updated for paper reproducability

.ipynb_checkpoints/README-checkpoint.md

@@ -12,7 +12,15 @@ license: mit
 
 This respository contains trained neural networks that can be used to predict the steady-state temperature profile.
 
-Step 1: Define the simulation parameters 
+Here is how to predict several profiles at once:
+Step 1: Define the simulation parameters in my_simulation_parameters.txt
 Step 2: The output is as follows  
-        - the depth profile (first column) and the temperature profile (second column)  
-        - corresonding plot of the temperature profile  
+        - outputs/.txt files: the depth profile (first column) and the temperature profile (second column)  
+        - outputs/.pdf files: corresonding plot of the temperature profile  
+
+
+Here's how to train a neural network: 
+Step 1: Ensure you have a working installation of PyTorch (https://pytorch.org/get-started/locally/)
+Step 2: cd train/
+Step 3 with GPU   : python train_profiles_mlp.py -l 5 -f 128 -a "selu" -gpu 0
+Step 3 without GPU: python train_profiles_mlp.py -l 5 -f 128 -a "selu" 

.ipynb_checkpoints/calculate_profiles-checkpoint.py

@@ -7,9 +7,11 @@ write_file   = True
 plot_profile = True
 #### Define outputs ####
 
+# load the saved mlp
 with open('numpy_networks/mlp_[128, 128, 128, 128, 128].pkl', 'rb') as file: 
     mlp = pickle.load(file) 
 
+# read the simulation parameters and ensure correct foramtting
 f_nn = "my_simulation_parameters.txt"    
 with open(f_nn) as fw:
     lines = fw.readlines()
@@ -32,6 +34,7 @@ for line in lines:
 if not len(r_list) == len(v_list) and len(r_list) == len(t_list):
     raise Exception("Ensure equal number of values for all parameters in " + f_nn)
 
+# Check parameter ranges and print appropriate warnings
 for i in range(len(r_list)):
     if r_list[i]<0 or  r_list[i]>9.5:
         warnings.warn('RaQ/Ra is outside the range of the training dataset') 

.ipynb_checkpoints/utils-checkpoint.py

@@ -25,7 +25,7 @@ def dimensionalize_fkv(x):
     return 10**(x*(1.9927988938926755-0.005251646002323797)+0.005251646002323797)
 
 def get_input(raq_ra, fkt, fkp, y_prof):
-    
+    # define input as (batch, [raq/ra, fkt, fkp and y])
     x = np.zeros((len(raq_ra)*len(y_prof), 4))
 
     cntr = 0
@@ -40,7 +40,9 @@ def get_input(raq_ra, fkt, fkp, y_prof):
     return x
 
 def get_profile(inp, mlp, num_sims=1, correction=True, prof_points=128):
+    # get predicted profile based on input
 
+    # forward network pass in pure python using saved weights
     num_layers = len(mlp)-1
     y_pred = inp
     res = []
@@ -59,11 +61,12 @@ def get_profile(inp, mlp, num_sims=1, correction=True, prof_points=128):
             res.append(y_pred)
                 
     y_pred = y_pred.reshape(num_sims, -1)
-    
+
+    # overwrite points at the boundary
     y_pred[:,0]  = 1.
     y_pred[:,-1] = 0.
 
-    if correction:
+    if correction:   # boundary layer corection
         inp = inp.reshape(num_sims, -1, inp.shape[-1])
     
         for sim_ind in range(num_sims):

Paper/los.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64f4894aef6463779d0de2527dce81c839cc5ad6e21358bc57910b807b89a64c
+size 195952322

Paper/overall_stats.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5422a53fa78d2ff9c7ae8ae0a064d972f2fd3fdbc58ccc1ef5ad7eb9dc647c3b
+size 577

Paper/simulations.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df59faeb9a989fa7751ebe13b513b0a3cf23bf6face190697c90f04567425cf9
+size 4628

Paper/x_p.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d76ada3a1d22cafcd67ac04a5acbc0798fe7e6626ffd367aed535df895da5924
+size 3537

Paper/y_p.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60058bfd5fcf2253384ca64602147641bdd50f4dc41505a132e255b5fbf73e0a
+size 131327

Paper/y_pred_krr.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d5946c988e5d38c6d18f435eea9b349aabddc254439e670d7fa9dd0183a6b69
+size 131327

Paper/y_pred_linear.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7413ff2d58d66c626c103560403025dca21b1f8615568b942eec596f712236cc
+size 131327

Paper/y_pred_nn_interpolation.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1702a46ed99540ef523bb67495b63f2f4eaa530538923e680c57f65832d87a73
+size 131327

Paper/y_pred_nn_nearestneighbor.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dda5a9c7d01651b0f00d547069aa4191bb24e39d8a2174b0cd82fb4d1cdc08b
+size 131327

Paper/y_pred_nn_pointwise.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54555229f98cb33f134b472259bdb5fc20cdd2ea99bdff056dbba2d3f3a2a039
+size 131327

Paper/y_prof.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74b064010ebed301b83d5a771864cddcf40972b92b7b93d9bad5be36b606fc6f
+size 1174

README.md

@@ -12,7 +12,15 @@ license: mit
 
 This respository contains trained neural networks that can be used to predict the steady-state temperature profile.
 
