Tseng’s Type Methods in Continuous and Discrete Time for Quasi-Variational Inequalities
Lkhamsuren Altangerel · 2025
Read Tseng’s Type Methods in Continuous and Discrete Time for Quasi-Variational Inequalities by Lkhamsuren Altangerel (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-04102.txt.
Introduction
Lkhamsuren Altangerel adapts in 2025 a known fix-point scheme to quasi-variational inequality problems in Hilbert spaces, in both continuous and discrete time. The work studies when equilibrium points exist for these problems. It proves convergence for the linked dynamical systems. Linear convergence of the split forms is then checked through examples, illustrations, and special cases.
Catalog record
| CREATOR | Lkhamsuren Altangerel |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-04102
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Tseng’s Type Methods in Continuous and Discrete Time for Quasi-Variational Inequalities Lkhamsuren Altangerel Abstract This paper presents an approach for obtaining approximate solutions to quasi-variational inequalities in a real Hilbert space by modifying Tseng’s scheme, which was originally designed for variational inequalities. The study explores the existence of equilibrium points and investigates convergence results related to dynamical systems. Linear convergence for discretized systems is examined through examples, illustrations, and special cases. This work was partially supported by a grant from the IMU-CDC and Simons Foundation 1 Introduction Quasi-variational inequalities represent a significant extension of classical variational inequalities, where the constraint set depends on the solution itself. They were initially introduced by Bensoussan et al. [ 3 ] in the context of impulse control problems and have since evolved into powerful mathematical tools with applications in economics, engineering, operations research, and partial differential equations. A standard sufficient condition for existence and uniqueness under Lipschitz continuous strongly monotone mapping was provided in [ 15 ] . For further study, we refer to [ 2 , 6 , 14 , 16 ] . There have been some results regarding the existence and uniqueness of solutions for quasi-variational inequalities under weaker assumptions, such as strongly pseudomonotonicity [ 12 , 13 ] . Additionally, the relationships with other models such as the generalized Nash equilibrium problems [ 6 ] , and multi-leader follower games [ 17 ] have been investigated. Regarding the applications of quasi-variational inequalities we refer to [ 19 ] in telecommunication networks, [ 18 ] in transportation networks, and [ 9 ] in image restoration. Recent advances in computational mathematics have led to the development of more efficient algorithms for solving QVIs. Antipin et al. [ 1 ] (also see [ 11 ] ) presented gradient projection and extragradient methods for solving QVIs under the assumptions of strong monotonicity and Lipschitz continuity of associated mappings. Others include inertial projection-type methods [ 5 , 20 , 21 , 23 ] , C-FISTA type projection algorithm [ 22 ] , and gradient-type projection methods [ 10 ] . Motivated by the above results, we introduce a new approach for solving QVI based on Tseng’s scheme ( [ 24 ] , also see [ 4 ] ), which has been investigated for variational inequality problems. Compared to existing algorithms, the main contributions of the paper can be highlighted as follows: • To the best of our knowledge the Tseng scheme for variational inequalities has been extended to quasi-variational inequalities, sparking broad research