
How can we do the PK algorithm choice better?
Why is the SCF converged to such crazy tolerences?
Read 4 index quantity from dpd without storing it in memory twice?
Does anyone actually read CI NO's or OPDMs off disk?


-----
David's list

How do we free strings, graphs, SigmaData
Occs array is created twice if alplist != betlist
In the TPDM code it looks like the symmetrize call used to assume that the orbials were not real
    i.e. pqrs != qpsr why is this?

We seem to get the following a lot:
(sem_iter): H0block_->H0b_diag[1][0] - H0block_->H0b_diag[2][0] = -0.393516 - 0.393516 = -0.787032 > 1.0E-8
Is this a phase issue?
This is more of a warning, maybe its finding the wrong state. 