-Step 1: Define the simulation parameters 
+Here is how to predict several profiles at once:
+Step 1: Define the simulation parameters in my_simulation_parameters.txt
 Step 2: The output is as follows  
-        - the depth profile (first column) and the temperature profile (second column)  
-        - corresonding plot of the temperature profile  
+        - outputs/.txt files: the depth profile (first column) and the temperature profile (second column)  
+        - outputs/.pdf files: corresonding plot of the temperature profile  
+
+
+Here's how to train a neural network: 
+Step 1: Ensure you have a working installation of PyTorch (https://pytorch.org/get-started/locally/)
+Step 2: cd train/
+Step 3 with GPU   : python train_profiles_mlp.py -l 5 -f 128 -a "selu" -gpu 0
+Step 3 without GPU: python train_profiles_mlp.py -l 5 -f 128 -a "selu" 

calculate_profiles.py

@@ -7,9 +7,11 @@ write_file   = True
 plot_profile = True
 #### Define outputs ####
 
+# load the saved mlp
 with open('numpy_networks/mlp_[128, 128, 128, 128, 128].pkl', 'rb') as file: 
     mlp = pickle.load(file) 
 
+# read the simulation parameters and ensure correct foramtting
 f_nn = "my_simulation_parameters.txt"    
 with open(f_nn) as fw:
     lines = fw.readlines()
@@ -32,6 +34,7 @@ for line in lines:
 if not len(r_list) == len(v_list) and len(r_list) == len(t_list):
     raise Exception("Ensure equal number of values for all parameters in " + f_nn)
 
+# Check parameter ranges and print appropriate warnings
 for i in range(len(r_list)):
     if r_list[i]<0 or  r_list[i]>9.5:
         warnings.warn('RaQ/Ra is outside the range of the training dataset') 

data/scaling_laws_data.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dcc41721a779493ba2d4f7517ef59794de61f79ac5c37a7944952b191cf3989
+size 16783233

data/scaling_laws_data_hot_start.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31c2d899a964b4f641148841a67dea83d863d98860b707297ec30bdfbc5acd9b
+size 22631031

data/sims.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f2c00c1fcc0e05799a37c8d657542d6cfcfd8b6f51ef192b9551e0b03dcbd97
+size 22319

data/snapshots.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae8f8d8545c7ab78db38e3d9b6b98eae74d21d2f004811a2cb81acffb4ecdfdd
+size 12437058

data/x_Tprev.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:351805bd60d4fd84c25112718702c24789545f3df5e18df65efe8b06e7bb77c0
+size 6938

data/x_modes.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4afe1c0dbc408af7cb3900c7116851e02e84f560c95ed10d9627134551c28a7
+size 3318

data/x_p.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4afe1c0dbc408af7cb3900c7116851e02e84f560c95ed10d9627134551c28a7
+size 3318

data/x_pointwise_orgres.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24405442e95cae6608a507e044b0414ee2983da71e6e9cf61ed9343dafdb0471
+size 524555

data/y_Tprev.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6eaeca58b64bb330ab8ad220f344f83a7773910f6dda2e98763e242bcccaa995
+size 18584253

data/y_modes.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1acfa7a99ed3edda7ef750dbd30889d56c954be11e889173265a50321d72dbe5
+size 33014

data/y_p.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60058bfd5fcf2253384ca64602147641bdd50f4dc41505a132e255b5fbf73e0a
+size 131327

data/y_pointwise_orgres.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d19c505b85811df63cba87046342b1ae5930e259251757f50131c16357cc0866
+size 131330

data/y_prof.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1be9d7e71bb22d9cddc3f6a4c2f363f4e7eb9879ed60cf2d8b8825f21f97f18
+size 2135

my_simulation_parameters.txt

@@ -0,0 +1,10 @@
+#### Define parameters ####
+
+# raq/ra betweel 0    and 10
+r_list = 5,7.5,10,
+
+# FKT between  1e+6 and 1e+10
+t_list = 1e+8,1e+9,1e+10,
+
+# FKV between    1e+0 and 1e+2
+v_list = 50,25,1,

