Think also about something similar to f(x) = x^N * sin(1/x) or g(x) = abs(f(x)), x from [eps...1], 0 < eps << 1, N>>1 (so f(x) is very small near eps) D. John Hassler wrote: > How do you know if there is a zero in your interval? How about: > y = x**2 + 1.e-999? > john