--- title: Incremental Global Asymptotic Stability (δ-GAS) url: https://www.emergentmind.com/topics/incremental-global-asymptotic-stability-delta-gas type: topic --- # Incremental Global Asymptotic Stability (δ-GAS) Searching arXiv for recent and foundational papers on incremental global asymptotic stability (\(\delta\)-GAS). Searching arXiv for "incremental global asymptotic stability delta-GAS". Incremental Global Asymptotic Stability, usually denoted \(\delta\)-GAS, is a trajectory-to-trajectory stability property: instead of asking whether solutions converge to a fixed equilibrium, it asks whether any two solutions driven by the same admissible input converge to one another according to a \(\mathcal{KL}\) estimate. In the Euclidean formulation this takes the form of a decay bound on \(\|x(t)-\tilde x(t)\|\); in coordinate-invariant formulations it is stated with respect to an arbitrary metric \(d\). Closely related terminology includes “incremental asymptotic stability,” “global incremental stability” (GIS), “convergence of all solutions,” and, in stronger cases, incremental exponential stability. Across the literature represented here, \(\delta\)-GAS appears as a state-space property for nonlinear control systems, as a contraction property for nonautonomous disturbed systems, and as a compositional certificate for large-scale networks [1207.0030] [1611.08322] [2206.12945] [2507.18141]. ## 1. Formal notion and principal variants For control systems \(\Sigma=(\mathbb{R}^n,U,\mathcal{U},f)\), the coordinate-invariant formulation introduced as \(\delta_{\exists}\)-GAS requires forward completeness and the existence of a metric \(d\) and a function \(\beta\in\mathcal{KL}\) such that, for any \(t\ge 0\), any \(x,x'\in\mathbb{R}^n\), and any common input \(v\in\mathcal U\), \[ d(\xi_{xv}(t),\xi_{x'v}(t))\le \beta(d(x,x'),t). \] The same-input requirement is structural: the property compares trajectories under identical exogenous signals, whereas input-mismatch belongs to incremental ISS-type notions. The paper introducing \(\delta_{\exists}\)-GAS is explicit that Angeli’s original Euclidean \(\delta\)-GAS is generally not coordinate invariant, and that allowing an arbitrary metric restores coordinate invariance [1207.0030]. A Euclidean state-space formulation is used for general nonlinear and piecewise-affine systems in the form \[ \|x(t)-\tilde x(t)\|\le \beta(\|x_0-\tilde x_0\|,t), \] for all \(x_0,\tilde x_0\in\mathcal X\), all \(t\ge 0\), and any common input \(u\in\mathcal U\). When \(\mathcal X=X=\mathbb R^n\), the property is called incrementally globally asymptotically stable. The same source distinguishes regional and global versions, and identifies incremental exponential stability as the stronger case in which \(\beta(r,t)\le d e^{-\lambda t}r\) for some \(d,\lambda>0\) [1611.08322]. For nonlinear time-varying perturbed systems \[ \dot x=f(x,t)+\delta(t), \] the logarithmic-norm treatment uses the term GIS and defines incremental asymptotic stability on a positively invariant set \(X\subset\mathbb R^n\) by the existence of \(\beta\in\mathcal{KL}\) such that \[ |x(t)-x^*(t)|\le \beta(|x(t_0)-x^*(t_0)|,t-t_0). \] When \(X=\mathbb R^n\), this is GIS; in standard control language, this corresponds to \(\delta\)-GAS in the norm-induced metric under consideration. That source also distinguishes the stronger estimate \[ |x(t)-x^*(t)|\le e^{-\alpha_0(t-t_0)}|x(t_0)-x^*(t_0)|, \] which is standard incremental global uniform exponential stability [2206.12945]. ## 2. Lyapunov, diagonal-set, and dissipativity characterizations A central theme in the modern treatment of \(\delta\)-GAS is that incremental stability can be reduced to ordinary stability of a lifted system. The coordinate-invariant framework constructs the product system \[ \widehat{\Sigma}=(\mathbb R^{2n},U,\mathcal U,\widehat f),\qquad \widehat f(\zeta,v)=\begin{bmatrix} f(\xi_1,v)\\ f(\xi_2,v)\end{bmatrix}, \] and studies the diagonal set \[ \Delta=\left\{ z\in\mathbb R^{2n}\mid \exists x\in\mathbb R^n:\ z=\begin{bmatrix}x\\x\end{bmatrix}\right\}. \] With the product metric \(\widehat d(z,z')=d(x_1,x_1')+d(x_2,x_2')\), the identity \[ \widehat d(z,\Delta)=d(x_1,x_2) \] turns inter-trajectory distance into distance-to-set. Under compactness of \(U\) and continuity of the metric fibers \(x\mapsto d(x,y)\), \(\delta_{\exists}\)-GAS is equivalent to the existence of a \(\delta_{\exists}\)-GAS Lyapunov function \(V(x,x')\) satisfying \[ \underline{\alpha}(d(x,x')) \le V(x,x') \le \overline{\alpha}(d(x,x')) \] and \[ \frac{\partial V}{\partial x}f(x,u)+\frac{\partial V}{\partial x'}f(x',u)\le -\kappa V(x,x'), \] for all \(x,x'\in\mathbb R^n\) and \(u\in U\). This converse Lyapunov theorem is one of the strongest structural results in the cited literature [1207.0030]. A related but distinct route uses dissipativity on an augmented system. For general nonlinear systems \(\dot x=f(x,u)\), \(y=h(x,u)\), incremental \(L_2\)-gain is treated as dissipativity of the augmented dynamics with state \((x,\tilde x)\), output difference \(h(x,u)-h(\tilde x,\tilde u)\), and supply rate \[ \bar w(u,\tilde u,\bar y)=\eta^2\|u-\tilde u\|^2-\|\bar y\|^2. \] When the compared inputs are equal, the dissipation inequality yields monotonicity of the incremental storage. To recover \(\delta\)-GAS from incremental \(L_2\)-gain, however, the paper requires additional uniform observability and uniform reachability assumptions. Under those assumptions, the storage admits lower and upper bounds in the state difference and strict finite-horizon decay, which is converted into a \(\mathcal{KL}\) estimate. Conversely, incremental asymptotic stability implies incremental \(L_2\)-gain only under differentiability and local Lipschitz regularity of the Jacobian. This establishes that incremental \(L_2\)-gain and \(\delta\)-GAS are closely related but not interchangeable without auxiliary hypotheses [1611.08322]. ## 3. Contraction, logarithmic norms, and Demidovich-type criteria For nonautonomous disturbed systems, \(\delta\)-GAS can be certified through matrix measures. In the perturbed system \[ \dot x=f(x,t)+\delta(t), \] with \(f\) continuously differentiable in \(x\), continuous in \(t\), \(\delta\) continuous, and solutions uniquely determined for all \(t\ge t_0\), the key assumption is the global Jacobian bound \[ \mu[J_x f(x,t)]\le -\alpha(t)\le -\alpha_0<0 \quad \text{for all } (x,t)\in\mathbb R^n\times [t_0,\infty), \] where \(\mu\) is the logarithmic norm associated with a chosen vector norm. The logarithmic norm is defined by \[ \mu[A]\triangleq \lim_{\theta\to 0^+}\frac{\|I+\theta A\|-1}{\theta}. \] For the Euclidean norm, \[ \mu_I[A]=\frac12\lambda_{\max}(A+A^T), \] and for a weighted Euclidean norm \(|x|_P=(x^TPx)^{1/2}\), \[ \mu_P[A]=\frac12\lambda_{\max}(\hat A+\hat A^T),\qquad \hat A=P_0AP_0^{-1}. \] Under the uniform negativity condition above, any two solutions satisfy \[ |x(t)-x^*(t)|\le e^{-\alpha_0(t-t_0)}|x(t_0)-x^*(t_0)|, \] which is stronger than abstract \(\mathcal{KL}\)-incremental stability and yields global uniform exponential incremental stability [2206.12945]. The proof proceeds by exact linearization of the trajectory difference through the mean-value identity \[ f(x,t)-f(x^*,t)=\left[\int_0^1 J_xf(x^*+\xi(x-x^*),t)\,d\xi\right](x-x^*), \] which produces a linear time-varying system \(\dot z=A(t)z\) for \(z=x-x^*\). Convexity of the matrix measure implies \(\mu[A(t)]\le -\alpha(t)\), and standard transition-matrix estimates then yield exponential contraction. A weaker assumption, \[ \mu[J_x f(x,t)]\le -\alpha(t),\qquad \int_{t_0}^{\infty}\alpha(\tau)\,d\tau=\infty, \] still gives \[ |x(t)-x^*(t)|\le e^{-\int_{t_0}^t\alpha(\tau)\,d\tau}|x(t_0)-x^*(t_0)|, \] which ensures asymptotic pairwise convergence but is generally not uniform and not exponential [2206.12945]. That same treatment generalizes the classical Demidovich criterion both “horizontally and vertically.” The horizontal generalization extends the contraction test from weighted Euclidean norms to arbitrary norms and corresponding matrix measures. The vertical generalization adds a result that identifies a common limit, namely the origin, under the additional asymptotic smallness condition \[ \frac{|f(0,t)+\delta(t)|}{\alpha(t)}\to 0. \] Under this condition every solution satisfies \(x(t)\to 0\), even when \(x=0\) is not an equilibrium of the perturbed system and may fail to be a solution of the nominal system. The paper states that the logarithmic-norm approach makes stability “a topological