changed to linear transformation and fixed support on uniform prior

desResLab · Mar 28, 2024 · a3beb4c · a3beb4c
1 parent b77eae5
commit a3beb4c
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 11 deletions.
diff --git a/linfa/tests/TP1_uniform_prior.py b/linfa/tests/TP1_uniform_prior.py
@@ -53,8 +53,8 @@ def run_test():
  exp.device = torch.device('cuda:0' if torch.cuda.is_available() and not exp.no_cuda else 'cpu')
 
  # Define transformation
- trsf_info = [['tanh', -20.0, 20.0, 500.0, 1500.0],
- ['tanh', -20.0, 20.0, -30000.0, -15000.0]]
+ trsf_info = [['linear', -20.0, 20.0, 500.0, 1500.0],
+ ['linear', -20.0, 20.0, -30000.0, -15000.0]]
  trsf = Transformation(trsf_info)
 
  # Apply the transformation

diff --git a/linfa/tests/TP1_uniform_prior_no_disc.py b/linfa/tests/TP1_uniform_prior_no_disc.py
@@ -53,8 +53,8 @@ def run_test():
  exp.device = torch.device('cuda:0' if torch.cuda.is_available() and not exp.no_cuda else 'cpu')
 
  # Define transformation
- trsf_info = [['tanh', -20.0, 20.0, 500.0, 1500.0],
- ['tanh', -20.0, 20.0, -30000.0, -15000.0]]
+ trsf_info = [['linear', -20.0, 20.0, 500.0, 1500.0],
+ ['linear', -20.0, 20.0, -30000.0, -15000.0]]
  trsf = Transformation(trsf_info)
 
  # Apply the transformation
@@ -163,14 +163,25 @@ def log_prior(calib_inputs, transform):
  adjust = transform.compute_log_jacob_func(calib_inputs)
  # Compute the calibration inputs in the physical domain
  phys_inputs = transform.forward(calib_inputs)
- # Define prior moments for uniform distribution
+ # Define upper and lower bounds for uniform distribution
  low = torch.tensor([500, -35.0E3])
  high = torch.tensor([1500, -10.0E3])
+ res = [] # Initialize
  # Eval log prior
- l1 = -np.log((high[0] - low[0])*(high[1] - low[1]))
- # Return
- res = l1 + adjust
- return res
+ for loopA, param_pairs in enumerate(phys_inputs):
+ if param_pairs[0] < low[0]:
+ l1 = 0
+ elif param_pairs[0] > high[0]:
+ l1 = 0
+ elif param_pairs[1] < low[1]:
+ l1 = 0
+ elif param_pairs[1] > high[1]:
+ l1 = 0
+ else:
+ l1 = -np.log((high[0] - low[0])*(high[1] - low[1]))
+ # Return
+ res.append(l1 + adjust[loopA])
+ return torch.tensor(res)
 
  exp.model_logprior = lambda x: log_prior(x, exp.transform)
 

diff --git a/run_plot_res.sh b/run_plot_res.sh
@@ -1,8 +1,8 @@
-python3 linfa/plot_res.py --folder results/ --name TP_15_uniform_prior --iter 25000 --picformat png
+python3 linfa/plot_res.py --folder results/ --name TP1_uniform_prior_no_disc --iter 5000 --picformat png
 # python3 linfa/plot_disc.py --folder results/ --name test_08_lf_w_disc_TP1_uniform_prior --iter 25000 --picformat png --mode histograms --num_points 10 --limfactor 1.0 --saveinterval 1000 --dropouts 10
 # python3 linfa/plot_disc.py --folder results/ --name test_19_lf_w_disc_TP15_rep_meas_dropout --iter 10000 --picformat png --mode discr_surface --num_points 10 --limfactor 1.0 --saveinterval 1000
 # python3 linfa/plot_disc.py --folder results/ --name test_08_lf_w_disc_TP1_uniform_prior --iter 25000 --picformat png --mode marginal_stats --num_points 10 --limfactor 1.0 --saveinterval 1000
-python3 linfa/plot_disc.py --folder results/ --name TP_15_uniform_prior --iter 25000 --picformat png --mode marginal_posterior --num_points 10 --limfactor 1.0 --saveinterval 1000
+python3 linfa/plot_disc.py --folder results/ --name TP1_uniform_prior_no_disc --iter 5000 --picformat png --mode marginal_posterior --num_points 10 --limfactor 1.0 --saveinterval 1000