specify resonance offset in optimal control

[Mengjia He]

Dear Ilya,

to cover the resonance bandwidth, I found the optimal control example in the Spinach defined multiple spins for the spin system and built a large-size Hamiltonian. Another method I can come up with is to build a small-size spin system but put the bandwidth into ensembles (control.drifts). Since I have a spin system, with 1H and 13C nuclei, each species has a bandwidth. Do you have an idea which method will compute faster?

thank you and best regards,
Mengjia

[kuprov]

Good timing! I've just implemented multiple offset channels in the developer version (follow the instructions on the download page to get it). New specification style is:

control.offsets={[offset values for channel A],[offset values for channel B],...};
control.off_ops={offset_operator_A,offset_operator_B,...};

Note that multiple offset channels add up combinatorially, so the calculation would rapidly slow down as you add more offset channels.

[Mengjia He]

thank you so much, I am lucky to try new features.

[kuprov]

You will also discover massively better parallel efficiency. One other thing you might also want to try is David Goodwin's very efficient Newton-Raphson implementations: use 'newton' or 'goodwin' instead of 'lbfgs'. For small spin systems and control sequences with fewer than 200 slices those methods converge much faster.

[Mengjia He]

that's absolutely good news for me, thank you for implementing this fancy algorithm！