outputs/profile_raq_ra10.0_fkt10000000000.0_fkv1.0.txt

@@ -0,0 +1,128 @@
+1.0   1.0
+0.99609375   0.9902098942725738
+0.9881562500000001   0.9703163994344438
+0.98021875   0.9698090412062678
+0.97228125   0.9691821542327205
+0.96434375   0.9684515420121397
+0.95640625   0.9677356340791369
+0.94846875   0.9671697830904316
+0.94053125   0.9667202236581387
+0.93259375   0.966311449336034
+0.92465625   0.9659481243331839
+0.91671875   0.9656773805469817
+0.90878125   0.9654725654458026
+0.90084375   0.9653150700204262
+0.89290625   0.9651916980635759
+0.88496875   0.965092961746218
+0.87703125   0.9650111687302013
+0.86909375   0.9649397652692542
+0.8611562500000001   0.9648770260555964
+0.85321875   0.9648247711847248
+0.84528125   0.9647767597243578
+0.83734375   0.9647278248222245
+0.82940625   0.9646797526182401
+0.82146875   0.9646316205912606
+0.81353125   0.9645885560525155
+0.80559375   0.9645405864163007
+0.79765625   0.9644910702591126
+0.78971875   0.9644372136707845
+0.78178125   0.9643808755612203
+0.77384375   0.964321872132771
+0.76590625   0.9642631270679868
+0.75796875   0.9642026502634682
+0.75003125   0.9641388550418537
+0.74209375   0.96407165722111
+0.7341562500000001   0.9640034450644884
+0.72621875   0.9639535615753994
+0.71828125   0.9638993714622504
+0.71034375   0.9638450955265502
+0.70240625   0.9638059849372029
+0.69446875   0.9637577866497685
+0.68653125   0.9636980049798857
+0.67859375   0.9636324429465546
+0.67065625   0.9635611509667682
+0.66271875   0.9634841701965359
+0.65478125   0.9634025293866947
+0.64684375   0.963330646310048
+0.63890625   0.9632707432324515
+0.63096875   0.9632034377171873
+0.62303125   0.9631275938242613
+0.61509375   0.963044092611668
+0.6071562500000001   0.9629610203510949
+0.59921875   0.9628686842131967
+0.59128125   0.9627672336670928
+0.58334375   0.9626563500660463
+0.57540625   0.962536023518184
+0.56746875   0.962420491607416
+0.55953125   0.9622991897704051
+0.55159375   0.9621680519091717
+0.54365625   0.9620267304304034
+0.53571875   0.9618756339504292
+0.52778125   0.961713923059125
+0.51984375   0.9615400932428618
+0.51190625   0.9613493372438985
+0.50396875   0.9611404267958391
+0.49603125000000003   0.9609165495281505
+0.48809375   0.9606885428813832
+0.48015625   0.9604565399427599
+0.47221875   0.9602097371188835
+0.46428125   0.9599469049699622
+0.45634375   0.9596665062472561
+0.44840625   0.959366632496811
+0.44046875   0.9590449217547321
+0.43253125000000003   0.9586984513785973
+0.42459375   0.958323597797175
+0.41665625   0.9579158517374129
+0.40871875   0.957513439179302
+0.40078125   0.9571143302567431
+0.39284375   0.9566899501927818
+0.38490625   0.9562374521378395
+0.37696875   0.955753466698022
+0.36903125000000003   0.9552340223880293
+0.36109375   0.9546744202766487
+0.35315625   0.9540690636607255
+0.34521875   0.9534112769062083
+0.33728125   0.9526930564699306
+0.32934375   0.9519047613605006
+0.32140625   0.9510122522654944
+0.31346875   0.9499836306702947
+0.30553125000000003   0.9488385276983663
+0.29759375   0.947556941139438
+0.28965625   0.9461514819098414
+0.28171875   0.9446168768590487
+0.27378125   0.9428482229388179
+0.26584375   0.9407955463080119
+0.25790625   0.9383916980923374
+0.24996875000000002   0.9355501727743388
+0.24203125   0.9321597717436331
+0.23409375   0.9280749815436454
+0.22615625   0.92409832729004
+0.21821875000000002   0.9199603312644494
+0.21028125   0.91529397107669
+0.20234375   0.9099618854041043
+0.19440625   0.9038215180307873
+0.18646875000000002   0.8967189006591215
+0.17853125   0.8884624483245174
+0.17059375   0.8788109964569801
+0.16265625   0.8674551794353741
+0.15471875000000002   0.8539938886343238
+0.14678125   0.8379068574130993
+0.13884375   0.8184845893551922
+0.13090625   0.8002555708629069
+0.12296875   0.7830705389169234
+0.11503125   0.7637560194052709
+0.10709375   0.7404082311824085
+0.09915625   0.7105735141347064
+0.09121875   0.6794464330052945
+0.08328125   0.6430006959339014
+0.07534375   0.6061079752629455
+0.06740625   0.5659042020226758
+0.05946875   0.5277315266129229
+0.05153125   0.4859035789367027
+0.04359375   0.4445176269133653
+0.03565625   0.39944274497494076
+0.02771875   0.31052209885431414
+0.01978125   0.22160145273368756
+0.01184375   0.13268080661306095
+0.00390625   0.04376016049243435
+0.0   0.0

outputs/profile_raq_ra5.0_fkt100000000.0_fkv50.0.txt

@@ -0,0 +1,128 @@
+1.0   1.0
+0.99609375   1.0004886307302538
+0.9881562500000001   1.0014815283741296
+0.98021875   1.000804723822155
+0.97228125   1.000350293388824
+0.96434375   1.0002550325461768
+0.95640625   1.0008654380922966
+0.94846875   1.0013652302040426
+0.94053125   1.001766314058726
+0.93259375   1.0020883822552678
+0.92465625   1.0023296302760354
+0.91671875   1.002461872518664
+0.90878125   1.0020987597468052
+0.90084375   1.0018112935862176
+0.89290625   1.0016318727877296
+0.88496875   1.0014263339050131
+0.87703125   1.0013376768427356
+0.86909375   1.0013275735204263
+0.8611562500000001   1.0013684925815804
+0.85321875   1.0014425840323204
+0.84528125   1.0016010617709805
+0.83734375   1.001813270476308
+0.82940625   1.0020246870888012
+0.82146875   1.0022308677315883
+0.81353125   1.0024308604725698
+0.80559375   1.0026193535512236
+0.79765625   1.0027957884153567
+0.78971875   1.0029601417827056
+0.78178125   1.0031127593519735
+0.77384375   1.0032561116829861
+0.76590625   1.0033889550368034
+0.75796875   1.0035116379965685
+0.75003125   1.0036248264745116
+0.74209375   1.0037291232564567
+0.7341562500000001   1.003820214014638
+0.72621875   1.0039010969261741
+0.71828125   1.0039746801239253
+0.71034375   1.0040414574335275
+0.70240625   1.0041018075512984
+0.69446875   1.0041560013675555
+0.68653125   1.0042044597671895
+0.67859375   1.0042474495055858
+0.67065625   1.0042852082893852
+0.66271875   1.0043164962792892
+0.65478125   1.0043387000734143
+0.64684375   1.0043610723879577
+0.63890625   1.004381003538
+0.63096875   1.004398923604277
+0.62303125   1.0044144376406412
+0.61509375   1.0044304460366351
+0.6071562500000001   1.0044636210329727
+0.59921875   1.0044923790380103
+0.59128125   1.004520233263561
+0.58334375   1.0045887493520003
+0.57540625   1.004679199743899
+0.56746875   1.004779877814393
+0.55953125   1.0048788693938613
+0.55159375   1.004977380138184
+0.54365625   1.0050756035254627
+0.53571875   1.0051737376253638
+0.52778125   1.005271987317915
+0.51984375   1.0053779926972155
+0.51190625   1.0054960240256412
+0.50396875   1.0055879712527709
+0.49603125000000003   1.0056687630323906
+0.48809375   1.0057530411581908
+0.48015625   1.0058562629680423
+0.47221875   1.005955859588927
+0.46428125   1.006049142126516
+0.45634375   1.0061392284693913
+0.44840625   1.0062227960462489
+0.44046875   1.0063014876598555
+0.43253125000000003   1.0063809537114619
+0.42459375   1.0064477385412693
+0.41665625   1.0065002233371656
+0.40871875   1.0065340278356054
+0.40078125   1.0065449910239257
+0.39284375   1.0065604716569034
+0.38490625   1.006545761987049
+0.37696875   1.006514220912114
+0.36903125000000003   1.006445797187465
+0.36109375   1.0063420660756848
+0.35315625   1.0061835811678699
+0.34521875   1.005955216668348
+0.33728125   1.0056493725988478
+0.32934375   1.0052531024438225
+0.32140625   1.0047476614201345
+0.31346875   1.0041090171355473
+0.30553125000000003   1.0033061411151785
+0.29759375   1.0022965449658188
+0.28965625   1.0010287716820465
+0.28171875   0.9994315147840812
+0.27378125   0.9971542852576364
+0.26584375   0.993085846984152
+0.25790625   0.9874956979334749
+0.24996875000000002   0.9803651503732914
+0.24203125   0.9712143155094474
+0.23409375   0.9593832352432593
+0.22615625   0.9448093523254151
+0.21821875000000002   0.9281858239988381
+0.21028125   0.9072652738070205
+0.20234375   0.8810591896616389
+0.19440625   0.8536958985978688
+0.18646875000000002   0.8242217024249135
+0.17853125   0.7970041482379427
+0.17059375   0.7680697876685222
+0.16265625   0.7341301061899966
+0.15471875000000002   0.7017719756017162
+0.14678125   0.6716856076715821
+0.13884375   0.6404048600148078
+0.13090625   0.6066679693481009
+0.12296875   0.573799169182174
+0.11503125   0.541222476564494
+0.10709375   0.5084939725792854
+0.09915625   0.472769769206234
+0.09121875   0.43567158985200666
+0.08328125   0.39883182158072766
+0.07534375   0.36379378245712773
+0.06740625   0.32625680593753936
+0.05946875   0.286219165783732
+0.05153125   0.2471830364120577
+0.04359375   0.20984231755040794
+0.03565625   0.1735799645976791
+0.02771875   0.13493902594052706
+0.01978125   0.096298087283375
+0.01184375   0.05765714862622295
+0.00390625   0.019016209969070892
+0.0   0.0

outputs/profile_raq_ra7.5_fkt1000000000.0_fkv25.0.txt

@@ -0,0 +1,128 @@
+1.0   1.0
+0.99609375   0.9990062610422076
+0.9881562500000001   0.9969869834799735
+0.98021875   0.9971232601553087
+0.97228125   0.9971806693281854
+0.96434375   0.9971845844770447
+0.95640625   0.9971461156385902
+0.94846875   0.9970699944828866
+0.94053125   0.9969707623390985
+0.93259375   0.9967311364452428
+0.92465625   0.9962571288038881
+0.91671875   0.9957798887799661
+0.90878125   0.9953726290532506
+0.90084375   0.9949482003998925
+0.89290625   0.9946372293678932
+0.88496875   0.994350193789516
+0.87703125   0.994082398444334
+0.86909375   0.9938673459886528
+0.8611562500000001   0.9936926261460569
+0.85321875   0.993549103310679
+0.84528125   0.9934299598039326
+0.83734375   0.9933408877236217
+0.82940625   0.9932748167728084
+0.82146875   0.9932240593252849
+0.81353125   0.9931845996072571
+0.80559375   0.9931535642345825
+0.79765625   0.9931284177384515
+0.78971875   0.9930763882075305
+0.78178125   0.993027270660629
+0.77384375   0.9929824997446136
+0.76590625   0.9929414748254256
+0.75796875   0.9929041445027907
+0.75003125   0.9928710699506768
+0.74209375   0.9928408956409854
+0.7341562500000001   0.9928062295169527
+0.72621875   0.9927731526457385
+0.71828125   0.9927418727153093
+0.71034375   0.9927124057458755
+0.70240625   0.9926843616027358
+0.69446875   0.9926577214129114
+0.68653125   0.9926324444557905
+0.67859375   0.992608483120658
+0.67065625   0.9925858197309926
+0.66271875   0.9925644421671426
+0.65478125   0.9925443623714286
+0.64684375   0.9925255878718212
+0.63890625   0.9925081328548363
+0.63096875   0.9924920175575421
+0.62303125   0.9924772674977019
+0.61509375   0.9924639128627663
+0.6071562500000001   0.9924519880277204
+0.59921875   0.9924414619655849
+0.59128125   0.9924248754213431
+0.58334375   0.9924091294843468
+0.57540625   0.9923949509101252
+0.56746875   0.9924055373747572
+0.55953125   0.9924226953546874
+0.55159375   0.9924412279591394
+0.54365625   0.9924609181073719
+0.53571875   0.9924822080796133
+0.52778125   0.992505191275782
+0.51984375   0.9925299638594424
+0.51190625   0.9925564376035475
+0.50396875   0.9925860465898036
+0.49603125000000003   0.9926185812893703
+0.48809375   0.9926533574477424
+0.48015625   0.9926880569201405
+0.47221875   0.9927248793493549
+0.46428125   0.9927643825455995
+0.45634375   0.9928067052937776
+0.44840625   0.992851761327506
+0.44046875   0.9928997556605149
+0.43253125000000003   0.9929506220835563
+0.42459375   0.9930046883440068
+0.41665625   0.9930610831698342
+0.40871875   0.9931189143921962
+0.40078125   0.9931812242068735
+0.39284375   0.9932488769681517
+0.38490625   0.9933222601182629
+0.37696875   0.9934017804362069
+0.36903125000000003   0.9935076371764416
+0.36109375   0.9936212772387668
+0.35315625   0.9937408126620867
+0.34521875   0.9938671203998567
+0.33728125   0.9940008199749039
+0.32934375   0.9941429866018373
+0.32140625   0.9942950177741342
+0.31346875   0.9944506020368925
+0.30553125000000003   0.994619583716713
+0.29759375   0.9947970240990528
+0.28965625   0.9949834978202933
+0.28171875   0.9951895224341948
+0.27378125   0.9954192703762218
+0.26584375   0.9956510966996391
+0.25790625   0.9958843935290972
+0.24996875000000002   0.9961171663976424
+0.24203125   0.9963453467837434
+0.23409375   0.9965568266194059
+0.22615625   0.99674747819695
+0.21821875000000002   0.9969482057070408
+0.21028125   0.9971208387999916
+0.20234375   0.9967090731311209
+0.19440625   0.9950787777951174
+0.18646875000000002   0.9930116113720479
+0.17853125   0.9902464116431152
+0.17059375   0.986576923533956
+0.16265625   0.9810600067528333
+0.15471875000000002   0.9733191107513938
+0.14678125   0.9633249834854362
+0.13884375   0.9503717984851009
+0.13090625   0.933303531509786
+0.12296875   0.9103604184926574
+0.11503125   0.8787642810144055
+0.10709375   0.8387919694348168
+0.09915625   0.7853416959477463
+0.09121875   0.715816483333058
+0.08328125   0.6491644646132442
+0.07534375   0.5822779840859292
+0.06740625   0.5214566188913672
+0.05946875   0.46451158897336337
+0.05153125   0.40186241330496747
+0.04359375   0.3343255941714209
+0.03565625   0.2648353775338124
+0.02771875   0.20587991224582963
+0.01978125   0.14692444695784684
+0.01184375   0.08796898166986407
+0.00390625   0.02901351638188129
+0.0   0.0

