Various small changes (#241)

* Remove lower bound from triangulated surface constraint

* Make addRefAxis work with zero x/y/zFraction

* Fix type of CompositeDV modeshapes

* Fix other fraction checks

* Convert `J_full` to array, not matrix

---------

Co-authored-by: ArshSaja <63115167+ArshSaja@users.noreply.github.com>

Author: A-CGray

pygeo/constraints/areaConstraint.py

@@ -38,7 +38,7 @@ def __init__(
         if geograd_parallel is None:
             raise ImportError("Geograd package must be installed to use triangulated surface constraint")
 
-        super().__init__(name, 2, -1e10, 0.0, scale, None, addToPyOpt)
+        super().__init__(name, 2, None, 0.0, scale, None, addToPyOpt)
 
         self.comm = comm
 

pygeo/parameterization/DVGeo.py

@@ -506,7 +506,7 @@ def addRefAxis(
                     "rot0axis": rot0axis,
                 }
             nAxis = len(curve.coef)
-        elif xFraction or yFraction or zFraction:
+        elif xFraction is not None or yFraction is not None or zFraction is not None:
             # Some assumptions
             #   - FFD should be a close approximation of geometry surface so that
             #       xFraction roughly corresponds to airfoil LE, TE, or 1/4 chord
@@ -594,22 +594,22 @@ def addRefAxis(
                             pts_vec[ct_, :] = p_rot
 
                     # Temporary ref axis node coordinates - aligned with main system of reference
-                    if xFraction:
+                    if xFraction is not None:
                         # getting the bounds of the FFD section
                         x_min = np.min(pts_vec[:, 0])
                         x_max = np.max(pts_vec[:, 0])
                         x_node = xFraction * (x_max - x_min) + x_min  # chordwise
                     else:
                         x_node = np.mean(pts_vec[:, 0])
 
-                    if yFraction:
+                    if yFraction is not None:
                         y_min = np.min(pts_vec[:, 1])
                         y_max = np.max(pts_vec[:, 1])
                         y_node = y_max - yFraction * (y_max - y_min)  # top-bottom
                     else:
                         y_node = np.mean(pts_vec[:, 1])
 
-                    if zFraction:
+                    if zFraction is not None:
                         z_min = np.min(pts_vec[:, 2])
                         z_max = np.max(pts_vec[:, 2])
                         z_node = z_max - zFraction * (z_max - z_min)  # top-bottom
@@ -1518,7 +1518,7 @@ def addCompositeDV(self, dvName, ptSetName=None, u=None, scale=None, prependName
             if ptSetName is None:
                 raise ValueError("If u and s need to be computed, you must specify the ptSetName")
             self.computeTotalJacobian(ptSetName)
-            J_full = self.JT[ptSetName].todense()  # this is in CSR format but we convert it to a dense matrix
+            J_full = self.JT[ptSetName].toarray()  # this is in CSR format but we convert it to a dense matrix
             u, s, _ = np.linalg.svd(J_full, full_matrices=False)
             scale = np.sqrt(s)
             # normalize the scaling