y2009a geometry status, take 2

We have found and corrected a bug in the TPC geometry which was masking off a significant amount of material in the TPC endcap.  The bug was that the TPC gas volume extended beyond its nominal z=210 cm limit, causing an overlap which geant could not properly resolve.  (It is still unclear why geant could not resolve this overlap).  The figures below show illustrate the old TPC geometry, the TPC geometry model developed by Yuri and Victor's implementation of Yuri's geometry within starsim.

1.0 Comparison of the Material Distribution Plots

Figure 1.1 shows the number of radiation lengths vs. eta for the "old" TPC geometry, used in many production tags.  
Figure 1.1 -- Material distribution of the tpce03 (tpcgeo2.g) version of the TPC.
Figure 1.2 shows the material distribution for Yuri's geometry.  Yuri's geometry was created going off of detailed TPC drawings, and checked by reconciling the total mass of the TPC with documented values.
Figure 1.2 -- Material distribution of Yuri's model of the TPC, based on ROOT TGeo geometry modeler.
Figure 1.3 shows Victor's implementation of Yuri's geometry within the starsim framework.  Differences in color coding of the histograms should be ignored.  However, detailed comparison shows that many of the structural details between the two geometries are similar.
Figure 1.3 -- Material distribution of Victor's agstar implementation of Yuri's TPC model.

2.0 Differences between the geometries

Figure 2.1 shows the difference between Yuri's model and Victor's model.  There are differences in detail in the structure... individual volumes may not be positioned at the correct place in one or both geometries, causing peaks and valleys in the material difference.  However, integrated over the range 0.5 to 2.5 in eta, the integrated difference of the mean radiation length is 0.7.
Figure 2.1 -- Difference between Yuri's and Victor's TPC models.
Figure 2.2 compares victor's TPC geometry to the "old" model.  Here, large differences are clearly visible.  In particular, there is significant material missing in the old geometry near eta=2.  The integrated mean difference is 12, significantly greater than in figure 2.3. 
Figure 2.2 -- Difference between Victor's TPC model and the "old" TPCe03 / tpcgeo2.g model.

3.0 Conclusion

Victor's TPCe04 (tpcgeo3.g) model of the TPC represents a significant improvement in the geometry model.  This model will be used in the y2009a production library.