06/18/2015 AgSTAR vs VMC Hit Comparisons

Plots shown below illustrate difference in tracking behavior between popstar and starsim.  Many more steps are required to propagate single tracks in a pythia event through identical geometries.  Possible explanations for this are (1) differences in tracking cuts which may remain between the applications; (2) differences in code between STAR's G3 and VMC G3; (3) Possibly due to difference in how TGeo handles MANY vs ONLY volumes.

Impact on the simulated detector response may be minimal.

Figure 1 -- Comparison on single steps through TPAD (tpc active volume) between popstar (black) and starsim (red).  Differences observed in number
and size of steps between the two simulations.




Figure 2 -- Sum of stepsize and energy deposit in TPAD over all steps taken.  Total pathlength in TPAD is consistent between popstar (black) and starsim (red).  Total energy deposited is in reasonable agreement, but differs in detail.



Figure 3 -- LEFT: Number of tracking steps in popstar at a given radius.  RIGHT: starsim.


Figure 4 -- Number of steps vs volume number.  (i.e. this is the number of tracking steps required to reach a given volume.  *NOT* the number of steps taken within a given volume.)  LEFT : popstar.  RIGHT: starsim.  X axis has been zoomed in on the range of TPC volumes.

---

06/17/2015 Notes

Implemented GSTPAR calls, associating tracking cuts (e.g. CUTGAM, CUTELE, etc...) to tracking media.
Fixed medium initialization bug.  IGAUTO flag is set to zero in starsim.  Negative values for deemax, stmin and stemax are set by default in all media in
starsim, which instructs G3 to calculate these values.

---

06/16/2015 TODO LIST

1) Use StIOInterface / maker to save hits in standard STAR format
2) Develop unit tests for popstar hits
3) Check code into offline area in CVS
4) Install latest G3 VMC
5) Install latest G4 VMC and G4
6) Implement legacy STAR geometries in StarAgMLMaker

---

06/15/2015 Possible BUG in TGeant3TGeo (gtonlyTGeo)

TGeant3TGeo's implementation of gtonly always returns 1, without checking to see if a volume's overlapping flag is set.  There is an exceedingly helpful comment left behind which suggests that TGeo geometry always returns only.  Sarcasm aside, we need to verify whether overlaps are handled properly in VMC.

06/15/2015 Bug in popstar -- medium parameters not being passed properly --

Update 1:  Not a bug, but a feature.  By default, step sizes are automatically calculated by G3.  Pavel disabled this in starsim. 
Update 2: We are using an older version of TGeant3, which lacks the functionality to disable automatic medium paramter calculation.  More recent version should be adopted.
Update 3: TGeant3::SetAUTO(0) disable auto configuration of step sizes.  However, results in a significantly longer simulation time.  Seconds --> (at least) minutes. 

Question: Does starsim handle this differently?  Yes, sort of.  Parameters which are flagged with a negative value are automatically computed even when IGAUTO is set to zero.  AgSTAR takes adavantage of this
to autocompute stmin, stmax and deemax.  AgML did not do this.  Rather all step sizes were specified. 

Fix: change default values for stmin, stemax and deemax to be consistent with starsim default.  i.e. negative.