1: \begin{abstract}

2: This paper presents a novel set-based computing method, called interval 

3: superposition arithmetic, for enclosing the image set of multivariate factorable

4: functions on a given domain. In order to construct such enclosures, the proposed

5: arithmetic operates over interval superposition models which are parameterized 

6: by a matrix with interval components. Every point in the domain of a factorable

7: function is then associated with a sequence of components of this matrix and the superposition,

8: i.e. Minkowski sum, of these elements encloses the image of the function at this

9: point. Interval superposition arithmetic has a linear runtime complexity with

10: respect to the number of variables. Besides presenting a detailed theoretical analysis 

11: of the accuracy and convergence properties of interval superposition arithmetic, 

12: the paper illustrates its advantages compared to existing set arithmetics

13: via numerical examples. 

14: \end{abstract}

15: