reported by Gary Hennigan
on tracker
http://sourceforge.net/tracker/index.php?func=detail&aid=3591132&group_id=38962&atid=423915
access to uninitialized memory,
trying to acquire the "time" of a projected auxiliary timestep in the past.
now using the following auxiliary timestep t[-1]
voltage(t[-1]) === voltage(t[0])
and
t[-1] === t[0] - (t[1] - t[0])
the details are not understood,
but thats certainly better than accessing uninitialized memory.