Jeff Mitchell was able to get the MVD code running in phool, so I know the problem is with me, not the code. He made three sample plots from his tests. These illustrate three of the key parameters the MVD is expected to return. First is the vertex position. This is a a little better than "typical" -- it is normal to have a sharp spike at zero separation -- meaning the vertex was found correctly. Usually, the vertex will be very close (within a few 100 microns) or else way off (many cm). In this case, the vertex appears to have been found correctly in every event.

The next plot of Jeff's plots shows the multiplicity from the MVD. I did not make the plot, but I assume that this is the quantity dMvdMultOut->totalmult, which is filled by mMvbdNdEta. The inner barrel of the MVD (from which this multiplicity comes) convers around 5 units of pseudo-rapidity and the average dN/deta is around 750 -- so we expect the total multiplicity to be around 3750, in good agreement with this plot.

The third of Jeff's plots shows dN/deta. This also looks reasonable.