preprocess/preprocess_profiles.ipynb

@@ -0,0 +1,470 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "8b63d618-1f46-49e8-b388-c4185624e58c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os \n",
+    "import torch\n",
+    "import numpy as np\n",
+    "from tabulate import tabulate\n",
+    "import random\n",
+    "from matplotlib import pyplot as plt\n",
+    "import pickle\n",
+    "from scipy.signal import find_peaks"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "2129fb21-fcea-40b4-ba3c-473aa2e6f1e2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pre = \"../data/\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "id": "ae2009a3-a989-4756-849a-1180bb7ba087",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [],
+   "source": [
+    "load_sims = False\n",
+    "\n",
+    "x = {}\n",
+    "y = {}\n",
+    "var_vec = [\"Tprev\"]\n",
+    "for var in var_vec:\n",
+    "    x[var] = {}\n",
+    "    y[var] = {}\n",
+    "    \n",
+    "for var in var_vec:\n",
+    "    with open(pre + 'x_' + var + '.pkl', 'rb') as file: \n",
+    "        x[var] = pickle.load(file) \n",
+    "    with open(pre + 'y_' + var + '.pkl', 'rb') as file: \n",
+    "        y[var] = pickle.load(file) \n",
+    "        "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "bff0409c-0adb-42df-96e3-2ace88051f6e",
+   "metadata": {},
+   "source": [
+    "### Identify train/cv/test"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 30,
+   "id": "22cbeaae-7421-4ae9-8042-079591939e07",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "97\n",
+      "15\n",
+      "16\n"
+     ]
+    }
+   ],
+   "source": [
+    "sims = torch.load(pre + \"/sims.pt\")\n",
+    "\n",
+    "extrapolation_sims = []\n",
+    "interpolation_sims = []\n",
+    "for si, sim in enumerate(sims):\n",
+    "    if si !=39 and si!=8: # and (si==0 or si==100 or si==120):\n",
+    "        #print(tabulate([[\"num\", \"dataset\", \"raq\", \"fkt\", \"fkp\", \"gr\", \"ar\"],\n",
+    "        #                sim[:-1]\n",
+    "        #               ]))\n",
+    "        ignr, ignr, raq, fkt, fkp, gr, ar, ignr = sim\n",
+    "\n",
+    "        #if (fkt < 5e+5 or fkt > 5e+8) and (fkp < 15 or fkp > 85) and (raq < 1.5 or raq > 8.5):\n",
+    "        if (fkt > 5e+9) or (fkp > 95) or (raq > 9.5):\n",
+    "            extrapolation_sims.append(si)\n",
+    "        else:\n",
+    "            interpolation_sims.append(si)\n",
+    "\n",
+    "random.seed(1992)\n",
+    "inds = {}\n",
+    "inds[\"test\"] = extrapolation_sims #+ random.choices(interpolation_sims, k=9)\n",
+    "\n",
+    "remain_inds = []\n",
+    "for inp in interpolation_sims:\n",
+    "    if inp not in inds[\"test\"]:\n",
+    "        remain_inds.append(inp)\n",
+    "        \n",
+    "inds[\"cv\"] = random.choices(remain_inds, k=16)\n",
+    "\n",
+    "inds[\"train\"] = []\n",
+    "for inp in remain_inds:\n",
+    "    if inp not in inds[\"test\"] and inp not in inds[\"cv\"]:\n",
+    "        inds[\"train\"].append(inp)\n",
+    "\n",
+    "inds[\"train\"] = np.unique(inds[\"train\"])\n",
+    "inds[\"cv\"] = np.unique(inds[\"cv\"])\n",
+    "inds[\"test\"] = np.unique(inds[\"test\"])\n",
+    "\n",
+    "print(len(inds[\"train\"]))#, sorted(inds[\"train\"]))\n",
+    "print(len(inds[\"cv\"]))#, sorted(inds[\"cv\"]))\n",
+    "print(len(inds[\"test\"]))#, sorted(inds[\"test\"]))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7ade165c-b896-4192-a4e6-d4ac96a6dd07",
+   "metadata": {},
+   "source": [
+    "### Write simulation parameters "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "id": "f3ec1278-58cb-445a-a8e4-112d0032c784",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "sims_table = [[\"Simulation\", \"Dataset\", \"RaQ/Ra\", \"FKT\", \"FKV\"]] \n",
+    "\n",
+    "for sim in sims:\n",
+    "    if sim[0] in inds[\"train\"]:\n",
+    "        an = \"train\"\n",
+    "    elif sim[0] in inds[\"cv\"]:\n",
+    "        an = \"cv\"\n",
+    "    elif sim[0] in inds[\"test\"]:\n",
+    "        an = \"test\"\n",
+    "    sims_table.append([sim[0], an, sim[2], sim[3], sim[4]])\n",
+    "\n",
+    "with open('../inputs/simulations.txt', 'w') as f:\n",
+    "    f.write(tabulate(sims_table))\n",
+    "\n",
+    "with open('../Paper/simulations.pkl', 'wb') as f:\n",
+    "    pickle.dump(sims_table, f)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4f93a6bb-3b87-4822-a58a-a105ca049c0c",
+   "metadata": {},
+   "source": [
+    "### Pointwise input preparation"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 34,
+   "id": "97e5f984-6dc9-4ec9-b770-3ca1a721f67c",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(26966, 4) (26966, 1)\n",
+      "(4170, 4) (4170, 1)\n",
+      "(4448, 4) (4448, 1)\n",
+      "(12416, 4) (12416, 1)\n",
+      "(1920, 4) (1920, 1)\n",
+      "(2048, 4) (2048, 1)\n"
+     ]
+    }
+   ],
+   "source": [
+    "x_pointwise = {}\n",
+    "y_pointwise = {}\n",
+    "\n",
+    "y_prof = torch.load(pre + \"/y_prof.pt\").flatten().numpy()\n",
+    "\n",
+    "y_prof = y_prof[::-1]\n",
+    "y_new = np.sort(np.concatenate((np.linspace(1,y_prof[15],100), \n",
+    "                                y_prof, np.linspace(y_prof[-10],y_prof[-1],50)), axis=0))[::-1]\n",
+    "\n",
+    "\n",
+    "for an in [\"train\", \"cv\", \"test\"]:\n",
+    "    x_pointwise[an] = np.zeros((len(inds[an])*y_new.shape[0], 4))\n",
+    "    y_pointwise[an] = np.zeros((len(inds[an])*y_new.shape[0], 1))\n",
+    "    \n",
+    "\n",
+    "    cntr = 0\n",
+    "    \n",
+    "    for i in inds[an]:\n",
+    "        #print(an, i)\n",
+    "        #u = y[\"uprev\"][i]*20\n",
+    "        #v = y[\"vprev\"][i]*20\n",
+    "        #vmag = np.sqrt(u[-50:,:]**2 + v[-50:,:]**2)\n",
+    "        #vmag = np.mean(vmag, axis=0)\n",
+    "        T = np.mean(y[\"Tprev\"][i], axis=0)\n",
+    "        T_new = np.interp(y_new, y_prof[::-1], T[::-1])\n",
