Erice Work List
We should start with a list of things we want to achive
Activity
I assume that we have at the time of Erice 1.6.0 released. All tasks are w.r.t. 1.9.9 or 2.0.0
-
Is it realistic to release 2.0.0 with proper documentation and regression tests.
-
Finish at least some if #45
-
Finish #46 (closed)
-
Immediate thoughts:
- Make sure manual is up-to-date.
- Double check multibunch tracking in OPAL-cycl (not sure if there are still issues there)
- Fix one issue with defining reference point in geometry for SAAMG solver
- Quo vadis, SAAMG solver? (If sticking around, maybe also tackle parallelism of it?)
- Unify units?
Edited by winklehner_d-
I assume that we have at the time of Erice 1.6.0 released. This is the last public release before 2.0.0
All tasks are w.r.t. 1.9.9 or 2.0.0 unless otherwise noted:
The goal should be 1.9.9 ready, i.e. the last developer version before 2.0.0
- Manual
- Finish #45 #46 (closed)
- minimum number of particles for statistic (*.stat file) write
- minimum number of particles for statistic H5
OPAL-cycl
- Remove duplicated/not used code in OAPL-cycl
- Double check multibunch tracking in OPAL-cycl (not sure if there are still issues there)
- Unify units OPAL-t / OPAL-cyc check also #80 ?
- OPAL-cyc has special files do we want to keep them or add the information into the h5 file?
SAAMG solver
- Fix one issue with defining reference point in geometry for SAAMG solver
- Quo vadis, SAAMG solver? (If sticking around, maybe also tackle parallelism of it?)
OPAL-T
- OPAL-T: make energy bins dynamic where the number of energy bins and the number of particles per bin can vary
Testing framework
- regression tests, with/without GPU
- unit tests, Jenkins etc
Distribution class
- split into stuff for the parser and implementation
- use separate classes for the different types
Edited by adelmann added To Do and removed ~13 labels
OPAL-cycl uses particle with ID=0 as referenz particle which is a bad choice for it.
- you newer know on which core your reference particle resides
- unintentional effects occur e.g. no output about ref particle, see here https://gitlab.psi.ch/OPAL/src/blob/master/src/Algorithms/ParallelCyclotronTracker.cpp#L1674
- the other cores have to be informed about position and momentum of reference particle
- the code is unnecessary complicated if you have to filter for you ref particle
- the ref particle contributes to the coulomb field but it is filtered from the statistics Consider using a separate particle not contained in PartBunch::R as reference.
Edited by krausIf I were to do a refactoring job on OPAL-cycl, I would think about making it a bit more structured. In particular, I have in mind that someone might want to add new routines for tracking (symplectic integrator?), different output routines, fancier materials models, etc. I would consider how one could structure the code so that it is relatively simple to add/modify these sorts of things. Probably it won't be done at Erice, but one could discuss the direction of travel.
mentioned in commit 0bc0cce9
mentioned in commit 9d2c771f
mentioned in commit 1cc7a6df
-
Erice Working List
Long(er) term projects
- Reference Particle(s) CLASS, for both flavours
- Refactoring OPAL-cycl
- many unit conversion (m,T)
Work for Erice
-
Quaternion (Daniel,AA OPAL-t) this is all unified in 2.0.0. Maybe part of Quaternion could inherited from boost
-
cleanup cyclotron units: (Daniel) a) internal x,y,z b) external use x,z,y later this can be set by a switch c) delete trackers done
-
timedependent E (B) a la Ring (Daniel, Chris)
-
Fix one issue with defining reference point in geometry for SAAMG solver (Achim/Daniel)
-
Fieldsolver sanity check (Andreas) done
Edited by adelmann marked this issue as related to #227
Let's close this one in favour of the more recent #227.
