Here is the memo that I promised on the conventions in PHOBOS. These conventions are, for
the most part, built into the code already and it's too late to change them without a lot
of pain.
The following is the convention for defining "Rows" and "Columns" for the PHOBOS Silicon detector elements. This convention will be the same for "hardware", which includes isolated sensors and for "software".
For a rectangular sensor, the number of rows is less than number of columns. E.g. plane A (a type 1 detector) has 22 rows (numbered from 0-21) and 70 columns (numbered from 0-69).
For the ring detector, the row # changes as you change phi, so an individual row runs radially out (or in) from the detector. The column number changes as you change radial distance. NOTE: This convention was chosen so that the ring detectors near phi=0 and phi=pi have the usual sense of rows running horizontally and columns running vertically.
When mounted on the frames, all sensors have the property that changing the row number changes the azimuthal angle phi.
NOTE: Software and hardware definitions of row0 and column 0 are necessarily different.
Sorry about that. In software, we want to match row and column number up to existing
software based on GEANT conventions and based on "physics" quantities like phi
and theta. In hardware, we must be able to describe an UNMOUNTED sensor which knows
nothing about GEANT or about the PHOBOS coordinate system.
0,0 is defined in the software so that the smallest row # corresponds to the most negative value of phi in the mounted sensor. This should be equivalent to the smallest "pady" or local y in the GEANT coordinate system for everything but the rings. row 0 corresponds to the smallest padphi in the rings.
UNFORTUNATELY this means that PHOBOS Y and PadY(or row #) run in opposite directions in the negative arm of the spectrometer. Yuck. Nevertheless, this is the definition and we'll stick to it.
The smallest column number is that column with the smallest value for theta in the spectrometer, the most negative z in the octagon/vertex, and the smallest r in the rings. This should also be the smallest "padx" or local X for the rectangular sensors and the smallest "padr" or local R for the rings.
I don't even know what this is. Pradeep will have to send out a description of the hardware convention for the "origin" of the row&column numbering. NOTE: The database already has an orientation parameter which explicitly relates the hardware and software origins for a given sensor, once it's mounted.
The final code will presumably make it "easy" to get at either definition of the origin for a given sensor because of the wonders of Object Oriented code, yada, yada, yada.
Mark
Memo was sent on Wednesday, July 15, 1998.