+    "\n",
+    "        #plt.figure()\n",
+    "        #plt.plot(T, y_prof)\n",
+    "        #plt.plot(T_new, y_new, 'kx')\n",
+    "        #plt.ylim([1,0])\n",
+    "        #plt.show()\n",
+    "\n",
+    "        for j in range(y_new.shape[0]):\n",
+    "            x_pointwise[an][cntr,:3] = x[\"Tprev\"][i]\n",
+    "            x_pointwise[an][cntr,3:4] = y_new[j]\n",
+    "            \n",
+    "            y_pointwise[an][cntr,0]   = T_new[j]\n",
+    "            cntr += 1        \n",
+    "    print(x_pointwise[an].shape, y_pointwise[an].shape)\n",
+    "\n",
+    "\n",
+    "\n",
+    "with open(pre + 'x_pointwise.pkl', 'wb') as file: \n",
+    "    pickle.dump(x_pointwise, file) \n",
+    "with open(pre + 'y_pointwise.pkl', 'wb') as file: \n",
+    "    pickle.dump(y_pointwise, file) \n",
+    "\n",
+    "\n",
+    "for an in [\"train\", \"cv\", \"test\"]:\n",
+    "    x_pointwise[an] = np.zeros((len(inds[an])*y_prof.shape[0], 4))\n",
+    "    y_pointwise[an] = np.zeros((len(inds[an])*y_prof.shape[0], 1))\n",
+    "    \n",
+    "    cntr = 0\n",
+    "    \n",
+    "    for i in inds[an]:\n",
+    "        T = np.mean(y[\"Tprev\"][i], axis=0)\n",
+    "\n",
+    "        for j in range(y_prof.shape[0]):\n",
+    "            x_pointwise[an][cntr,:3] = x[\"Tprev\"][i]\n",
+    "            x_pointwise[an][cntr,3:4] = y_prof[j]\n",
+    "            \n",
+    "            y_pointwise[an][cntr,0]   = T[j]\n",
+    "            cntr += 1        \n",
+    "    print(x_pointwise[an].shape, y_pointwise[an].shape)\n",
+    "\n",
+    "\n",
+    "\n",
+    "with open(pre + 'x_pointwise_orgres.pkl', 'wb') as file: \n",
+    "    pickle.dump(x_pointwise, file) \n",
+    "with open(pre + 'y_pointwise_orgres.pkl', 'wb') as file: \n",
+    "    pickle.dump(y_pointwise, file) "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "id": "2875ca3d-079b-4bda-ad0d-29081aa82fee",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "### Full profile input"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 36,
+   "id": "40d86b1d-8db5-405c-a305-f7fb23c297dc",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "train 0\n",
+      "train 3\n",
+      "train 4\n",
+      "train 5\n",
+      "train 6\n",
+      "train 7\n",
+      "train 9\n",
+      "train 10\n",
+      "train 11\n",
+      "train 12\n",
+      "train 13\n",
+      "train 14\n",
+      "train 16\n",
+      "train 18\n",
+      "train 19\n",
+      "train 20\n",
+      "train 21\n",
+      "train 22\n",
+      "train 23\n",
+      "train 24\n",
+      "train 25\n",
+      "train 26\n",
+      "train 27\n",
+      "train 28\n",
+      "train 29\n",
+      "train 30\n",
+      "train 31\n",
+      "train 33\n",
+      "train 34\n",
+      "train 35\n",
+      "train 36\n",
+      "train 37\n",
+      "train 41\n",
+      "train 43\n",
+      "train 44\n",
+      "train 45\n",
+      "train 46\n",
+      "train 47\n",
+      "train 48\n",
+      "train 49\n",
+      "train 50\n",
+      "train 51\n",
+      "train 52\n",
+      "train 53\n",
+      "train 54\n",
+      "train 56\n",
+      "train 61\n",
+      "train 62\n",
+      "train 63\n",
+      "train 64\n",
+      "train 65\n",
+      "train 66\n",
+      "train 67\n",
+      "train 70\n",
+      "train 71\n",
+      "train 72\n",
+      "train 73\n",
+      "train 74\n",
+      "train 75\n",
+      "train 78\n",
+      "train 79\n",
+      "train 80\n",
+      "train 81\n",
+      "train 82\n",
+      "train 84\n",
+      "train 88\n",
+      "train 89\n",
+      "train 90\n",
+      "train 91\n",
+      "train 96\n",
+      "train 97\n",
+      "train 99\n",
+      "train 100\n",
+      "train 101\n",
+      "train 102\n",
+      "train 103\n",
+      "train 104\n",
+      "train 106\n",
+      "train 107\n",
+      "train 108\n",
+      "train 109\n",
+      "train 110\n",
+      "train 111\n",
+      "train 113\n",
+      "train 114\n",
+      "train 115\n",
+      "train 116\n",
+      "train 117\n",
+      "train 119\n",
+      "train 120\n",
+      "train 121\n",
+      "train 123\n",
+      "train 124\n",
+      "train 125\n",
+      "train 126\n",
+      "train 128\n",
+      "train 129\n",
+      "(97, 3) (97, 128)\n",
+      "cv 2\n",
+      "cv 17\n",
+      "cv 32\n",
+      "cv 38\n",
+      "cv 40\n",
+      "cv 57\n",
+      "cv 59\n",
+      "cv 60\n",
+      "cv 76\n",
+      "cv 83\n",
+      "cv 92\n",
+      "cv 95\n",
+      "cv 98\n",
+      "cv 105\n",
+      "cv 122\n",
+      "(15, 3) (15, 128)\n",
+      "test 1\n",
+      "test 15\n",
+      "test 42\n",
+      "test 55\n",
+      "test 58\n",
+      "test 68\n",
+      "test 69\n",
+      "test 77\n",
+      "test 85\n",
+      "test 86\n",
+      "test 87\n",
+      "test 93\n",
+      "test 94\n",
+      "test 112\n",
+      "test 118\n",
+      "test 127\n",
+      "(16, 3) (16, 128)\n"
+     ]
+    }
+   ],
+   "source": [
+    "x_p = {}\n",
+    "y_p = {}\n",
+    "\n",
+    "for an in [\"train\", \"cv\", \"test\"]:\n",
+    "    x_p[an] = np.zeros((len(inds[an]),3))\n",
+    "    y_p[an] = np.zeros((len(inds[an]),128))\n",
+    "\n",
+    "    cntr = 0\n",
+    "    for i in inds[an]:\n",
+    "        print(an, i)\n",
+    "        T = np.mean(y[\"Tprev\"][i], axis=0)\n",
+    "\n",
+    "        x_p[an][cntr,:] = x[\"Tprev\"][i]\n",
+    "        y_p[an][cntr,:] = T\n",
+    "        cntr += 1        \n",
+    "    \n",
+    "    print(x_p[an].shape, y_p[an].shape)\n",
+    "\n",
+    "\n",
+    "\n",
+    "with open(pre + 'x_p.pkl', 'wb') as file: \n",
+    "    pickle.dump(x_p, file) \n",
+    "with open(pre + 'y_p.pkl', 'wb') as file: \n",
+    "    pickle.dump(y_p, file) "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "960cc6f5-de5e-4788-aaaa-65e06aa33a42",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.10.9"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

