|
Signed Data (Text SHA1) |
Dear Users,
in case you used the ZigZag current deposition scheme and *all* of the
following criteria apply to you, you will be affected by a bug that René
Widera discovered today:
- 2D3V
- ZigZag current deposition scheme for charged species
- relevant v.z component in particles (e.g., warm plasma start)
https://github.com/ComputationalRadiationPhysics/picongpu/pull/823
Currents along the z-direction where over-estimated by a factor 2.
The ZigZag current deposition scheme was a generalization of three
papers that we implemented past year as an alternative to the Esirkepov
scheme. It provides about a factor 1.6 speedup due to reduced memory
access and scales better than the Esirkepov implementation with number
of particles per cell.
It is *not* enabled by default yet.
If you did not change the value (UsedParticleCurrentSolver) in
`species.param` , which was in earlier versions in
`componentsConfig.param`, you will *not* be affected.
As soon as #823 is merged, `dev` will not be affected any more. The fix
will also be back-ported to the upcoming `0.1.0` release of PIConGPU.
Best regards,
Axel
--
Axel Huebl
Phone +49 351 260 3582
https://www.hzdr.de/crp
Computational Radiation Physics
Laser Particle Acceleration Division
Helmholtz-Zentrum Dresden - Rossendorf e.V.
Bautzner Landstrasse 400, 01328 Dresden
POB 510119, D-01314 Dresden
Vorstand: Prof. Dr.Dr.h.c. R. Sauerbrey
Prof. Dr.Dr.h.c. P. Joehnk
VR 1693 beim Amtsgericht Dresden
|
Content Unaltered as verified By: |
Huebl, Axel <a.huebl@hzdr.de> |
|
|