SWIFTsim merge requestshttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests2024-02-13T12:31:21Zhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1777Draft: Streamlined MAGMA2 implementation2024-02-13T12:31:21ZMatthieu SchallerDraft: Streamlined MAGMA2 implementationThis is a reimplementation of the MAGMA2 branch with:
- Simpler symmetric matrix handling
- Only 1st order reconstruction of `v` and `u`.
- Only basic viscosity and diffusion terms as in the Rosswog 2020 implementation
Questions:
...This is a reimplementation of the MAGMA2 branch with:
- Simpler symmetric matrix handling
- Only 1st order reconstruction of `v` and `u`.
- Only basic viscosity and diffusion terms as in the Rosswog 2020 implementation
Questions:
- Do we need more advanced terms? SPHENIX-like? Or the Rosswog entropy-based viscosity switch?
- What implementation do we want? Traditional or Gasoline?
Todo:
- [ ] Document the matrix stuff
- [ ] 1D and 2D cases
- [ ] Scale-factors
- [ ] Strange corner cases (e.g. no ngb)
- [ ] Tests in galaxy settings
- [ ] Properly implement the symmetric functions.
- [ ] Fix documentation. It still mentions `Minimal` and `SPHENIX` in places.https://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1776adding balsara eq 1 in ODI scheme2023-10-07T09:12:24ZFederico Andrés Stasyszynadding balsara eq 1 in ODI schemeTesting new modality of pushes. Jusr add Balsara switch eq 1 in ODI.
I also add some additional fields in the standard plot.py of Roberts flowTesting new modality of pushes. Jusr add Balsara switch eq 1 in ODI.
I also add some additional fields in the standard plot.py of Roberts flowMatthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1775Testing and putting Balsara switch into ODI2023-10-06T12:19:35ZFederico Andrés StasyszynTesting and putting Balsara switch into ODIadditionally changed plot routine in Roberts Flow to show other fieldsadditionally changed plot routine in Roberts Flow to show other fieldshttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1774skip particle timestep data collection in space_split.c when running without RT2023-11-09T16:40:19ZMladen Ivkovicskip particle timestep data collection in space_split.c when running without RTIssue reported by Yolan in !1749 .
For new progeny cells, when running without RT, particle time bins are zero. Then those time bins get collected to determine the time step vars for the progeny cell, leading to wrong results.
This oug...Issue reported by Yolan in !1749 .
For new progeny cells, when running without RT, particle time bins are zero. Then those time bins get collected to determine the time step vars for the progeny cell, leading to wrong results.
This oughtta fix it, but I haven't had time to test it yet.
ToDo's Mladen:
* [x] Testing this fix
* [x] Check that there aren't other places where this could be an issue
* [x] run the bloody formatterMatthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1773Add barotropic EoS2023-10-06T14:55:04ZMatthieu SchallerAdd barotropic EoSAdd barotropic EoS that is used in some MHD problems.
Todo:
- [x] Add reference to the paper where this is taken from,
- [x] Fix the few equations labelled as "IS THIS CORRECT",
- [x] Fix the Doxygen comments to have the correct equa...Add barotropic EoS that is used in some MHD problems.
Todo:
- [x] Add reference to the paper where this is taken from,
- [x] Fix the few equations labelled as "IS THIS CORRECT",
- [x] Fix the Doxygen comments to have the correct equations.
- [x] Fix the example/parameter_examples.yml file to have a description of the two parameters.Matthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1771Add a function so that we can mark memory as deliberately leaked at the end o...2023-10-07T09:12:57ZMatthieu SchallerAdd a function so that we can mark memory as deliberately leaked at the end of the program.Alternative to !1767.
Implements #867.Alternative to !1767.
Implements #867.Matthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1770Mhd canvas2023-10-02T16:38:07ZFederico Andrés StasyszynMhd canvasFederico Andrés StasyszynFederico Andrés Stasyszynhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1769Homogenized Version of the codes2023-09-29T16:53:59ZFederico Andrés StasyszynHomogenized Version of the codesA small clean up to for the different schemes.
Same nomenclature and variable names.
Still some inconsistencies.A small clean up to for the different schemes.
Same nomenclature and variable names.
Still some inconsistencies.Matthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1768Mhd canvas2023-09-29T09:50:22ZFederico Andrés StasyszynMhd canvasFederico Andrés StasyszynFederico Andrés Stasyszynhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1767Add an allocation function that disables the memory sanitizer.2023-10-05T14:37:31ZMatthieu SchallerAdd an allocation function that disables the memory sanitizer.Implements #867.
This is useful for the allocation call in space_getcells() since we don't (can't) clean up the allocations properly.Implements #867.
This is useful for the allocation call in space_getcells() since we don't (can't) clean up the allocations properly.Peter W. DraperPeter W. Draperhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1766added read Vz factor from the yml files.2023-09-28T16:03:40ZFederico Andrés Stasyszynadded read Vz factor from the yml files.double to float in forcing terms
printout messagedouble to float in forcing terms
printout messagehttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1765Avoid crash when writing healpix maps caused by HDF5 chunk size larger than t...2023-11-16T09:09:19ZJohn HellyAvoid crash when writing healpix maps caused by HDF5 chunk size larger than the datasetFixes #862.
When nside is small the lightcone map HDF5 dataset written by one MPI rank can be smaller than the chunk size. In this case we need to reduce the chunk size to the dataset size or else H5Dcreate() will fail.
This merge requ...Fixes #862.
When nside is small the lightcone map HDF5 dataset written by one MPI rank can be smaller than the chunk size. In this case we need to reduce the chunk size to the dataset size or else H5Dcreate() will fail.
This merge request also disables chunking of the map datasets when no filters are in use.Matthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1764Fix jet initial velocities2023-09-28T12:18:17ZFilip HuskoFix jet initial velocitiesFixes #866.Fixes #866.Matthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1763Draft: Add ICC Milan optimization and change AMD architecture to corei7-avx.2023-10-20T11:57:43ZPeter W. DraperDraft: Add ICC Milan optimization and change AMD architecture to corei7-avx.Add ICC Milan CPU ids and optimization.
Change the basic architecture to corei7-avx from core-avx2 as tests show that to be 20% faster for gravity only and the same speed for hydro.Add ICC Milan CPU ids and optimization.
Change the basic architecture to corei7-avx from core-avx2 as tests show that to be 20% faster for gravity only and the same speed for hydro.Peter W. DraperPeter W. Draperhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1762Fix new time-step for spin-jet scheme2023-09-19T12:35:13ZFilip HuskoFix new time-step for spin-jet schemeI have tested out changes introduced in the recent MR with all the SPIN-JET updates, but one issue did not come up until I ran high-res tests down to low enough redshifts in a large enough volume. Specifically, one black hole gets very l...I have tested out changes introduced in the recent MR with all the SPIN-JET updates, but one issue did not come up until I ran high-res tests down to low enough redshifts in a large enough volume. Specifically, one black hole gets very low time-steps. I managed to trace the issue to the new time-step, which is computed in the following way in lines 1354-1356 in `black_holes.h`:
```
const float dt_redirection = 0.1 * m_warp(bp, constants, props) /
bp->accretion_rate * j_BH(bp, constants) /
j_warp(bp, constants, props) / sine;
```
The problem is in the fact that `m_warp` and `j_warp` are defined as `float`s (and they have been all this time), but their values can be very small (e.g. `m_warp` can be 10^-5 Msol). With this fix they are changed to be `double`s, and I have confirmed this stops the problematic BH from having very small time-steps.https://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1761Fix eagle tracers output2023-09-19T07:45:22ZMladen IvkovicFix eagle tracers outputRecent change in tracers/EAGLE/tracers_io.h introduced a bug: The count for the elements for star outputs is incorrect, and misses several indices. This leads to HDF5 crashing output at snapshot zero because it cannot create a dataset wi...Recent change in tracers/EAGLE/tracers_io.h introduced a bug: The count for the elements for star outputs is incorrect, and misses several indices. This leads to HDF5 crashing output at snapshot zero because it cannot create a dataset without a name, nor without any content.Matthieu SchallerMatthieu Schallerhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1760Correct Units2023-09-15T12:14:38ZFederico Andrés StasyszynCorrect Units-Added correct Units to Dinduction fede and VP
-added vecpotenetial to the gametized cloud-Added correct Units to Dinduction fede and VP
-added vecpotenetial to the gametized cloudhttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1759Mhd canvas2023-09-15T11:24:03ZFederico Andrés StasyszynMhd canvashttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1758Mhd canvas2023-09-15T10:24:35ZFederico Andrés StasyszynMhd canvashttps://gitlab.cosma.dur.ac.uk/swift/swiftsim/-/merge_requests/1757Mhd canvas2023-09-14T13:33:32ZFederico Andrés StasyszynMhd canvas