Sph projections, slices, gridding: backend and off-axis

doc/source/analyzing/generating_processed_data.rst

@@ -54,16 +54,17 @@ the transformation of a variable mesh of points consisting of positions and
 sizes into a fixed-size array that appears like an image.  This process is that
 of pixelization, which yt handles transparently internally.  You can access
 this functionality by constructing a
-:class:`~yt.visualization.fixed_resolution.FixedResolutionBuffer` and supplying
+:class:`~yt.visualization.fixed_resolution.FixedResolutionBuffer`
+and supplying
 to it your :class:`~yt.data_objects.data_containers.YTSelectionContainer2D`
 object, as well as some information about how you want the final image to look.
 You can specify both the bounds of the image (in the appropriate x-y plane) and
 the resolution of the output image.  You can then have yt pixelize any field
 you like.
 
 .. note:: In previous versions of yt, there was a special class of
-          FixedResolutionBuffer for off-axis slices.  This is no longer
-          necessary.
+          FixedResolutionBuffer for off-axis slices.  This is still used
+          for off-axis SPH data projections: OffAxisFixedResolutionBuffer.
 
 To create :class:`~yt.data_objects.data_containers.YTSelectionContainer2D` objects, you can
 access them as described in :ref:`data-objects`, specifically the section
@@ -99,7 +100,10 @@ this, see :ref:`saving-grid-data-containers`.
 In the FITS case, there is an option for setting the ``units`` of the coordinate system in
 the file. If you want to overwrite a file with the same name, set ``clobber=True``.
 
-The :class:`~yt.visualization.fixed_resolution.FixedResolutionBuffer` can even be exported
+The :class:`~yt.visualization.fixed_resolution.FixedResolutionBuffer`
+(and its
+:class:`~yt.visualization.fixed_resolution.OffAxisProjectionFixedResolutionBuffer`
+subclass) can even be exported
 as a 2D dataset itself, which may be operated on in the same way as any other dataset in yt:
 
 .. code-block:: python

doc/source/quickstart/4)_Data_Objects_and_Time_Series.ipynb

@@ -337,6 +337,8 @@
     "\n",
     "There are two different types of covering grids: unsmoothed and smoothed.  Smoothed grids will be filled through a cascading interpolation process; they will be filled at level 0, interpolated to level 1, filled at level 1, interpolated to level 2, filled at level 2, etc.  This will help to reduce edge effects.  Unsmoothed covering grids will not be interpolated, but rather values will be duplicated multiple times.\n",
     "\n",
+    "For SPH datasets, the covering grid gives the SPH-interpolated value of a field at each grid cell center. This is done for unsmoothed grids; smoothed grids are not available for SPH data.\n",
+    "\n",
     "Here we create an unsmoothed covering grid at level 2, with the left edge at `[0.0, 0.0, 0.0]` and with dimensions equal to those that would cover the entire domain at level 2.  We can then ask for the Density field, which will be a 3D array."
    ]
   },
@@ -385,13 +387,13 @@
  ],
  "metadata": {
   "kernelspec": {
-   "name": "python3",
    "display_name": "Python 3.9.5 64-bit ('yt-dev': pyenv)",
    "metadata": {
     "interpreter": {
      "hash": "14363bd97bed451d1329fb3e06aa057a9e955a9421c5343dd7530f5497723a41"
     }
-   }
+   },
+   "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {

doc/source/visualizing/plots.rst

@@ -293,8 +293,8 @@ argument. Optionally, a ``north_vector`` can be specified to fix the orientation
 of the image plane.
 
 .. note:: Not every data types have support for off-axis slices yet.
-   Currently, this operation is supported for grid based data with cartesian geometry.
-   In some cases (like SPH data) an off-axis projection over a thin region might be used instead.
+   Currently, this operation is supported for grid based and SPH data with cartesian geometry.
+   In some cases an off-axis projection over a thin region might be used instead.
 
 .. _projection-plots:
 
@@ -433,6 +433,8 @@ by applying the
 In this use case, the volume renderer casts a set of plane parallel rays, one
 for each pixel in the image.  The data values along each ray are summed,
 creating the final image buffer.
+For SPH datsets, the coordinates are instead simply rotated before the axis-aligned
+projection function is applied.
 
 .. _off-axis-projection-function:
 
@@ -652,10 +654,6 @@ simply pass ``all`` as the first argument of the field tuple:
 Additional Notes for Plotting Particle Data
 -------------------------------------------
 
-An important caveat when visualizing particle data is that off-axis slice plotting is
-not available for any particle data. However, axis-aligned slice plots (as described in
-:ref:`slice-plots`) will work.
-
 Since version 4.2.0, off-axis projections ares supported for non-SPH particle data.
 Previous to that, this operation was only supported for SPH particles. Two historical
 workaround methods were available for plotting non-SPH particles with off-axis

nose_ignores.txt

@@ -45,3 +45,5 @@
 --ignore-file=test_set_log_level\.py
 --ignore-file=test_field_parsing\.py
 --ignore-file=test_disks\.py
+--ignore-file=test_offaxisprojection_pytestonly\.py
+--ignore-file=test_sph_pixelization_pytestonly\.py

tests/tests.yaml

@@ -37,7 +37,7 @@ answer_tests:
     - yt/frontends/amrvac/tests/test_outputs.py:test_riemann_cartesian_175D
     - yt/frontends/amrvac/tests/test_outputs.py:test_rmi_cartesian_dust_2D
 
-  local_arepo_011:  # PR 4419
+  local_arepo_013:  # PR 4939
     - yt/frontends/arepo/tests/test_outputs.py:test_arepo_bullet
     - yt/frontends/arepo/tests/test_outputs.py:test_arepo_tng59
     - yt/frontends/arepo/tests/test_outputs.py:test_arepo_cr
@@ -92,7 +92,7 @@ answer_tests:
     - yt/frontends/flash/tests/test_outputs.py:test_wind_tunnel
     - yt/frontends/flash/tests/test_outputs.py:test_fid_1to3_b1
 
-  local_gadget_009:  # PR 3258
+  local_gadget_010:  # PR 4939
     - yt/frontends/gadget/tests/test_outputs.py:test_iso_collapse
     - yt/frontends/gadget/tests/test_outputs.py:test_pid_uniqueness
     - yt/frontends/gadget/tests/test_outputs.py:test_bigendian_field_access
@@ -108,7 +108,7 @@ answer_tests:
   local_gdf_002:
     - yt/frontends/gdf/tests/test_outputs_nose.py:test_sedov_tunnel
 
-  local_gizmo_008:  # PR 2909
+  local_gizmo_009:  # PR 4939
     - yt/frontends/gizmo/tests/test_outputs.py:test_gizmo_64
 
   local_halos_012:  # PR 3325
@@ -118,7 +118,7 @@ answer_tests:
     - yt/frontends/gadget_fof/tests/test_outputs.py:test_fields_g5
     - yt/frontends/gadget_fof/tests/test_outputs.py:test_fields_g42
 
-  local_owls_008:  # PR 2909
+  local_owls_009:  # PR 4939
     - yt/frontends/owls/tests/test_outputs.py:test_snapshot_033
     - yt/frontends/owls/tests/test_outputs.py:test_OWLS_particlefilter
 
@@ -130,7 +130,7 @@ answer_tests:
     - yt/visualization/tests/test_particle_plot.py:test_particle_phase_answers
     - yt/visualization/tests/test_callbacks.py:test_axis_manipulations
 
-  local_tipsy_009:  # PR 2909
+  local_tipsy_010:  # PR 4939
     - yt/frontends/tipsy/tests/test_outputs.py:test_pkdgrav
     - yt/frontends/tipsy/tests/test_outputs.py:test_gasoline_dmonly
     - yt/frontends/tipsy/tests/test_outputs.py:test_tipsy_galaxy
@@ -221,5 +221,7 @@ other_tests:
      - "--exclude-test=yt.frontends.gdf.tests.test_outputs.TestGDF"
      - "--exclude-test=yt.frontends.adaptahop.tests.test_outputs"
      - "--exclude-test=yt.frontends.stream.tests.test_stream_particles.test_stream_non_cartesian_particles"
+     - "--ignore-file=test_offaxisprojection_pytestonly\\.py"
+     - "--ignore-file=test_sph_pixelization_pytestonly\\.py"
   cookbook:
      - 'doc/source/cookbook/tests/test_cookbook.py'

yt/data_objects/construction_data_containers.py

@@ -649,6 +649,8 @@ class YTCoveringGrid(YTSelectionContainer3D):
     level : int
         The resolution level data to which data will be gridded. Level
         0 is the root grid dx for that dataset.
+        (The grid resolution will be simulation size / 2**level along
+         each grid axis.)
     left_edge : array_like
         The left edge of the region to be extracted.  Specify units by supplying
         a YTArray, otherwise code length units are assumed.
@@ -995,14 +997,15 @@ def _fill_sph_particles(self, fields):
 
         smoothing_style = getattr(self.ds, "sph_smoothing_style", "scatter")
         normalize = getattr(self.ds, "use_sph_normalization", True)
+        kernel_name = getattr(self.ds, "kernel_name", "cubic")
 
         bounds, size = self._get_grid_bounds_size()
 
         period = self.ds.coordinates.period.copy()
         if hasattr(period, "in_units"):
             period = period.in_units("code_length").d
-        # TODO maybe there is a better way of handling this
-        is_periodic = int(any(self.ds.periodicity))
+        # check periodicity per dimension
+        is_periodic = self.ds.periodicity
 
         if smoothing_style == "scatter":
             for field in fields:
@@ -1036,6 +1039,7 @@ def _fill_sph_particles(self, fields):
                         pbar=pbar,
                         check_period=is_periodic,
                         period=period,
+                        kernel_name=kernel_name,
                     )
                     if normalize:
                         pixelize_sph_kernel_arbitrary_grid(
@@ -1051,6 +1055,7 @@ def _fill_sph_particles(self, fields):
                             pbar=pbar,
                             check_period=is_periodic,
                             period=period,
+                            kernel_name=kernel_name,
                         )
 
                 if normalize: