1. Sam Skillman
  2. yt

Commits

Show all
Author Commit Message Date Builds
Matthew Turk
Attempting to deal with boundary issues. This is still not functional.
Matthew Turk
Reversing the iteration of fields and cells seems to corect the disparity.
Matthew Turk
Although still providing the incorrect answer for queries of density, this is getting closer to correctly identifying oct indices with positions in the file.
Matthew Turk
Identifying octs to fill is still incorrect; however, this changeset correctly feeds the results of the fill operation back to the user.
Matthew Turk
Attempting to prototype hydro field reading in RAMSES. Disabled ART importing.
Matthew Turk
Merging in Chris's changes.
Christopher Moody
nothing works. first pass at refactor ART frontend.
Matthew Turk
Setting up some structures for RAMSES hydro var reading.
Matthew Turk
Adding fcoords, which is revealing that somehow the coordinates all seem to run from 0..0.5 in the test data, indicating an off-by-one offset in the level information.
Matthew Turk
[octree] Fix a pernicious segfault. Removed a bunch of the domain indices stuff. A base chunk is now identified by a single mask that is shared between chunks. Add ires and icoords. Projections now actually work in manual mode, and given the correct (1.0 everywhere) answer for RAMSES data.
Matthew Turk
Adding a RAMSESDomainSubset for chunking data and storing indices. Starting to rethink the RAMSES data chunking scheme. icoords doesn't quite work because it doesn't apply a selector.
Matthew Turk
[octree] Octrees for RAMSES are now allocated on a domain-by-domain basis. This means that we also have to keep a local versus global counter when iterating over local_ind (which we could perhaps make more elegant somehow). Additionally, we also insert any previously uninserted components of the mesh during the reading of a single domain. So we can construct the entire tree, regardless of which domain we are currently reading. In practice this *could* lead to better lo…
Matthew Turk
[octree] Moving count_cells to base class
Matthew Turk
Set up initial chunking, counting (of octs, domain-octs, and cells) and so on for RAMSES data. Octs now have positions associated with them, but I think we might be able to remove that at some point. The tricky part is that we don't necessarily want to walk from the root for things like counting.
Matthew Turk
Split the RAMSES octree container into a subclass and add decorators for selection routines for octs.
Matthew Turk
RAMSES tree generation and oct handling now works; selectors can be applied (by hand) to the Octree handler.
Matthew Turk
Cleanup of a substantial portion of the Octree reading code for RAMSES. Use the oct_container now, and mostly if not completely pure-python. Add a fortran_utils for reading fortran records.
Matthew Turk
Adding some utilities for handling fortran files.
Matthew Turk
Octree geometry selection seems to work now.
Matthew Turk
A few modifications to the oct geometry container, so it now allocates exactly the correct number of octs in single-domain calculations. Some of the math for identifying children was backwards, too.
Matthew Turk
Splitting Oct definitions into multiple files, attempting a selection routine.
Matthew Turk
Fixed an off-by-one error.
Matthew Turk
Initial import of an oct container.
Matthew Turk
A few fixes to get things to import.
Matthew Turk
Merging from tip of dev branch, to keep up to date.
Matthew Turk
Restore compilation temporarily
Matthew Turk
Adding a forgotten setup.py and removing a validate spatial from particle_mass in enzo.
Matthew Turk
A few minor fixes to the hilbert index utility functions.
Matthew Turk
Add a few cython decorators, which (sadly) have little impact on performance in our particular use case.
Matthew Turk
Beginning conversion of RAMSES to octree geometry handling. Have refactored some of the RAMSES routines, which will not work presently. Moving toward directly handling of Octs.
  1. Prev
  2. Next