diff --git a/smt/applications/mfk.py b/smt/applications/mfk.py
index 0a3847e2c..d113e3ed0 100644
--- a/smt/applications/mfk.py
+++ b/smt/applications/mfk.py
@@ -123,7 +123,10 @@ def _initialize(self):
             desc="Turning this option to True, forces variance to zero at HF samples ",
         )
         declare(
-            "noise0", 1e-6, types=(float, list), desc="Initial noise hyperparameters"
+            "noise0",
+            [1e-6],
+            types=(list, np.ndarray),
+            desc="Initial noise hyperparameters",
         )
         self.name = "MFK"
 
@@ -208,8 +211,8 @@ def _new_train(self):
         self.X_norma_all = [(x - self.X_offset) / self.X_scale for x in X]
         self.y_norma_all = [(f - self.y_mean) / self.y_std for f in y]
 
-        if isinstance(self.options["noise0"], float):
-            self.options["noise0"] = self.nlvl * [self.options["noise0"]]
+        if len(self.options["noise0"]) == 1:
+            self.options["noise0"] = self.nlvl * self.options["noise0"]
         noise0 = self.options["noise0"].copy()
 
         if (
@@ -227,7 +230,7 @@ def _new_train(self):
         theta0 = self.options["theta0"].copy()
 
         for lvl in range(nlevel):
-            self.options["noise0"] = noise0[lvl]
+            self.options["noise0"] = [noise0[lvl]]
             self.options["theta0"] = theta0[lvl, :]
 
             self.X_norma = self.X_norma_all[lvl]
diff --git a/smt/applications/mfkpls.py b/smt/applications/mfkpls.py
index 4870b30b8..66417cfa7 100644
--- a/smt/applications/mfkpls.py
+++ b/smt/applications/mfkpls.py
@@ -57,7 +57,12 @@ def _initialize(self):
             values=(True, False),
             desc="noise evaluation flag",
         )
-        declare("noise0", 1e-6, types=float, desc="Initial noise hyperparameter")
+        declare(
+            "noise0",
+            [1e-6],
+            types=(list, np.ndarray),
+            desc="Initial noise hyperparameter",
+        )
         self.name = "MFKPLS"
 
     def _componentwise_distance(self, dx, opt=0):
diff --git a/smt/surrogate_models/krg_based.py b/smt/surrogate_models/krg_based.py
index ad46cfbd6..737e11ddf 100644
--- a/smt/surrogate_models/krg_based.py
+++ b/smt/surrogate_models/krg_based.py
@@ -180,8 +180,8 @@ def _new_train(self):
             self._specific_train()
         else:
             if self.options["eval_noise"]:
-                self.noise = self.optimal_theta[self.nx :]
-                self.optimal_theta = self.optimal_theta[0 : self.nx]
+                self.noise = self.optimal_theta[self.D.shape[1] :]
+                self.optimal_theta = self.optimal_theta[0 : self.D.shape[1]]
         # if self.name != "MGP":
         #     del self.y_norma, self.D
 
@@ -244,8 +244,8 @@ def _reduced_likelihood_function(self, theta):
         noise = 0
         tmp_var = theta
         if self.options["eval_noise"]:
-            theta = tmp_var[0 : self.nx]
-            noise = tmp_var[self.nx :]
+            theta = tmp_var[0 : self.D.shape[1]]
+            noise = tmp_var[self.D.shape[1] :]
         r = self._correlation_types[self.options["corr"]](theta, self.D).reshape(-1, 1)
 
         if self.options["is_noise_het"]:
@@ -1066,12 +1066,14 @@ def _check_param(self):
 
         # FIXME: _check_param should be overriden in corresponding subclasses
         if self.name in ["KPLS", "KPLSK", "GEKPLS", "MFKPLS", "MFKPLSK"]:
-
             d = self.options["n_comp"]
         elif self.name in ["MGP"]:
             d = self.options["n_comp"] * self.nx
         else:
             d = self.nx
+        # self.n_features_x = d
+        # if self.name in ["KPLS", "MFKPLSK"]:
+        #     self.n_features_x = self.nx
 
         if self.name in ["MGP"]:
             if self.options["corr"] != "act_exp":