The purpose of creating this feature in YT is to analyze field
-properties that surround
s the dark matter halo. Originally, this was
+properties that surround dark matter halo. Originally, this was
usually done with the sphere 3D container, but since many halo
particles are linked together in a more elongated shape, I thought it
would be better to use an ellipsoid 3D container to wrap around the
-particles. This way less of the empty-of-particle space around the
+particles. This way less of the empty-of-particle space around the
halo would be included when doing the analysis of field properties
where the particles are suppose to occupy.
the y axis that will align the first semi-principle axis with the x
axis once it is aligned in the x-z plane.)
-Once those parameters are determine
s, the function "ellipsoid" will
+Once those parameters are determine, the function "ellipsoid" will
return the 3D object, and users will be able to get field attributes
-from the data object just they would from spheres, cylinders etc.
+from the data object just they would from spheres, cylinders etc.
And calculates the ellipsoid information needed for the 3D container.
What I usually do is get this information from the halo finder output
-files in the .h5 HDF5 binary format. I load them in using the
+files in the .h5 HDF5 binary format. I load them in using the
LoadHaloes() function instead of reading in the ASCII output.
haloes = LoadHaloes(pf, 'MyHaloList')
Once the halo information is saved you can load it into the data
-object haloes, you can get loop over the list of haloes and do
+object haloes, you can get loop over the list of haloes and do
smallest ellipsoid that contains the particles as possible. To
start at the center of the particles based on position will require
an O(:math:`N^2`) operation, right now I'm trying to limit
- everything to O(:math:`N`) operations.
+ everything to O(:math:`N`) operations. If particle count does not
+ get too large, I may implement the O(:math:`N^2`) operation.
* Currently the list of haloes can be analyzed using object
parallelism (one halo per core), but I'm not sure if haloes will get
big enough soon that other forms of parallelism will be needed to