Block-Preserving Contraction

• Block-preserving contraction is the preservation of key block structures during contraction analysis, ensuring that essential operator or system components remain intact under mapping transformations.
• It uses spectral set theory and operator factorization conditions, such as the generalized Douglas–Muhly–Pearcy criterion, to characterize when block matrices remain contractions.
• The approach extends to nonlinear feedback systems via Kronecker formulations and block-preserving small-gain theorems, enabling scalable convergence analysis in networked dynamical systems.

Block-preserving contraction refers to structural conditions and matrix inequalities governing contraction properties for block-structured matrices or dynamical systems, where essential features or blocks in the formulation remain preserved under the contraction operation. This principle manifests in two major areas: spectral set theory for operator block matrices associated with the annulus, and higher-order contraction theory for block-partitioned nonlinear dynamical systems—especially, $2$-contraction for feedback interconnections. Both domains rigorously characterize when certain block operators or interconnected systems maintain contraction, using criteria that exploit the block decomposition and preserve key block-wise relations under mapping or evolution.

1. Spectral Set Theory and Block-Preserving $\mathbb{A}_r$-Contractions

A $\mathbb{A}_r$-contraction is a bounded Hilbert space operator $T$ for which the closure of the annulus $$\mathbb{A}_r := \{z \in \mathbb{C}: r \le |z| \le 1\}$$, for $0$\sigma(T) \subseteq \mathbb{A}_r$ (spectrum inclusion) and for every rational function $f$ with poles off $\mathbb{A}_r$,

$\|f(T)\| \leq \sup\{|f(z)|: z \in \mathbb{A}_r\}.$

Agler's theorem provides an operator-theoretic characterization: $T$ is an $\mathbb{A}_r$-contraction if for all $\epsilon \in (0,1)$ and $a \in \mathbb{T}$ (the unit circle),

$$\operatorname{Re} \Gamma_\epsilon(aT) := \frac{1}{2} \bigl(\Gamma_\epsilon(aT) + \Gamma_\epsilon(aT)^*\bigr) \ge 0,$$

where $\Gamma_\epsilon(z)$ is a specific power-series holomorphic on a slightly larger annulus.

For block operators, the classical Douglas–Muhly–Pearcy (DMP) condition gives a necessary and sufficient criterion for when a $2\times2$ block matrix $$T_X = \begin{bmatrix} T_1 & X \ 0 & T_2 \end{bmatrix}$$ is a contraction: Both $T_1$ and $T_2$ must be contractions, and there must exist a contraction $C$ such that $X = D_{T_1^*} C D_{T_2}$, with $D_T = (I - T^*T)^{1/2}$ the defect operator (Pal et al., 2023).

The block-preserving criterion in the annulus setting aims to generalize the DMP result: Given $$\widetilde{T}_Y = \begin{bmatrix} T_1 & Y \ 0 & T_2 \end{bmatrix}$$, under the "block-preserving" hypothesis that $Y = X (T_1-T_2)$ for $X$ commuting with $T_1,T_2$ (i.e., $T_1-T_2$ invertible and $X = Y(T_1-T_2)^{-1}$), necessary and sufficient conditions can again be formulated in terms of operator-valued factorization and functional calculus identities for blocks (Pal et al., 2023).

2. Block Factorization Principles in the Annulus Setting

The block-preserving $\mathbb{A}_r$-contraction result states:

Let $T_1$, $T_2$ be commuting $\mathbb{A}_r$-contractions and $X$ commute with both. Define $$T_X = \begin{bmatrix} T_1 & X (T_1-T_2) \ 0 & T_2 \end{bmatrix}$$. Then $T_X$ is an $\mathbb{A}_r$-contraction if and only if either (a) or (b) below holds for every $\epsilon \in (0,1)$ and $a \in \mathbb{T}$:

(a) Existence of a contraction $K = K(\epsilon,a)$ in $B(H)$ such that

$$X \bigl[\Gamma_\epsilon(aT_1) - \Gamma_\epsilon(aT_2)\bigr] = \left[\operatorname{Re}\Gamma_\epsilon(aT_1)\right]^{1/2} K \left[\operatorname{Re}\Gamma_\epsilon(aT_2)\right]^{1/2}.$$

(b) Existence of a unitary $U=U(\epsilon,a)$ on $H \oplus H$ satisfying a block dilation equation for the corresponding numerical range.

This generalizes to the case where $T_1-T_2$ is invertible for arbitrary $Y \in B(H)$: $X$, defined as $Y (T_1 - T_2)^{-1}$, must satisfy the above factorization. These factorization conditions directly encode block structure preservation via commutation and multiplicative relationships.

