Dear HPS Software,
The notes from yesterday's software meeting are on Confluence:
Please let me know if you have anything to show for next week's meeting.
Attending: Omar, Narges, Matt, Takashi, Denny, Ebrahim, Maurik
- Coordinate Transformations -- Maurik -- PDF FILE** Notes from meeting:
- It seems it would be nice to use the LAB frame for everything. This would avoid confusion about the coordinate system we are in. It would make things easier when dealing with survey data etc.
- The rotations involved seem relatively simple, but we have no clear idea of how much work would be really involved in the lcsim framework to get this done. *** Tracking currently has the magnetic field "hard wired" in the z direction.
- The hits would need to be rotated from LAB frame to SLIC frame.
- Not clear how much the multiple scattering and such would complicate this.
- Decision on this item will be postponed until we get some input from Norman, Jeremy and Rich.
- The alternative is to do all the geometry and tracking on the SLIC frame, and then rotate the resulting physics vectors to the LAB frame before writing out a DST.
- CED -- All
We should discuss and then vote on whether we want to encourage Dave Heddle to invest some time into creating an HPS version of his Event Display system. If we decide we want this, someone needs to sit down and hammer out what/how we want things displayed.** Notes from meeting:
- Everyone agreed that CED12 seems a very useful program. All are in favor of encouraging Dave Heddle to create a version of this code for HPS.
- Matt will present some ideas on how we would like the tracker hits to be displayed for this program.
- Clara -- All
We never decided one way or another on the Clara system. This is probably not for the test run, but if it is something we want for the full HPS run, we need to decide on this soon.** Notes from meeting:
- It appears that the CLARA framework may help in the problem of distributing the data over a large number of nodes, something we need for the Level 3 trigger.
- It is not clear how exactly the CLARA framework would be implemented for HPS. *** Some thoughts: Vardan would be providing some wrappers for the existing lcsim code. Specific blocks of lcsim would be implemented as a CLARA service.
- Key to successful implementation is identifying the code blocks to turn into a service and the communication required between these services.
- This is again something were we would need Norman, Jeremy and Rich to comment.
- Omar will look into CLARA to see if it can be useful for the Level 3 trigger.
- Merging Events -- Matt** Matt showed us some code that will merge many 2ns simulated events into longer events, allowing for a negative offset. This looks promising for us to get long pulse trains.
- Maurik remarked that he tried something like this, using a rolling buffer where for each next event to be analyzed, one new event is popped onto the buffer, one is dropped and all the other events are rolled by one slot. This is efficient in that you do not "waste" MC events. In the case of a trigger the buffer is flusher. However, it turned out that it becomes rather difficult to get the statistics correct.
- A similar problem would be if you randomly mixed the events. You would need to be very careful not to use the same event (which may cause a trigger) in the trigger window twice.