Index: /issm/trunk-jpl/src/m/classes/inversion.py
===================================================================
--- /issm/trunk-jpl/src/m/classes/inversion.py	(revision 13092)
+++ /issm/trunk-jpl/src/m/classes/inversion.py	(revision 13093)
@@ -37,4 +37,8 @@
 		self.vel_obs                     = float('NaN')
 		self.thickness_obs               = float('NaN')
+
+		#set defaults
+		self.setdefaultparameters()
+
 		#}}}
 	def __repr__(self):
@@ -78,5 +82,5 @@
 		#parameter to be inferred by control methods (only
 		#drag and B are supported yet)
-		self.control_parameters=['FrictionCoefficient']
+		self.control_parameters='FrictionCoefficient'
 
 		#number of steps in the control methods
@@ -85,5 +89,5 @@
 		#maximum number of iteration in the optimization algorithm for
 		#each step
-		self.maxiter_per_step=20*ones(self.nsteps)
+		self.maxiter_per_step=20*numpy.ones(self.nsteps)
 
 		#the inversed parameter is updated as follows:
@@ -92,13 +96,13 @@
 		#inversed parameter (10^8 for B, 50 for drag) and can be decreased
 		#after the first iterations
-		self.gradient_scaling=50*ones(self.nsteps)
+		self.gradient_scaling=50*numpy.ones(self.nsteps)
 
 		#several responses can be used:
-		self.cost_functions=101*ones(self.nsteps)
+		self.cost_functions=101*numpy.ones(self.nsteps)
 
 		#step_threshold is used to speed up control method. When
 		#misfit(1)/misfit(0) < self.step_threshold, we go directly to
 		#the next step
-		self.step_threshold=.7*ones(self.nsteps) #30 per cent decrement
+		self.step_threshold=.7*numpy.ones(self.nsteps) #30 per cent decrement
 
 		#stop control solution at the gradient computation and return it? 
@@ -108,5 +112,5 @@
 		#if J[n]-J[n-1]/J[n] < criteria, the control run stops
 		#NaN if not applied
-		self.cost_function_threshold=NaN #not activated 
+		self.cost_function_threshold=float('NaN')    #not activated 
 
 		return self