Miranda reader does not work with serial data in newer format

[akshaysubr]

Miranda reader with the newer format and a single processor run returns a zero array for all fields including coordinates. For a 128x128x1 2D problem, looks like self.px, seld.py, self.pz == 0, 0, 1 with self.nx, seld.ny, self.nz == 128, 128, 1 and self.ax, seld.ay, self.az == 129, 129, 1. Maybe an issue with handling of ghost point indexing in the reader?

[akshaysubr]

Changing

FloATPy/floatpy/readers/miranda_reader.py

Lines 99 to 101 in 9628b87

	self.ax -= numpy.minimum(self.px-1,1)
	self.ay -= numpy.minimum(self.py-1,1)
	self.az -= numpy.minimum(self.pz-1,1)

to

# Fix the local indices                      
if self.nx > 1:                              
    self.ax -=  1 #numpy.minimum(self.px-1,1)
if self.ny > 1:                              
    self.ay -=  1 #numpy.minimum(self.py-1,1)
if self.nz > 1:                              
    self.az -=  1 #numpy.minimum(self.pz-1,1)

seems to fix the issue. @flow-phys could you check if this fix is correct?

[flow-phys]

@akshaysubr , this looks about right, though I'd need to check that this doesn't break the parallel version.

[akshaysubr]

Fixed this issue in the latest pull request. Checked that the fix works with the parallel reader as well. Closing this issue.