The LPP website has provided some answers to my question.
The highest quoted experimental fusion yield/gross input energy I saw quoted on the website was 0.067% from, as I understand it, several years ago. (This is not my ‘compression efficiency’ number, but close enough.)
However, LPP’s paper “Theory and experimental program for p-B11Fusion with the Dense Plasma Focus” explains why this low experimental number is not surprising.
http://lawrencevilleplasmaphysics.com/images/stories/theory_and_experimental_program_for_focus_fusion__lpp_jan2011.pdf
As most readers here probably already know, it describes the device as exploiting a series of four “natural instabilities in the plasma, with each instability further concentrating the plasma and the magnetic field produced by the currents running through the plasma” into the plasmoid.
So optimizing the efficiency of the compression process requires the optimization of these four natural instabilities in sequence, where the efficiency of each is dependent on the plasma’s characteristics produced by the action of the previous instability.
This is an immense solution space with, based on the instability of plasmas, I expect superficially chaotic solution volumes.
I can now see how it is possible, even after 40 years, that recent advances and insights and the resulting improved simulations can reveal new ways to optimize the compression process in such a deceptively simple device.