Choose timezone
Your profile timezone:
BLonD code development meeting
→
Europe/Zurich
864/1-B04 (CERN)
General news:
- Password change for blond
- Joint publication:
- About half of the contributions drafted, to be finished by end of next week
- Proofread during 3-22 February, use comments to each other or discuss directly
- Simon: multiple-beam implementation
- Originally for longitudinal painting
- Sum coordinates; need to have the same superparticle ratio and particle type for all
- Simon: data check class
- check type of input data
- Simon: rf modulation implemented
- all tested with unittests
- Gitlab mirror - BLonD repository (only) at the moment, to the BLonD group
- Deepcopy - discussed by Kostis
Kostis: MPI tests and Assembler
- Simplified user interface for MPI version
- To run in hpc batch, see README on BLonD-mpi
- Execute with mpirun -n numberProcesses
- Markus: experience with SPS simulations: simulation of 1-3 days reduced to 1 h
- Some difference in numpy FFT vs FFTW3 FFT; the first bunch suffers the most: profiles evolve differently
- Check with low intensity, does it converge when not at the limit of stability?
- Assembler: first draft done
- Will always append the tracker
- Old method with maps still maintained for now as is
- Single RF station only for now
- Rename "tracker" with something more speaking
- Some general stuff:
- -Ofast flag issues with fast_sine function
- Improved compile.py: tries to use existing files to get errors at compile time rather than during run time
- Added examples as unittests for integration tests
- Deepcopy() issue
- Causes memory issues, esp. in MPI
- Used in bunch generation methods and trackers.utilities
- Selectively copy what is needed, not a deepcopy of the entire object
- Or reassign from scratch all parameters, initialising new classes
- GPU cluster with 4 servers operational soon, accessible by anyone
- BLonD and BLonD-mpi should be merged at some point
There are minutes attached to this event.
Show them.