1: \begin{definition}

2: \label{def:evalrelaxslide}

3: Consider a processing chain as in Definition~\ref{def:processing-chain} and let $\ell\inint{1}{L - 1}$ be a fixed layer index.

4: With Theorem~\ref{thm:sliding-subsignal} let $f_j\colon M_{j - 1}^{c_j}\to N_j$ be the unique functions which fulfill $T_j = \Slide_{f_j}$ for all $j\inint{1}{L}$.

5: For $j\inint{0}{L}$, the operator $\EvalRelaxSlide_j^\ell\colon\cup_{\ROI + k_L^* - k_\ell^*}(M_0)\to\cup_1(M_j)$, 

6: \begin{displaymath}

7:   \xi\mapsto

8:   \begin{cases}

9:     \xi\text{,} & \text{if } j = 0\text{,}\\

10:     \Stride_{g_j}(\Slide_{f_j}(\EvalRelaxSlide_{j - 1}^\ell(\xi)))\text{, } & \text{if } j\inint{1}{\ell}\text{,}\\

11:     \Fragmentation_{k_j}(\Slide_{g_j}(\Slide_{f_j}(\EvalRelaxSlide_{j - 1}^\ell(\xi))))\text{, } & \text{if } j\inint{\ell + 1}{L}\text{,}

12:   \end{cases}

13: \end{displaymath}

14: applies the processing chain in a fashion which is \emph{mixed of a relaxed and a sliding fashion}.

15: Divisibility requirements for well-definedness of this operator are detailed below.

16: \end{definition}

17: