Updated for pushing to HelioML notebook; added HelioML files in folde…

…r, including new gif
PaulJWright · Nov 28, 2018 · 73dc06c · 73dc06c
1 parent 04c4b9b
commit 73dc06c
Show file tree

Hide file tree

Showing 20 changed files with 1,507 additions and 333 deletions.
diff --git a/DeepEM.ipynb b/DeepEM.ipynb
diff --git a/DeepEM_CNN_HelioML.pth b/DeepEM_CNN_HelioML.pth
diff --git a/HelioML_Files/.ipynb_checkpoints/HelioML_intro-checkpoint.ipynb b/HelioML_Files/.ipynb_checkpoints/HelioML_intro-checkpoint.ipynb
diff --git a/HelioML_Files/DeepEM.ipynb b/HelioML_Files/DeepEM.ipynb
diff --git a/HelioML_Files/DeepEM.md b/HelioML_Files/DeepEM.md
@@ -0,0 +1,16 @@
+
+DeepEM: A Deep Learning Approach for DEM Inversion
+============================================
+*by Paul Wright, Mark Cheung, Rajat Thomas, Richard Galvez, Alexandre Szenicer, Meng Jin, Andrés Muñoz-Jaramillo, and David Fouhey*
+
+
+In this chapter we will introduce a deep learning approach for Differential Emission Measure (DEM) Inversion: DeepEM. <i>For this notebook</i>, `DeepEM` is a trained on one set of <i>SDO</i>/AIA observations (six optically-thin channels; 6 x N x N) and DEM solutions (in 18 temperature bins from logT = 5.5 - 7.2 K, 18 x N x N; [Cheung et al 2015](https://doi.org/10.1088/0004-637X/807/2/143)) at a resolution of 512 x 512 (N = 512) using a 1x1 2D Convolutional Neural Network with a single hidden layer.
+
+The `DeepEM` example presented here takes every DEM solution with no regards to the quality or existence of the solution. As will be demonstrated, when this method is trained with a single set of <i>SDO</i>/AIA images and DEM solutions, `DeepEM` solutions have a similar fidelity to Basis Pursuit (with a significantly increased computation speed—on the order of 10 million DEM solutions per second), and additionally, `DeepEM` finds positive solutions for every pixel.
+
+This notebook was developed with [PyTorch](https://pytorch.org/), and `DeepEM` is currently being under development at https://github.com/PaulJWright/DeepEM.
+
+<IMG SRC="./1/docs/DeepEM_171_211.gif" width="100%">
+
+Example of the `DeepEM` solutions for logT ~ 5.9 K and 6.3 K in comparison to the Basis Pursuit (and <i>SDO</i>/AIA images) for the same temperature.
+
diff --git a/HelioML_Files/DeepEM_CNN_HelioML.pth b/HelioML_Files/DeepEM_CNN_HelioML.pth
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-01-27.aia.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-01-27.aia.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-01-27.emcube.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-01-27.emcube.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-01-27.status.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-01-27.status.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-02-22.aia.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-02-22.aia.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-02-22.emcube.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-02-22.emcube.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-02-22.status.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-02-22.status.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-03-20.aia.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-03-20.aia.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-03-20.emcube.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-03-20.emcube.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_DEM_2011-03-20.status.npy b/HelioML_Files/DeepEM_Data/AIA_DEM_2011-03-20.status.npy
diff --git a/HelioML_Files/DeepEM_Data/AIA_Resp.npy b/HelioML_Files/DeepEM_Data/AIA_Resp.npy
diff --git a/HelioML_Files/HelioML_intro.ipynb b/HelioML_Files/HelioML_intro.ipynb
diff --git a/HelioML_Files/docs/DeepEM_171_211.gif b/HelioML_Files/docs/DeepEM_171_211.gif
diff --git a/Misc/DeepEM_171_211.gif b/Misc/DeepEM_171_211.gif
diff --git a/readme.md b/readme.md
@@ -16,9 +16,9 @@ The `DeepEM` method presented here takes every DEM solution with no regards to t
 
 This notebook was developed with [PyTorch](https://pytorch.org/).
 
-<IMG SRC="./Misc/DeepEMGif2.gif" width="100%">
+<IMG SRC="./Misc/DeepEM_171_211.gif" width="100%">
 
-Example of the `DeepEM` solution for logT ~ 5.9 K, in comparison to the Sparse Inversion solution for the same temperature. Additional Gifs can be found in `Misc/`
+Example of the `DeepEM` solution for logT ~ 5.9 K and 6.3 K, in comparison to the Sparse Inversion solution, and the SDO/AIA images for the same temperature. Additional Gifs can be found in `Misc/`.
 
 
 Directory Structure