Thanks Kristoffer, indeed that works. 

I'm slightly baffled by the initialisation process. In my session I had 
already used fill! successfully on a matrix already on the device, so 
presumed everything had been initialised. Clearly that's not enough.


Reply via email to