bug fixes with torch.linalg.solve() and default variance calculation

lbl-camera · Sep 14, 2022 · 8b98c33 · 8b98c33
1 parent 00da667
commit 8b98c33
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 8 deletions.
diff --git a/fvgp/gp.py b/fvgp/gp.py
@@ -127,9 +127,9 @@ def __init__(
         #######prepare variances##################
         ##########################################
         if variances is None:
-            self.variances = np.ones((self.y_data.shape)) * abs(np.mean(self.y_data) / 100.0)
+            self.variances = np.ones((self.y_data.shape)) * (np.mean(abs(self.y_data)) / 100.0)
             logger.warning("CAUTION: you have not provided data variances in fvGP, "
-                           "they will be set to 1 percent of the data values!")
+                           "they will be set to 1 percent of the |data values|!")
         elif np.ndim(variances) == 2:
             self.variances = variances[:,0]
         elif np.ndim(variances) == 1:
@@ -198,7 +198,7 @@ def update_gp_data(
         #######prepare variances##################
         ##########################################
         if variances is None:
-            self.variances = np.ones((self.y_data.shape)) * abs(self.y_data / 100.0)
+            self.variances = np.ones((self.y_data.shape)) * (np.mean(abs(self.y_data)) / 100.0)
         elif np.ndim(variances) == 2:
             self.variances = variances[:,0]
         elif np.ndim(variances) == 1:
@@ -525,7 +525,7 @@ def _optimize_log_likelihood(self,starting_hps,
                        radius = deflation_radius,
                        num_epochs = max_iter,
                        constraints = constraints)
-
+            print("gfdsdd: ", np.array(starting_hps).reshape(1,-1))
             obj = opt.optimize(dask_client = dask_client, x0 = np.array(starting_hps).reshape(1,-1))
             res = opt.get_final()
             hyperparameters = res["x"][0]
@@ -731,9 +731,9 @@ def solve(self, A, b):
                     logger.error("torch.solve() on cpu did not work")
                     logger.error("reason: ", str(e))
                     #x, qr = torch.lstsq(b,A)
-                    x, qr = torch.linalg.lstsq(A,b)
+                    x, res, rank, s = torch.linalg.lstsq(A,b)
                 except Exception as e:
-                    logger.error("torch.solve() and torch.lstsq() on cpu did not work; falling back to numpy.lstsq()")
+                    logger.error("torch.solve() and torch.lstsq() on cpu did not work; falling back to numpy.linalg.lstsq()")
                     logger.error("reason: {}", str(e))
                     x,res,rank,s = np.linalg.lstsq(A.numpy(),b.numpy())
                     return x
@@ -750,7 +750,7 @@ def solve(self, A, b):
                     #x, qr = torch.lstsq(b,A)
                     x = torch.linalg.lstsq(A,b)
                 except Exception as e:
-                    logger.error("torch.solve() and torch.lstsq() on gpu did not work; falling back to numpy.lstsq()")
+                    logger.error("torch.solve() and torch.lstsq() on gpu did not work; falling back to numpy.linalg.lstsq()")
                     logger.error("reason: ", str(e))
                     x,res,rank,s = np.linalg.lstsq(A.numpy(),b.numpy())
                     return x

diff --git a/requirements.txt b/requirements.txt
@@ -6,7 +6,7 @@ numpy
 matplotlib
 dask >= 2021.6.2
 distributed >= 2021.6.2
-hgdl == 2.0.2
+hgdl == 2.0.3
 notebook
 plotly
 loguru