We noticed a bug in the visualization notebook DOFS calculation. This only occurs when using clustering. In make_gridded_posterior.py we map the calculated sensitivity values to the corresponding state vector elements. Eg. if element 3 has a sensitivity of .5, and is composed of 10 grid cells, then all 10 grid cells are set to .5. However, in visualization_notebook.ipynb the way we calculate DOFS is np.nansum(avkern_ROI.values) which sums every .25 degree grid cell. Per the example, we are effectively overcounting the sensitivity of element 3 by 10x.
To fix this we simply use the un-gridded inversion result from inversion_result.nc to calculate DOFS.
Credit to Alex Oort Alonso for flagging this bug!
Confirmed that this fix works in the Permian test case (with no clustering).
Name and Institution (Required)
Name: Lucas Estrada Institution: Harvard ACMG
Describe the update
We noticed a bug in the visualization notebook DOFS calculation. This only occurs when using clustering. In
make_gridded_posterior.py
we map the calculated sensitivity values to the corresponding state vector elements. Eg. if element 3 has a sensitivity of .5, and is composed of 10 grid cells, then all 10 grid cells are set to .5. However, invisualization_notebook.ipynb
the way we calculate DOFS isnp.nansum(avkern_ROI.values)
which sums every .25 degree grid cell. Per the example, we are effectively overcounting the sensitivity of element 3 by 10x.To fix this we simply use the un-gridded inversion result from
inversion_result.nc
to calculate DOFS.Credit to Alex Oort Alonso for flagging this bug!
Confirmed that this fix works in the Permian test case (with no clustering).