Geoffrey_Coram

There is a "bsource" element that allows arbitrary expressions, which could probably handle this.
I'd be concerned about the expression you have, though. Eg, on the first iteration of the dc solution, many simulators start with a vvector that is all zeros. This would probably mean that y=0 on the very first iteration, and the resulting division by zero would prevent the simulator from knowing how to change that voltage. Newton's method could also produce a negative current (x) during one iteration.