interest in this direction. • Continuous and discrete time algorithms have been investigated for quasi-variational inequalities in the general case with set-valued mappings. • Similar to the inertial projection-type methods, the projection and mapping values will be evaluated once in each iteration. The outline of the paper is as follows. Section 2 will review the definitions and preliminary results used in this paper. In Section 3, we will present the existence of equilibrium points and convergence results related to dynamical systems. Section 4 is dedicated to examining linear convergence for discretized systems, with some examples, illustrations, and special cases included. The paper concludes with a mention of research ideas that can be considered in the future. 2 Preliminaries Let \mathcal{H} be a real Hilbert space with inner product \langle\cdot,\cdot\rangle , and topology induced by the norm \|x|=\sqrt{\langle x,x\rangle} . Let F:\mathcal{H}\rightarrow\mathcal{H} a nonlinear operator, and K:\mathcal{H}\rightrightarrows\mathcal{H} be a set-valued mapping such that for each x\in\mathcal{H},~K(x) is a nonempty closed convex subset of \mathcal{H}. The Quasi-Variational Inequality consists of finding x^{*}\in K(x^{*}) such that (QVI)\hskip 28.45274pt\langle F(x^{*}),y-x^{*}\rangle\geq 0~~\forall y\in K(x^{*}). Remark 1 . In the case of a single set K , the problem (QVI) reduces to the classical variational inequality of finding x^{*}\in K (VI)\hskip 28.45274pt\langle F(x^{*}),y-x^{*}\rangle\geq 0~~\forall y\in K. Definition 1 . Let F:\mathcal{H}\rightarrow\mathcal{H} be a given mapping. The mapping F is said to be: (i) L -Lipschitz continuous with L>0 if \|F(x)-F(y)\|\leq L\|x-y\|~~\forall x,y\in\mathcal{H}. (ii) monotone if \langle x-y,F(x)-F(y)\rangle\geq 0~~\forall x,y\in\mathcal{H}. (iii) \rho -strongly monotone if there exists \rho>0 such \langle x-y,F(x)-F(y)\rangle\geq\rho\|x-y\|^{2}~~\forall x,y\in\mathcal{H}. The projection P_{C}:\mathcal{H}\to C onto the nonempty, closed, convex set C\subseteq\mathcal{H} is defined by P_{C}(x)=\mbox{argmin}\{\|x-y\|:~y\in C\}. Note that for each x\in\mathcal{H}, the implicit projection P_{K(x)} is nonexpensive, i.e., \|P_{K(x)}(u)-P_{K(x)}(v)\|\leq\|u-v\|~~\forall u,v\in\mathcal{H}. Proposition 1 . [ 15 ] Let F:\mathcal{H}\rightarrow\mathcal{H} be L -Lipschitz continuous and \rho -strongly monotone on \mathcal{H}. Assume that K:\mathcal{H}\rightrightarrows\mathcal{H} is a set-valued mapping with nonempty, closed and convex values. Setting \gamma:=\frac{L}{\rho}\geq 1 and assume that there is some constant 0\leq l\leq\frac{1}{\gamma(\gamma+\sqrt{\gamma^{2}-1})} such that \|P_{K(x)}(z)-P_{K(y)}(z)\|\leq l\|x-y\|~~\forall x,y,z\in\mathcal{H}, (1) then the quasi-variational inequality (QVI) has a unique solution x^{*}. Remark 2 . In [ 14 ] , it was shown that uniqueness is guaranteed as long as 0\leq l\leq\frac{1}{\gamma}. Lemma 2.1 . Let K:\mathcal{H}\rightrightarrows\mathcal{H} be a set-valued mapping with nonempty, closed and convex subset in \mathcal{H}. x^{*}\in K(x^{*}) is solution of (QVI) if and only if for any \alpha>0 it holds x^{*}=P_{K(x^{*})}(x^{*}-\alpha F(x^{*})). Lemma 2.2 . Let z\in\mathcal{H}. The necessary and sufficient characterizations of the projection are P_{K(x)}(z)\in K(x)\Leftrightarrow~\langle P_{K(x)}(z)-z,y-P_{K(x)}(z)\rangle\geq 0~~\forall y\in K(x), or, equivalently \langle z-P_{K(x)}(z),y-P_{K(x)}(z)\rangle\leq 0~~\forall y\in K(x). Lemma 2.3 . Under the assumptions of Proposition 1 it holds \|P_{K(x)}(x-\lambda F(x))-P_{K(y)}(y-\lambda F(y))\|\leq\theta\|x-y\|\quad\forall