Step by Step Guide: If you want to train using the 8 x 8 matrices jump to step 4.
-
Use dataInput.py to load in data from the Ensight-3D simulation of a mixed, turbulent engine.
a. Enter in the path to the files you are reading in.
b. Enter in the name of the values.
-
Use normalizeInputData.py to normalize the values read in by the previous step.
-
Run either alternateCNNShape.py or cnnFromTurbulentEngine.py to train a network to predict the variance fields given the smoothed fields.
a. alternateCNNShape.py uses the direct convolutional shape used in the Seltz Domingo paper.
b. cnnFromTurbulentEngine.py uses the U-Net model proposed by Ronneberger's team.
-
Use dataInput8By8.py to load in data from the Ensight-3D simulation of a mixed, turbulent engine.
a. Enter in the path to the files you are reading in.
b. Enter in the name of the values.
-
Use normalizeInputData8By8.py to normalize the values read in by the previous step.
-
Run either alternateCNNShape8x8.py to train the network.