mismatch between the ouput ground state energy and mps

[jiangtong1000]

The optimize_mps returns the macro_iteration_result and res_mps. If we evaluate the energy by res_mps.expectation(mpo), the energy is not necessarily equal to min(macro_iteration_result). It is not even necessarily contained in macro_iteration_result. For instance, if the minimal energy is not obtained in the last sweep, but in the middle sweeps (energy does not decrease monotonically with the number of sweeps), then we obtainmin(macro_iteration_result)$\neq$res_mps.expectation(mpo)

[jiangtong1000]

for instance, M=10 for H-4 (sto-3g basis, r0=1.6),
Try to evaluate <MPS|H|MPS>, and it does not equivalent to any value appear in the Figure. I am looking into the code to check the bugs

[jiangtong1000]

ummm, I think the lowest energy happens in the middle because of using the percent in the first several sweeps, so I was wondering if should modify the convergence condition by excluding the results of the first several sweeps. Although this can be seen in very small systems such as I mentioned above.

Renormalizer/renormalizer/mps/gs.py

Lines 123 to 130 in bac8bed

	# check if convergence
	if isweep > 0 and percent == 0:
	v1, v2 = sorted(macro_iteration_result)[:2]
	if np.allclose(
	v1, v2, rtol=mps.optimize_config.e_rtol, atol=mps.optimize_config.e_atol
	):
	logger.info("DMRG has converged!")
	break