quemb.molbe.mbe.BE.optimize

BE.optimize(solver='MP2', method='QN', only_chem=False, use_cumulant=True, conv_tol=1e-06, relax_density=False, J0=None, nproc=1, ompnum=4, max_iter=500, trust_region=False, solver_args=None)

BE optimization function

Interfaces BEOPT to perform bootstrap embedding optimization.

Parameters:

  • solver (str) – High-level solver for the fragment, by default ‘MP2’

  • method (str) – Optimization method, by default ‘QN’

  • only_chem (bool) – If true, density matching is not performed – only global chemical potential is optimized, by default False

  • use_cumulant (bool) – Whether to use the cumulant energy expression, by default True.

  • conv_tol (float) – Convergence tolerance, by default 1.e-6

  • relax_density (bool) – Whether to use relaxed or unrelaxed densities, by default False This option is for using CCSD as solver. Relaxed density here uses Lambda amplitudes, whereas unrelaxed density only uses T amplitudes. c.f. See http://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C34df668afbHW-7216-1405+00.htm for the distinction between the two

  • max_iter (int) – Maximum number of optimization steps, by default 500

  • nproc (int) – Total number of processors assigned for the optimization. Defaults to 1. When nproc > 1, Python multithreading is invoked.

  • ompnum (int) – If nproc > 1, ompnum sets the number of cores for OpenMP parallelization. Defaults to 4

  • J0 (ndarray[Tuple[int, ...], dtype[floating]] | None) – Initial Jacobian.

  • trust_region (bool) – Use trust-region based QN optimization, by default False

Return type:

None