Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Reduce memory usage in Jacobian creation #280

Draft
wants to merge 3 commits into
base: dev
Choose a base branch
from
Draft

Reduce memory usage in Jacobian creation #280

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
88a183e
reduce memory usage by jacobian creation
eastjames Oct 5, 2024
22d6b0a
cleanup
eastjames Oct 5, 2024
c9d1178
merge conflicts
eastjames Oct 5, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 8 additions & 9 deletions src/inversion_scripts/operators/operator_utilities.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,15 +131,14 @@ def read_geoschem(date, gc_cache, n_elements, config, build_jacobian=False):
pert_simulations_dict[run_num] = [sv_elem]
else:
pert_simulations_dict[run_num].append(sv_elem)

gc_date = pd.to_datetime(date, format="%Y%m%d_%H")
gc_date = pd.to_datetime(date, format='%Y%m%d_%H')
ds_all = [
concat_tracers(k, gc_date, config, v, n_elements)
for k, v in pert_simulations_dict.items()
]
ds_sensi = xr.concat(ds_all, "element")
ds_sensi.load()

ds_sensi = xr.concat(ds_all, 'element')

sensitivities = ds_sensi["ch4"].values
# Reshape so the data have dimensions (lon, lat, lev, grid_element)
sensitivities = np.einsum("klji->ijlk", sensitivities)
Expand All @@ -149,7 +148,6 @@ def read_geoschem(date, gc_cache, n_elements, config, build_jacobian=False):
ds_emis_base = concat_tracers(
"0001", gc_date, config, [0], n_elements, baserun=True
)
ds_emis_base.load()
dat["emis_base_ch4"] = np.einsum("klji->ijlk", ds_emis_base["ch4"].values)

# get OH base, run RunName_0000
Expand All @@ -158,7 +156,6 @@ def read_geoschem(date, gc_cache, n_elements, config, build_jacobian=False):
ds_oh_base = concat_tracers(
"0000", gc_date, config, [0], n_elements, baserun=True
)
ds_oh_base.load()
dat["oh_base_ch4"] = np.einsum("klji->ijlk", ds_oh_base["ch4"].values)

return dat
Expand Down Expand Up @@ -220,11 +217,13 @@ def concat_tracers(run_id, gc_date, config, sv_elems, n_elements, baserun=False)
if baserun:
keepvars = ["SpeciesConcVV_CH4"]

ds_concat = xr.concat([dsmf[v] for v in keepvars], "element").rename("ch4")
ds_concat = xr.concat([dsmf[v] for v in keepvars], 'element').rename('ch4')
ds_concat = ds_concat.astype(np.float32, casting='same_kind', copy=False)
ds_concat = ds_concat.to_dataset().assign_attrs(dsmf.attrs)
ds_concat = ds_concat.isel(time=gc_date.hour, drop=True) # subset hour of interest
if not baserun:
ds_concat = ds_concat.assign_coords({"element": sv_elems})
ds_concat = ds_concat.assign_coords({'element':sv_elems})
ds_concat.load()
return ds_concat


Expand Down