What I posted before is probably overkill. If you write these particular problems to do the subtraction at the integer level (2105-1105) and then divide at the end by the appropriate power of 10, that might be a cleaner way to avoid machine rounding errors.

You are currently using guest access (Log in)