make_grid_info - Part 1 #79
Conversation
… be able to use logging, move the latlon specific stuff from make_grid_info into its own function, documents the members of the hgridobj
| self.jsc = None | ||
| self.jec = None | ||
|
|
||
| def __init_numpy_arrays__(self, nsuper, narea, ndx, ndy): |
There was a problem hiding this comment.
do the arrays need to be initialized beforehand in python?
There was a problem hiding this comment.
ok init is a bad name, __allocate_numpy_arrays__?
This is allocating memory for all of the 1D members of HGridObj
They are set in pyfrenctools.make_hgrid_wrappers.create_*, which is why they are allocated here.
There was a problem hiding this comment.
I think the allocation can be abstracted into pyfrenctools.make_hgrid_wrappers.create_* (do the allocation there and return the appropriate arrays)
There was a problem hiding this comment.
The way I see this turning out is
-
For lat/lon grids in lonlat_grid.py
-> grid_obj = HGridObj() (initiliazed the hgrid object)
-> grid_obj.make_grid_info_lat_lon(...) (set up the arrays to hold the data)
-> pyfrenctools.make_hgrid_wrappers.create_regular_lonlat_grid( ...) (fill in the arrays)
-> grid_obj.write_out_hgrid(...) (write out the data) -
Simlarly for cubesphere grids in gnomonic_grid.py
-> grid_obj = HGridObj() (initiliazed the hgrid object)
-> grid_obj.make_grid_info_gnomonic(...) (set up the arrays to hold the data)
-> pyfrenctools.make_hgrid_wrappers.create_regular_lonlat_grid( ...) (fill in the arrays)
-> grid_obj.write_out_hgrid(...) (write out the data)
make_grid_info_lat_lon and make_grid_info_gnomonic are going to share code like __allocate_numpy_arrays__
make_grid_info_gnomonic has to do extra stuff (i.e grid_obj.x has a size of nlon * nlon * 6 for non nesting grids and nlon * nlon * 6 + nest_x * nest_y) for grids with nests
Let's talk about this tomorrow :)
|
|
||
| self.logger.debug(f"make_grid_info: Creating a lat-lon with nlon={nlon} and nlat={nlat}") | ||
|
|
||
| self.nxl = np.array(nlon, dtype=np.int32) |
There was a problem hiding this comment.
is nxl, nyl just scalars? They probably can just be self.nxl = nlon...
There was a problem hiding this comment.
They are 1D integers of size equal to the number of tiles. In this lat/lon case they are just one. The code that writes the output is currently expecting them to be arrays.
FMSgridtools/fmsgridtools/make_hgrid/hgridobj.py
Lines 456 to 457 in b2cfa66
There was a problem hiding this comment.
ooo.... that should be changed...
(yup, unnecessarily complicated)
There was a problem hiding this comment.
Yes, there is a lot of unnecessarily complicated code.
I am just tackling this one step at a time.
| ntiles = 6 | ||
| ratio = 2 | ||
| grid_size = 96 | ||
| nlon = np.array([grid_size], dtype=np.int32) |
There was a problem hiding this comment.
no need for scalars to be "np.array"'s
2 participants
This PR: