Choose timezone
Your profile timezone:
BLonD code development meeting
→
Europe/Zurich
CERN
CERN
Video conferencing only
Participants: Helga, Theo, Kostis, Panagiotis, Markus, Alex, Luis, Giulia, Ivan, Heiko, Simon, Mihaly
News
Helga: BLonD PR#163, added new documentation
PR message:
New documentation for SPS OTFB and LHC OTFB
Uploaded source files for images in docs
Updated auto-moduling of toolbox folder
Added new drift equation formulation to description of equations of motion
* will merge it
* Also uploaded on github pages
* Temporal website, Helga will add us to the authors, then we can edit, add applications
* We keep it minimal and simple, we point to the github repo for further info.
Alex Q for website: Drupal now, there is a push for a different one (MKDocs).
* Drupal looks consistent with CERN websites and more generic, we stay with it.
BLonD_common
* PR38 from Alex: Correction to find_potential_wells_cubic if the max is too close to edge
* will add unittests in the next PR
* a library of all failing edge cases
* Also try to figure out a way to store these input data for the tests
* merge this PR
* One large PR from Simon
PR39: New version of ring and rf_station objects, plus a few other bits, https://github.com/blond-org/blond_common/pull/39
* very volatile, cannot add unittests yet
* maybe new branch for experimentation, before merged in the master
* Discussion about the potential well sorting order and size
* Going through the commits, file by file
* Alex added some __EXAMPLES
* some commits have been discussed in previous meetings
* only the barebones for ring and the tracker, that we extend in other codes.
* No need to leave it unmerged for too long, we can merge it.
* Helga to Simon: Set up some usage examples, and in next meetings we can have a look and figure out what needs to stay in blond_common and what goes to the blond_tracker.
* What will be added in beam_dynamics folder?
- tracking but without tracking, what is in beams in blond
- Alex: overlapping with what is in blond now.
- Helga: bucket should go to rf functions
- maybe rename it to something different, maybe evolution
- not really needed if we move the bucket to rf functions
- Heiko suggests that we dedicate a discussion to a single object, we move more slowly but in more detail.
* Helga: Simon and Alex show us some use case examples, next week.
* Simon showing usage example slides (they are on indico)
* Simon showing his gitlab repo
* Simon uploaded his slides
* Helga: For next time, share more examples, and we have a detailed look in the ring object, define the minimal subset.
GPU Development
Panagiotis slides
* Linear interpolation kick ~15x speedup in the GPU
* kick function, depends on the n_rf: ranges from 10x to 25x (n_rf 1 to 5)
* drift 3 solvers: simple, legacy,
* simple --> ~25x
* legacy --> ~25x
* full solver --> ~30x
* Next time stats for the synch rad
* Ivan
- merge all the loop of synch radiation
- Kostis: Have another look at the synch rad for the CPU
- Panagiotis: n_kick? Ivan In principle 1.
Recap for next meeting (in one week):
- Discuss ring object, Alex & Simon use cases
- Kostis Sync rad for CPU
- Panagiotis GPU for sync rad + integration
There are minutes attached to this event.
Show them.