1: \begin{abstract}

2: We address the problem of designing simultaneous input and state interval observers %is presented for %mixed monotone 

3: for Lipschitz \yong{continuous} nonlinear systems with \yongz{rank-deficient feedthrough,} unknown inputs and bounded noise signals. Benefiting from the existence of nonlinear decomposition functions and affine abstractions, \yongs{

4: our proposed observer recursively computes the maximal and minimal elements of the estimate intervals that are proven to contain the true states and unknown inputs. %and leverages the output/measurement signals to shrink the intervals by eliminating estimates that are incompatible with the measurements. 

5: %are compatible with output/measurement signals, \moham{in the sense that given the observation, at each step, our observer iteratively updates the interval estimates, by iteratively solving a tractable linear program (LP) that returns \emph{local} parallel affine abstraction (bounding) functions for the observation vector field.}

6: Moreover, we provide \mohamm{necessary and} sufficient conditions for the %\emph{nontrivial} 

7: existence and \mohamm{sufficient conditions for the} stability (i.e., uniform boundedness of the sequence of estimate interval widths) of the designed observer, %. Furthermore, we  provide several sufficient conditions for stability of the designed observer (i.e., uniform boundedness of the sequence of estimate interval widths), using a derived Lipschitz-like property for decomposition (nonlinear bounding) functions. We also 

8: and} show that the input interval estimates are tight, given the state intervals and decomposition functions. % and convergence of the predicted upper-bounding sequence of interval lengths. 

9: % Finally, the effectiveness of our approach is demonstrated using an illustrative example.

10:  \end{abstract}

11: