A priori worst-case error bounds for floating-point computations | IEEE Conference Publication | IEEE Xplore