Semicontinuity conditions for set-valued equilibrium problems and applications to set optimization problems

From MaRDI portal





This paper studies parametric set-valued equilibrium problems without assuming the ordered cone has nonempty interior or that the objective map has compact values. The author introduces a new quasiconvexity property for set-valued maps and establishes sufficient conditions for the semicontinuity and continuity of approximate solution maps. The results are then applied to set optimization problems using the possibly less order relation, relaxing standard convexity and compactness assumptions from the literature.\N\N\textbf{Comments}\N\N\begin{itemize}\N\item[1.] \textbf{Novelty claim unclear:} The abstract and introduction emphasize ``without assuming the solid condition of the ordered cone and the compact values, but similar relaxations already appear in [\textit{L. Q. Anh} et al., Optim. Lett. 18, No. 8, 1907--1924 (2024; Zbl 08055204)] and [\textit{M. Durea} and \textit{E.-A. Florea}, J. Optim. Theory Appl. 200, No. 3, 1286--1308 (2024; Zbl 1566.49019); J. Nonlinear Var. Anal. 8, No. 4, 533--547 (2024; Zbl 07870914)]. Please clarify the specific new contribution.\N\N\item[2.] \textbf{Definition 1 vs. standard approximation:} The proposed \(\epsilon\)-solution uses \(\epsilon \mathbb{B}_+\) instead of \(\epsilon e\) with \(e \in \operatorname{int}\mathcal{C}\). While this removes the interior requirement, the equivalence in Remark 1(b) assumes \(\mathbb{B}_+ \subset \mathbb{S}_+ - \mathcal{C}\). Is this always valid without interior conditions?\N\N\item[3.] \textbf{Theorem 1 proof gap:} The proof constructs \(S(\epsilon,\lambda)\) with \(\operatorname{int}\mathbb{B}_+\) but later claims \(\operatorname{Sol}_{+}(\mathrm{SEP})(\epsilon_0,\lambda_0) \subset \operatorname{cl} S(\epsilon_0,\lambda_0)\). The quasiconvexity argument uses \(\operatorname{int}\mathbb{B}_+\) on one side and \(\mathbb{B}_+\) on the other, please justify this mismatch.\N\N\item[4.] \textbf{Theorem 2 closedness assumption:} The proof uses that \(-\mathbb{B}_+ + \mathcal{C}\) is closed. Is this always true for arbitrary closed convex cones \(\mathcal{C}\)? If not, additional conditions may be needed.\N\N\item[5.] \textbf{Lemma 5 proof issue:} The implication from \(z_t = tz_2 + (1-t)z_1 + c\) to \(z_t + (1-t)r_1b_1 + tr_2b_2 \in \mathcal{C}\) assumes \(c \in \mathcal{C}\) and the sum of \(\mathcal{C}\)-elements stays in \(\mathcal{C}\). Fine, but the final conclusion uses \(\operatorname{int}\mathbb{B}_+\), please verify the interior preservation under convex combination.\N\N\item[6.] \textbf{Example 1 verification incomplete:} The claim ``\(r_1 \geq 2\) and \(r_2 > 2\) is stated without derivation. Please show the explicit computation for readers.\N\N\item[7.] \textbf{Corollary 1 assumption (ii):} ``\((\pm \mathcal{C})\)-continuous means both \(\mathcal{C}\)-usc and \((-\mathcal{C})\)-usc? This needs clarification, as the proof only uses \((-\mathcal{C})\)-continuity.\N\N\item[8.] \textbf{Typo in Lemma 1 reference:} The citation ``Hu and Papageorgiou 1997, p. 37 is correct, but Lemma 1(b) has a formatting issue with the definition of \(\liminf\).\N\N\item[9.] \textbf{Notation inconsistency:} \(\mathbb{R}_{++}\) appears in the abstract and assumptions, but \(\mathbb{R}_{++}\) is never explicitly defined (only \(\mathbb{R}_+\) and \(\mathbb{R}_{++}\) mentioned on page 4, line 4). Define clearly.\N\N\item[10.] \textbf{Equation numbering missing:} Several displayed equations (e.g., after ``Consequently on page 8, the inequality on page 9) have no numbers, making referencing difficult.\N\N\item[11.] \textbf{Example 2 computation:} The final solution set \(\left[-\sqrt{\epsilon + \lambda^2},\sqrt{\epsilon + \lambda^2}\right]\) seems independent of \(y\), but \(F\) depends on \(y^2 - x^2\). Please verify.\N\end{itemize}\N\N\textbf{Recommendation}\N\NThe paper presents interesting extensions, but the novelty must be sharpened, and several technical gaps in the proofs (especially regarding interior points and closedness properties) need addressing.



Cites work









This page was built for publication: Semicontinuity conditions for set-valued equilibrium problems and applications to set optimization problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6834995)