Added training data from sparse inversions of flare productive AR 12017.

[fluxtransport]

No description provided.

[fluxtransport]

Note: The training data has lgt = [5.5, .... 7.4], 21 elements instead of 18. This is not compatible with AIA_Resp.npy in the current repository.

[fluxtransport]

Reading genx files requires sunpy.

[PaulJWright]

Are you able to move the files to an additional_examples folder or the similar, and update/reset your repo to the master at PaulJWright/DeepEM?

[fluxtransport]

Hi @PaulJWright do you mean the notebook and/or the genx files?

[PaulJWright]

Hi @PaulJWright do you mean the notebook and/or the genx files?

Just the notebook, really! I would prefer not overwriting the HelioML version.

Cheers,
Paul

[fluxtransport]

I don't even need you to approve the merge. However, I'd like us to use the genx files as a starting point.

So really this is the part that is important:

import glob
from sunpy.io.special import read_genx
aia_files = glob.glob('./DeepEM_Data/*_2014-03-29T1[89]*genx')
aia_files = sorted(aia_files)
print(aia_files)

k = 0
for f in aia_files:
dump = read_genx(f)
if k == 0:
X = np.zeros((len(aia_files), dump['IMG'].shape[0], dump['IMG'].shape[1], dump['IMG'].shape[2]))
y = np.zeros((len(aia_files), dump['EMCUBE'].shape[0], dump['EMCUBE'].shape[1], dump['EMCUBE'].shape[2]))
status = np.zeros((len(aia_files), dump['STATUS'].shape[0], dump['STATUS'].shape[1]))
lgtaxis = dump['LGTAXIS']
nlgT = lgtaxis.shape[0]

X[k] = dump['IMG']
y[k] = dump['EMCUBE']
status[k] = dump['STATUS']
k = k+1

Scale the inputs and outputs

X = img_scale(X)
y = em_scale(y)