stats/MLP_stats.txt

@@ -1,23 +1,23 @@
-------------  ---------  -------  -------  ---------
-architecture  mae train  mae cv   diff     diff + cv
-[32, 2]       0.01158    0.01249  0.00091  0.01294
-[64, 2]       0.01077    0.01379  0.00302  0.0153
-[128, 2]      0.00997    0.01177  0.00179  0.01266
-[256, 2]      0.01       0.01226  0.00226  0.01339
-[32, 3]       0.00922    0.01167  0.00245  0.01289
-[64, 3]       0.00878    0.01149  0.00271  0.01284
-[128, 3]      0.00834    0.01009  0.00175  0.01097
-[256, 3]      0.0082     0.00927  0.00108  0.00981
-[32, 4]       0.00797    0.00915  0.00118  0.00974
-[64, 4]       0.00652    0.00821  0.00169  0.00905
-[128, 4]      0.00592    0.0083   0.00237  0.00948
-[256, 4]      0.00681    0.00794  0.00113  0.0085
-[32, 5]       0.00877    0.01209  0.00332  0.01375
-[64, 5]       0.00584    0.0084   0.00256  0.00968
-[128, 5]      0.00608    0.00857  0.00249  0.00981
-[256, 5]      0.00682    0.00814  0.00133  0.00881
-[32, 6]       0.00799    0.01453  0.00654  0.0178
-[64, 6]       0.0118     0.01206  0.00026  0.01219
-[128, 6]      0.00607    0.00949  0.00342  0.0112
-[256, 6]      0.00738    0.01015  0.00277  0.01153
-------------  ---------  -------  -------  ---------
+------------  ---------  -------  --------  ---------
+architecture  mae train  mae cv   diff      diff + cv
+[32, 2]       0.01167    0.01348  0.00181   0.01438
+[64, 2]       0.01003    0.01141  0.00138   0.01211
+[128, 2]      0.00979    0.0115   0.00171   0.01236
+[256, 2]      0.00975    0.01108  0.00133   0.01174
+[32, 3]       0.00849    0.00934  0.00085   0.00976
+[64, 3]       0.00601    0.00769  0.00167   0.00852
+[128, 3]      0.00572    0.00727  0.00155   0.00805
+[256, 3]      0.00624    0.00729  0.00105   0.00782
+[32, 4]       0.00642    0.00773  0.00131   0.00838
+[64, 4]       0.00513    0.00599  0.00086   0.00642
+[128, 4]      0.00439    0.00553  0.00114   0.0061
+[256, 4]      0.0045     0.00584  0.00133   0.0065
+[32, 5]       0.00709    0.00935  0.00226   0.01048
+[64, 5]       0.00741    0.00838  0.00097   0.00886
+[128, 5]      0.00442    0.00541  0.00098   0.0059
+[256, 5]      0.00469    0.0071   0.00241   0.00831
+[32, 6]       0.01028    0.01264  0.00236   0.01383
+[64, 6]       0.01263    0.00999  -0.00264  0.00999
+[128, 6]      0.00277    0.00861  0.00584   0.01154
+[256, 6]      0.00694    0.00751  0.00057   0.00779
+------------  ---------  -------  --------  ---------