notion”; technically, this refers to formulation with respect to arbitrary norm-induced topologies rather than a restriction to quadratic Euclidean metrics, not to coordinate invariance in the differential-geometric sense [2206.12945]. ## 4. Computational verification and compositional certification For piecewise-affine systems, \(\delta\)-GAS can be certified by finite-dimensional convex conditions. The continuous-time PWA model \[ \dot x(t)=A_i x(t)+a_i+B_i u(t),\qquad y(t)=C_i x(t)+c_i+D u(t)\quad \text{for } x(t)\in X_i, \] is analyzed through an augmented system on \((x,\tilde x)\) and a continuous piecewise-quadratic incremental Lyapunov or storage function. On diagonal cells \(X_{ii}\), the storage must reduce to a quadratic form in \(x-\tilde x\); on off-diagonal cells \(X_{ij}\), it may depend piecewise-quadratically on \((x,\tilde x,1)\). The main theorem gives LMIs involving regional matrices \(P_i\), \(\bar P_{ij}\), S-procedure multipliers, continuity-enforcing matrices, and positive scalars \(\sigma_1,\sigma_2,\sigma_3\). These LMIs enforce lower and upper quadratic bounds and strict decrease along same-input trajectories, yielding an incremental exponential estimate and therefore \(\delta\)-GAS when \(X=\mathbb R^n\). The method is explicitly presented as less conservative than single-quadratic certificates, but it remains sufficient, assumes no Zeno behavior, requires a common feedthrough matrix \(D\), and may fail when local subsystem matrices \(A_i\) are non-Hurwitz [1611.08322]. A more recent verification direction is fully data-driven and compositional. For an interconnected discrete-time network of unknown nonlinear homogeneous subsystems \[ x_i(k+1)=f_i(x_i(k),w_i(k)), \] with each \(f_i\) continuous and homogeneous of degree one, subsystem-level incremental stability is treated as \(\delta\)-ISS with respect to the internal inputs \(w_i\). Each subsystem Lyapunov candidate is parameterized as \[ \mathcal S_i(q_i,x_i,x_i')=\sum_{j=1}^r q_i^j\,p_i^j(x_i,x_i'), \] with homogeneous basis functions. Because the dynamics are unknown, the exact robust optimization program is replaced by a scenario optimization program based on one-step transition data normalized to the unit sphere. If the optimizer \(\mathcal S_i^*\) satisfies the deterministic correctness condition \[ \mu_{\mathcal N_i}^*+\mathscr L_i\varepsilon_i\le 0, \] where \(\varepsilon_i\) is a covering radius on the unit sphere and \(\mathscr L_i\) is a Lipschitz bound for the constraint functions, then \(\mathcal S_i^*\) is a valid subsystem \(\delta\)-ISS Lyapunov function [2507.18141]. Network-level \(\delta\)-GAS is then obtained through a small-gain composition. Defining \[ \Gamma=\operatorname{diag}(1-\gamma_1^*,\ldots,1-\gamma_{\mathcal M}^*),\qquad \hat\delta_{ij}=\frac{\rho_i^*}{\underline{\alpha}_j^*}, \] the compositional condition is \[ \mathds 1_{\mathcal M}^\top(-\Gamma+\Delta)<0. \] Under that condition, \[ \mathcal V(q,x,x')=\sum_{i=1}^{\mathcal M}\mathcal S_i^*(q_i,x_i,x_i') \] is an incremental Lyapunov function for the full interconnection and certifies \(\delta\)-GAS. The paper emphasizes that, for fixed subsystem dimensions, the compositional approach yields sample complexity growing linearly with subsystem count, whereas the monolithic approach exhibits exponential growth. Its large-scale example treats a homogeneous ring network with \(10000\) subsystems and reports successful certification with a correctness guarantee [2507.18141]. ## 5. Delay systems, stochastic analogues, and controller synthesis Time-delay systems make the uniformity aspect of \(\delta\)-GAS especially delicate. The paper on retarded time-delay systems does not define incremental stability directly, but it is relevant because incremental arguments are commonly reduced to ordinary GAS or UGAS of a product or error system. For \[ \dot x(t)=f(x_t), \] with state in \(C^0([-r,0];\mathbb R^n)\), it proves that \((GAS)\wedge(RFC)\Rightarrow(UGAS)\). It also proves that, in the stronger spaces \(X=W^{1,p}([-r,0])\) or \(X=C^{0,1-1/p}([-r,0])\) with \(p\in(1,+\infty]\), forward completeness implies RFC, and consequently \(GAS\text{-}X\Rightarrow UGAS\text{-}X\). The same source provides Lyapunov-Krasovskii characterizations of UGAS-X and, for \(W^{1,p}\), \(1