Some structures are generated with nan's in cell vectors

[pmrv]

For some inputs randspg seems to generate structures where the cell vectors contain nan's. I've attached an example below.

I'm using the version 0.0.1 from conda, which I assume points to this repo's HEAD, since it's newer than the last commit here.

Example config file:

composition=Mg1
spacegroups=1-230
latticeMins=1.8, 1.8, 1.8, 30, 30, 30
latticeMaxes=3.6, 3.6, 3.6, 150, 150, 150
minVolume=5.832000000000001
maxVolume=46.656000000000006
numOfEachSpgToGenerate=1
setMinRadii=1.8
scalingFactor=1
maxAttempts=100
outputDir=randSpgOut
verbosity=r
forceMostGeneralWyckPos=FALSE

Example structure with nans

Mg1 -- randSpg with spg of: 2
1.00000
    3.124722616095241    0.000000000000000                 -nan
   -1.497237672060954    1.026086982099370                 -nan
    2.311760653395992    4.608278902740905                 -nan
   Mg
    1
Direct
  0.000000000000000  0.500000000000000  0.000000000000000

The log file and the stdout/stderr did not report any problems.

It's easy enough for me to filter these structures, but I thought maybe something's off with my input parameters or so. Thanks already for looking into this.

[pmrv]

I found that restricting the lattice angles to (much more reasonably) [60,120] makes this disappear, so I'm assuming the problem is somewhere in the generation of triclinic and monoclinic cells (applying a rotation somewhere maybe?). That's ok for my application, so this is a very non-critical issue.