepper. I noticed that attempting anything other than full matrix diagonalization produced unrealistic spectra, because the hyperfine splitting is here probably as large or larger than the exchange and/or dipolar interaction. But using Opt.Matrix I waited for hours before shutting down the simulation and couldn't even get one powder spectrum.
I have been able to simulate single crystal spectra with this system though, and they take a few seconds each. My question is, are there simulation options which would allow me to decrease considerably the simulation time? The first think I'm going to try is nKnots, but as this is a considerably anisotropic spectrum I don't know how much I will be able to reduce it.
If you could help me it would be greatly appreciated!
Best regards,
Nicolas

[Stefan Stoll]

Here are a few suggestions:

• You might be hitting the memory limit on your computer. Check your memory usage while you are runing the powder simulation.
• Try reducing Opt.nKnots - both numbers.
• It could also be that there are a lot of forbidden transitions with non-negligible amplitude such that EasySpin wants to include them. Try increasing Opt.Threshold a bit.
• You can also run the crystal sim for a few selected orientation and ask pepper for the list of transitions, then use that list for the powder simulation.
• Try to reduce the nuclear spin to see if you can get the powder sim running, e.g. replace Co by 55Mn (just for testing).
• Try Opt.Sparse=true. This instructs EasySpin to use sparse matrices instead of full matrices internally. It could help, or not.

[niconeuman]

Dear Prof. Stoll
thank you very much for these recommendations.
I will test them one by one.
Best regards,
Nicolas