- We can estimate this error assuming
**f**and**g**are independent. - NOTE: We add, not multiply errors.
- Suppose we use events.
- Therefore, use of
**2N**events gets the same fractional error in**I**as in , , if these each use**N**events and have . - So do NOT think of (NI.18) as a Monte Carlo over theory and for
each theory (model) a Monte Carlo over experiment.
- Rather, (NI.18) is a single Monte Carlo where ``random
**x**'' plus ``random**y**'' form a single random point in multidimensional space.

Geoffrey Fox, Northeast Parallel Architectures Center at Syracuse University, gcf@npac.syr.edu