3. Block-Preserving $2$-Contraction in Nonlinear Feedback Interconnections

$k$-contraction theory provides a generalization of classical contraction analysis in dynamical systems. For a system $\dot{x} = f(t,x)$ with Jacobian $J(t,x)$, the $k$th additive compound $J^{[k]}$ relates to the contraction of $k$-dimensional structures. For $k=2$, all bounded trajectories of a time-invariant $2$-contracting system converge to equilibrium points, potentially supporting multistationarity.

In the context of block-preserving contraction, new block-Kronecker product and sum formulas for the $2$-compounds are derived for block-partitioned matrices:

$$X = \begin{pmatrix}A & B \ C & D\end{pmatrix},$$

with blocks $A, B, C, D$ of appropriate dimensions. The $2$-multiplicative compound $X^{(2)}$ and $2$-additive compound $X^{[2]}$ can be constructed efficiently using appropriately defined block Kronecker operations and permutation/selection matrices, enabling tractable contraction analysis for interconnected systems (Ofir et al., 2024).

4. Sufficient Conditions via Block-Preserving Small-Gain Theorems

For a feedback interconnection

$\dot{x} = f(x, z), \qquad \dot{z} = g(x, z),$

sufficient conditions for $2$-contraction are formulated using measures of the diagonal Jacobian blocks ($c_x^{[2]}(f), c_z^{[2]}(g)$), the induced matrix measures of off-diagonal blocks, and bounds on their respective norms. A hierarchical method constructs a Metzler "comparison" matrix from the induced norms and matrix measures of the individual blocks; then, a network small-gain theorem for Metzler matrices applies. The critical inequalities involve both diagonal contraction and constraints on off-diagonal block strengths, capturing the preservation of the block structure and preventing destabilization from feedback (Ofir et al., 2024).

A summary table of the main sufficient condition:

Condition Type	Notation / Formula
Diagonal $2$-contraction	$c_x^{[2]}(f) < 0,\quad c_z^{[2]}(g) < 0$
Joint $1$-contraction	$c_x(f) + c_z(g) < 0$
Off-diagonal small-gain	$$\| \partial_z f \| \cdot \| \partial_x g \| < \frac12 \frac{[c_x(f)+c_z(g)]c_x^{[2]}(f)c_z^{[2]}(g)}{c_x^{[2]}(f) + c_z^{[2]}(g)}$$

5. Specializations and Concrete Examples

Special cases illustrate the general block-preserving theory:

• Identical diagonal blocks (e.g., $$T_X = \begin{bmatrix}T & X \ 0 & T\end{bmatrix}$$ for operator contractions, or symmetric feedback patterns in nonlinear systems) reduce the factorization conditions to simpler operator identities or contraction inequalities.
• Scalar examples in the annulus setting: For $H = \mathbb{C}$ and scalar $T_1 = \lambda, T_2 = \mu$ in $\mathbb{A}_r$, setting $Y = t (\lambda - \mu)$, the block-preserving criterion shows that $$\widetilde{T}_Y = \begin{pmatrix}\lambda & t(\lambda-\mu) \ 0 & \mu\end{pmatrix}$$ is an $\mathbb{A}_r$-contraction if and only if $|t| \le 1$ (Pal et al., 2023).
• FitzHugh–Nagumo network: For $N$ coupled FHN units with Laplacian coupling matrix $R$, and in the context of $2$-contraction with block-preserved structure, explicit inequalities in system and network parameters ($b, c, \lambda_2$) can guarantee convergence of all bounded trajectories to the equilibria set, even in the presence of multistationarity (Ofir et al., 2024).

6. Methodological Significance and Implications

Block-preserving contraction results leverage the intrinsic block structure to yield sharp necessary and sufficient conditions for contraction, both in operator-theoretic and dynamical systems contexts. The operator-theoretic route uses matrix positivity, functional calculus, and dilation tricks (e.g., Halmos' unitary dilation) to encode block factorization. In the nonlinear setting, Kronecker formulations and hierarchical contraction exploit the block structure for scalability and transparency of the multistationary regime.

A plausible implication is that these block-preserving conditions enable modular certification of contraction-type properties in large-scale or networked systems, as well as in the spectral theory of operator tuples where block coupling is algebraically constrained.

• "The $2 \times 2$ block matrices associated with an annulus" (Pal et al., 2023)
• "A sufficient condition for 2-contraction of a feedback interconnection" (Ofir et al., 2024)