[Eeglablist] When your nan isn't your amica

[John Johnson]

I have a particular subject for whom AMICA is terminating.
I checked the mailing list archives, but only saw one report of this, 
and no answer.
I'm running AMICA with the pcakeep option set to getrank()'s rank of the 
data.

What's the best course of action?

```
iter   585 lrate =  0.5000000000 LL =  -2.0261266821 nd =  0.0004927834, 
D =   0.51974E+01  0.22712E+01  (  1.79 s,   0.7 h)
  Hessian not positive definite, using natural gradient
  iter   586 lrate =  0.0250000000 LL =            NaN nd =           
NaN, D =   0.51974E+01  0.22712E+01  (  1.76 s,   0.7 h)
  Got NaN! Exiting ...
... done. Execution time:   0.30 h
```

Thanks,
John

John T. Johnson
PhD Candidate - Cognitive Motor Control Laboratory
Lab TA NEURO 3010 Neuroscience Methods
School of Biological Sciences
Georgia Institute of Technology
john.johnson at gatech.edu
he/him/his