<P> Because of negative zero (and also when the rounding mode is upward or downward), the expressions − (x − y) and (− x) − (− y), for floating - point variables x and y, cannot be replaced by y − x . However (− 0) + x can be replaced by x with rounding to nearest (except when x can be a signaling NaN). </P> <P> Some other special rules: </P> <Ul> <Li> − 0 = − 0 (\ displaystyle (\ sqrt (- 0)) = - 0 \, \!) </Li> <Li> − 0 − ∞ = + 0 (\ displaystyle (\ frac (- 0) (- \ infty)) = + 0 \, \!) (follows the sign rule for division) </Li> <Li> x − 0 = − ∞ (\ displaystyle (\ frac (\ left x \ right) (- 0)) = - \ infty \, \!) (for non-zero x (\ displaystyle x), follows the sign rule for division) </Li> <Li> ± 0 × ± ∞ = NaN (\ displaystyle (\ pm 0) \ times (\ pm \ infty) = (\ mbox (NaN)) \, \!) (Not a Number or interrupt for indeterminate form) </Li> <Li> ± 0 ± 0 = NaN (\ displaystyle (\ frac (\ pm 0) (\ pm 0)) = (\ mbox (NaN)) \, \!) </Li> </Ul> <Li> − 0 = − 0 (\ displaystyle (\ sqrt (- 0)) = - 0 \, \!) </Li>

There are two zero's in the binary signed magnitude representation of integer numbers