{data → train}/mlp.py

@@ -8,13 +8,22 @@ from matplotlib.lines import Line2D
 import math
 
 def get_lr(optimizer):
+    '''Function to get learning rate'''
     for param_group in optimizer.param_groups:
         return param_group['lr']
 
 class MLP(nn.Module):
     def __init__(self, f_i: int, f_o: int, act_fn: object = nn.SELU, f=[], insert_in=[4], freq_encoding=False):
-        
         super().__init__()
+        '''
+        Feedforward neural network model in Pytorch.
+        f_i: input filters
+        f_o: output dimension
+        act_fn: activate function
+        f: list of filters in hidden layers
+        insert_in: if insert inputs into layer numbers
+        freq_encoding: if use frequency encoding
+        '''
 
         self.insert_in = insert_in
         self.layers = nn.ModuleList()
@@ -63,6 +72,7 @@ class MLP(nn.Module):
         return x
 
 def one_epoch_mlp(mlp, epoch, loader, optimizer, device, is_train=False):
+    '''Function to run an epoch in Pytorch with a standard optimizer'''
     running_loss = 0.
     counter = 1
     loss_fn = torch.nn.L1Loss() #reduction="none")
@@ -94,6 +104,7 @@ def one_epoch_mlp(mlp, epoch, loader, optimizer, device, is_train=False):
     return running_loss/counter
 
 def one_epoch_mlp_lbfgs(mlp, epoch, loader, optimizer, device, is_train=False):
+    '''Function to run an epoch in Pytorch with a lbfgs'''
     running_loss = 0.
     counter = 1
     loss_fn = torch.nn.L1Loss() #reduction="none")
@@ -181,6 +192,7 @@ class Siren(nn.Module):
         out = self.dropout(out)
         return out
 
+# Siren network
 class SirenMLP(nn.Module):
 
     def __init__(self,

{data → train}/train_profiles_mlp.py

@@ -6,8 +6,8 @@ import matplotlib.pyplot as plt
 import torch
 from mlp import *
 import argparse
-from datasetio import *
-from torch.utils.data import TensorDataset
+#from datasetio import *
+from torch.utils.data import TensorDataset, DataLoader
 
 import copy
 import pickle
@@ -16,8 +16,7 @@ import time
 # In[ ]:
 
 
-data_dir = "/plp_scr1/agar_sh/data/TPH/"
-nn_dir = "/plp_user/agar_sh/PBML/pytorch/TPH/MLP/trained_networks/"
+nn_dir = "./"
 
 
 # In[ ]:
@@ -26,7 +25,7 @@ nn_dir = "/plp_user/agar_sh/PBML/pytorch/TPH/MLP/trained_networks/"
 run_cell = True
 if run_cell:
     parser = argparse.ArgumentParser(description='Train mlp')
-    parser.add_argument("-gpu", "--gpu_number", type=int, help="specify gpu number")
+    parser.add_argument("-gpu", "--gpu_number", type=int, help="specify gpu number", default=-1)
     parser.add_argument("-a", "--act_fn", type=str, help ="activation function")
     parser.add_argument("-l", "--num_layers", type=int, help ="activation function")
     parser.add_argument("-f", "--f_h", type=int, help ="filters")
@@ -53,7 +52,10 @@ f_nn = "mlp_profile_pointwise_" + str(f_h) + "_" + str(num_layers) + "_" + act_f
 if not os.path.isdir(nn_dir + f_nn):
     os.mkdir(nn_dir + f_nn)
 
-device = torch.device("cuda:" + str(gpu_number)) if torch.cuda.is_available() else torch.device("cpu")
+if gpu_number >=0 and torch.cuda.is_available():
+    device = torch.device("cuda:" + str(gpu_number))
+else:
+    device = torch.device("cpu")
 
 epoch         = 0
 start_lr      = 1e-3
@@ -80,7 +82,7 @@ with open(nn_dir + "mlp.txt", 'w') as writer:
 dataset = {}
 loader = {}
 batches = {}
-pre = "/plp_user/agar_sh/PBML/pytorch/TPH/MLP/profiles/"
+pre = "../data/"
 with open(pre + 'x_pointwise.pkl', 'rb') as file: 
     x_pointwise = pickle.load(file) 
 with open(pre + 'y_pointwise.pkl', 'rb') as file: 

train/trained_networks/mlp_profile_[128]_2_selu/mlp.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47bf6d6c36d171d50ada7c24a10f740164f712c846a6028bf58f1ddd636dabae
+size 267488

train/trained_networks/mlp_profile_[128]_3_selu/mlp.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35dbc140c6235cf3c1dcb2d820838d75eaa4cb67fde44069ec7fb9b22c299ce6
+size 400016