Fix incorrect types in pypesto.result.profile.ProfilerResult

pypesto/profile/util.py

@@ -183,11 +183,11 @@ def fill_profile_list(
 
     # create blank profile
     new_profile = ProfilerResult(
-        x_path=optimizer_result["x"],
+        x_path=optimizer_result["x"][..., np.newaxis],
         fval_path=np.array([optimizer_result["fval"]]),
         ratio_path=np.array([np.exp(global_opt - optimizer_result["fval"])]),
-        gradnorm_path=gradnorm,
-        exitflag_path=optimizer_result["exitflag"],
+        gradnorm_path=np.array([gradnorm]),
+        exitflag_path=np.array([optimizer_result["exitflag"]]),
         time_path=np.array([0.0]),
         time_total=0.0,
         n_fval=0,

pypesto/result/profile.py

@@ -52,9 +52,9 @@ def __init__(
         x_path: np.ndarray,
         fval_path: np.ndarray,
         ratio_path: np.ndarray,
-        gradnorm_path: np.ndarray = np.nan,
-        exitflag_path: np.ndarray = np.nan,
-        time_path: np.ndarray = np.nan,
+        gradnorm_path: np.ndarray = None,
+        exitflag_path: np.ndarray = None,
+        time_path: np.ndarray = None,
         time_total: float = 0.0,
         n_fval: int = 0,
         n_grad: int = 0,
@@ -64,19 +64,41 @@ def __init__(
         super().__init__()
 
         # initialize profile path
-        x_shape = x_path.shape
-        if len(x_shape) == 1:
-            self.x_path = np.zeros((x_shape[0], 1))
-            self.x_path[:, 0] = x_path[:]
+        if not x_path.ndim == 2:
+            raise ValueError("x_path must be a 2D array.")
+
+        self.x_path = x_path.copy()
+        self.fval_path = fval_path.copy()
+        self.ratio_path = ratio_path.copy()
+
+        if gradnorm_path is None:
+            self.gradnorm_path = np.full(x_path.shape[1], np.nan)
+        else:
+            self.gradnorm_path = gradnorm_path.copy()
+
+        if exitflag_path is None:
+            self.exitflag_path = np.full(x_path.shape[1], np.nan)
         else:
-            self.x_path = np.zeros((x_shape[0], x_shape[1]))
-            self.x_path[:, :] = x_path[:, :]
-
-        self.fval_path = np.asarray(fval_path)
-        self.ratio_path = np.asarray(ratio_path)
-        self.gradnorm_path = np.asarray(gradnorm_path)
-        self.exitflag_path = np.asarray(exitflag_path)
-        self.time_path = np.asarray(time_path)
+            self.exitflag_path = exitflag_path.copy()
+
+        if time_path is None:
+            self.time_path = np.full(x_path.shape[1], np.nan)
+        else:
+            self.time_path = time_path.copy()
+
+        if (
+            not self.x_path.shape[1]
+            == len(self.fval_path)
+            == len(self.ratio_path)
+            == len(self.gradnorm_path)
+            == len(self.exitflag_path)
+            == len(self.time_path)
+        ):
+            raise ValueError(
+                "x_path, fval_path, ratio_path, gradnorm_path, exitflag_path, "
+                "time_path must have the same length."
+            )
+
         self.time_total = time_total
         self.n_fval = n_fval
         self.n_grad = n_grad
@@ -131,26 +153,12 @@ def append_profile_point(
         n_hess:
             Number of Hessian evaluations performed to find `x`.
         """
-
-        # short function to append to numpy vectors
-        def append_to_vector(field_name, val):
-            field_new = np.zeros(self[field_name].size + 1)
-            field_new[0:-1] = self[field_name]
-            field_new[-1] = val
-            self[field_name] = field_new
-
-        # write profile path
-        x_new = np.zeros((self.x_path.shape[0], self.x_path.shape[1] + 1))
-        x_new[:, 0:-1] = self.x_path
-        x_new[:, -1] = x
-        self.x_path = x_new
-
-        # append to other paths
-        append_to_vector("fval_path", fval)
-        append_to_vector("ratio_path", ratio)
-        append_to_vector("gradnorm_path", gradnorm)
-        append_to_vector("exitflag_path", exitflag)
-        append_to_vector("time_path", time)
+        self.x_path = np.hstack((self.x_path, x[..., np.newaxis]))
+        self.fval_path = np.hstack((self.fval_path, fval))
+        self.ratio_path = np.hstack((self.ratio_path, ratio))
+        self.gradnorm_path = np.hstack((self.gradnorm_path, gradnorm))
+        self.exitflag_path = np.hstack((self.exitflag_path, exitflag))
+        self.time_path = np.hstack((self.time_path, time))
 
         # increment the time and f_eval counters
         self.time_total += time

pypesto/store/read_from_hdf5.py

@@ -32,7 +32,7 @@ def read_hdf5_profile(
         specifies the profile index that is read
         from the HDF5 file
     """
-    result = ProfilerResult(np.array([]), np.array([]), np.array([]))
+    result = ProfilerResult(np.empty((0, 0)), np.array([]), np.array([]))
 
     for profile_key in result.keys():
         if profile_key in f[f'/profiling/{profile_id}/{parameter_id}']:

test/visualize/test_visualize.py

@@ -178,8 +178,8 @@ def create_profile_result():
     result = create_optimization_result()
 
     # write some dummy results for profiling
-    ratio_path_1 = [0.15, 0.25, 0.7, 1.0, 0.8, 0.35, 0.15]
-    ratio_path_2 = [0.1, 0.2, 0.7, 1.0, 0.8, 0.3, 0.1]
+    ratio_path_1 = np.array([0.15, 0.25, 0.7, 1.0, 0.8, 0.35, 0.15])
+    ratio_path_2 = np.array([0.1, 0.2, 0.7, 1.0, 0.8, 0.3, 0.1])
     x_path_1 = np.array(
         [
             [2.0, 2.1, 2.3, 2.5, 2.7, 2.9, 3.0],