Maybe a more apt comparison would be to compare all the analytical expressions in CalcZAF, but this time, fit to alpha factors as seen here (since that is how we obtain fast matrix corrections from Penepma):
Summary of All Calculated (averaged) Matrix Corrections:
NIST SRM C2402 (Hastelloy C) (test beta fluorescence)
LINEMU Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV
Elemental Weight Percents:
ELEM: Fe Ni Cr Mo W Co Si Mn V Cu TOTAL
1 7.300 50.742 16.232 16.524 3.806 1.500 .850 .640 .220 .190 98.004 Armstrong/Love Scott (default)
2 7.211 50.412 15.904 17.345 4.518 1.500 .850 .640 .220 .190 98.790 Conventional Philibert/Duncumb-Reed
3 7.316 50.705 16.291 16.664 4.145 1.500 .850 .640 .220 .190 98.522 Heinrich/Duncumb-Reed
4 7.267 50.583 16.145 16.450 3.886 1.500 .850 .640 .220 .190 97.730 Love-Scott I
5 7.297 50.736 16.201 16.512 4.013 1.500 .850 .640 .220 .190 98.159 Love-Scott II
6 7.236 50.578 15.925 17.478 4.371 1.500 .850 .640 .220 .190 98.988 Packwood Phi(pz) (EPQ-91)
7 7.249 52.029 15.819 16.623 3.984 1.500 .850 .640 .220 .190 99.104 Bastin (original) Phi(pz)
8 7.260 50.583 16.079 17.049 4.188 1.500 .850 .640 .220 .190 98.559 Bastin PROZA Phi(pz) (EPQ-91)
9 7.270 50.670 16.092 16.937 4.057 1.500 .850 .640 .220 .190 98.426 Pouchou and Pichoir-Full (Original)
10 7.290 50.767 16.131 16.953 4.083 1.500 .850 .640 .220 .190 98.625 Pouchou and Pichoir-Simplified (XPP)
AVER: 7.269 50.781 16.082 16.853 4.105 1.500 .850 .640 .220 .190 98.491
SDEV: .032 .452 .154 .360 .214 .000 .000 .000 .000 .000 .430
SERR: .010 .143 .049 .114 .068 .000 .000 .000 .000 .000
MIN: 7.211 50.412 15.819 16.450 3.806 1.500 .850 .640 .220 .190 97.730
MAX: 7.316 52.029 16.291 17.478 4.518 1.500 .850 .640 .220 .190 99.104
Very similar results to the normal analytical expressions at least, but it's interesting that the best total (99.532) is from the Penepma fast Monte Carlo method...