[R] Catch errors

[Gang Chen]

I run a linear mixed-effects model in a loop

for (i in 1:N) {
    fit.lme <- lme(Y ~ X1*X2, random = ~1|subj, Model[i]);
}

As the data in some iterations are all (or most) 0's, leading to the  
following error message from lme:

Error in chol((value + t(value))/2) : the leading minor of order 1 is  
not positive definite

What is a good way to catch the error without spilling on the screen  
so that I can properly stuff the corresponding output with some  
artificial value such as 0?

Thanks,
Gang