x,y\in\mathcal{H}, (2) where \theta=l+\sqrt{1-2\lambda\rho+\lambda^{2}L^{2}}. Proof. Let x,y\in\mathcal{H}. Then, it holds \displaystyle\|P_{K(x)}(x-\lambda F(x))-P_{K(y)}(y-\lambda F(y))\| \displaystyle\leq \displaystyle\|P_{K(x)}(x-\lambda F(x))-P_{K(x)}(y-\lambda F(y))\|+\|P_{K(x)}(y-\lambda F(y))-P_{K(y)}(y-\lambda F(y))\| \displaystyle\leq \displaystyle\|x-\lambda F(x)-(y-\lambda F(y))\|+l\|x-y\|. On the other hand, we have \displaystyle\|x-\lambda F(x)-(y-\lambda F(y))\|^{2} \displaystyle= \displaystyle\|x-y\|^{2}+\lambda^{2}\|F(x)-F(y)\|^{2}-2\lambda\langle x-y,F(x)-F(y)\rangle \displaystyle\leq \displaystyle(1+\lambda^{2}L^{2}-2\lambda\rho)\|x-y\|^{2}, or, equivalently \|x-\lambda F(x)-(y-\lambda F(y))\|\leq\sqrt{1+\lambda^{2}L^{2}-2\lambda\rho}\|x-y\|. (3) By considering that \theta=l+\sqrt{1+\lambda^{2}L^{2}-2\lambda\rho}, it holds \|P_{K(x)}(x-\lambda F(x))-P_{K(y)}(y-\lambda F(y))\|\leq\theta\|x-y\|. (4) ∎ 3 Dynamical system of the Tseng type Let T:\mathcal{H}\rightarrow\mathcal{H} be a continuous mapping, and x:(0,+\infty)\rightarrow\mathcal{H}. We recall the stability concepts of an equilibrium point of the general dynamical system \dot{x}(t)=T(x(t))~~t\geq 0. (5) Definition 2 . (i) A point x^{*}\in\mathcal{H} is an equilibrium point for ( 5 ) if T(x^{*})=0. (ii) An equilibrium point x^{*} of ( 5 ) is stable if, \forall\varepsilon>0,~\exists\delta>0 such that, \forall x_{0}\in B(x^{*},\delta), the solution x(t) of the dynamical system with x(0)=x_{0} exists and x(t)\in B(x^{*},\varepsilon):=\{y\in\mathcal{H}|~\|x^{*}-y\|<\varepsilon\}~~\forall t>0. (iii) An equilibrium point x^{*} of ( 5 ) is exponentially stable if \exists\varepsilon>0 and constants k>0,~\delta>0 such that, for each solution x(t) of ( 5 ) with x(0)\in B(x^{*},\varepsilon), one has \|x(t)-x^{*}\|\leq k\|x(0)-x^{*}\|e^{-\delta t}~~\forall t\geq 0. We consider the following dynamical system of the Tseng type \displaystyle y(t)=P_{K(x(t))}(x(t)-\lambda F(x(t))) \displaystyle\dot{x}(t)+x(t)=y(t)+\lambda(F(x(t))-F(y(t))) \displaystyle x(0)=x_{0}, where \lambda>0 and x_{0}\in\mathcal{H}. The system is reduced to a dynamical system of type \dot{x}=f(x), where f(x)=P_{K(x)}(x-\lambda F(x))+\lambda(F(x)-F(P_{K(x)}(x-\lambda F(x)))-x. (6) Theorem 3.1 . Let the assumptions of Proposition 1 be fulfilled. Then the dynamical system ( 6 ) has an unique solution. Proof. \displaystyle\|f(x)-f(z)\|\leq\|z-x\|+\|P_{K(z)}(z-\lambda F(z))-P_{K(x)}(x-\lambda F(x))\| \displaystyle+ \displaystyle\lambda\|F(z)-F(x)\|+\lambda\|F(P_{K(z)}(z-\lambda F(z))-F(P_{K(x)}(x-\lambda F(x))\| \displaystyle\leq \displaystyle(1+\lambda L)\|z-x\|+(1+\lambda L)\|P_{K(z)}(z-\lambda F(z))-P_{K(x)}(x-\lambda F(x))\| \displaystyle\leq \displaystyle(1+\lambda L)\|z-x\|+(1+\lambda L)\|P_{K(z)}(z-\lambda F(z))-P_{K(x)}(z-\lambda F(z))\| \displaystyle+ \displaystyle(1+\lambda L)\|P_{K(x)}(z-\lambda F(z))-P_{K(x)}(x-\lambda F(x))\| \displaystyle\leq \displaystyle(1+l)(1+\lambda L)\|z-x\|+(1+\lambda L)\|z-\lambda F(z)-(x-\lambda F(x)\|. Using ( 3 ), it becomes \|f(x)-f(z)\|\leq(1+\lambda L)(1+l+\sqrt{1-2\rho\lambda+\lambda^{2}L^{2}})\|z-x\|=(1+\lambda L)(1+\theta)\|z-x\|, (7) which shows that f is a Lipchitz continues, and the existence and uniqueness follows from the classical Cauchy-Lipschitz-Picard theorem ( [ 7 ] ). ∎ Theorem 3.2 . Let the assumptions of Proposition 1 be fulfilled. Assume that (1+\lambda L)(1+\theta)<2. Then the dynamical system ( 6 ) converges to the solution of the (QVI) at the rate \|x(t)-x^{*}\|\leq\|x_{0}-x^{*}\|e^{\Lambda t}, where x^{*} is the equilibrium point of the system. Proof. We consider the Lyapunov