A study on ℱ \mathscr{F} -simultaneous approximative τ \tau -compactness property in Banach spaces Syamantak Das, Tanmoy Paul Abstract . Veselý (1997) studied Banach spaces that admit f f -centers for finite subsets of the space. In this work, we introduce the concept of ℱ \mathscr{F} -simultaneous approximative τ \tau -compactness property ( τ \tau - ℱ \mathscr{F} -SACP in short) for triplets ( X , V , 𝔉 ) (X,V,\mathfrak{F}) , where X X is a Banach space, V V is a τ \tau -closed subset of X X , 𝔉 \mathfrak{F} is a subfamily of closed and bounded subsets of X X , ℱ \mathscr{F} is a collection of functions, and τ \tau is the norm or weak topology on X X . We characterize reflexive spaces with the Kadec-Klee property using triplets with τ \tau - ℱ \mathscr{F} -SACP. We investigate the relationship between τ \tau - ℱ \mathscr{F} -SACP and the continuity properties of the restricted f f -center map. The study further examines τ \tau - ℱ \mathscr{F} -SACP in the context of C ​ L ​ U ​ R CLUR spaces and explores various characterizations of τ \tau - ℱ \mathscr{F} -SACP, including connections to reflexivity, Fréchet smoothness, and the Kadec-Klee property. 2000 Mathematics Subject Classification Primary 41A28, 46B20 Secondary 41A65, 41A50 August 9, 2026 1. Introduction 1.1. Prerequisites: We list some standard notations used in this study: X denotes a Banach space, and by a subspace Y of X , we refer to a closed linear subspace. B_{X} and S_{X} represent the closed unit ball and unit sphere of X , respectively. \mathcal{F}(X),\penalty\ \mathcal{CC}(X) and \mathcal{CB}(X) denote the collection of all nonempty finite, closed and convex, closed and bounded subsets of X , respectively. For F\in\mathcal{F}(X) , card(F) denotes the cardinality of the set F . For f\in S_{X^{*}} , we define J_{X}(f)=\{x\in S_{X}:f(x)=1\} . NA(X) denotes the set of all f\in S_{X^{*}} such that f attains its norm on B_{X} . The real line is assumed to correspond to the scalar field of the space. In this study, \tau denotes the norm or weak topology on X . For F\in\mathcal{CB}(X) , we define a subclass \ell_{\infty}^{+}(F) by \ell_{\infty}^{+}(F)=\{\varphi:F\rightarrow\mathbb{R}_{\geq 0}:\underset{t\in F}{\sup}\penalty\ \varphi(t)<\infty\} . On the set \ell_{\infty}^{+}(F) , we consider a coordinate-wise ordering defined as: for \varphi_{1},\varphi_{2}\in\ell_{\infty}^{+}(F) , \varphi_{1}\leq\varphi_{2} if and only if \varphi_{1}(t)\leq\varphi_{2}(t) for all t\in F . Definition 1.1 . For F\in\mathcal{CB}(X) and a function f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} , we state that (a) f is monotone if for \varphi_{1},\varphi_{2}\in\ell_{\infty}^{+}(F) , \varphi_{1}\leq\varphi_{2} implies f(\varphi_{1})\leq f(\varphi_{2}) . (b) f is weakly strictly monotone if f is monotone and for \varphi_{1},\varphi_{2}\in\ell_{\infty}^{+}(F) , \varphi_{1}(t)<\varphi_{2}(t) for all t\in F implies f(\varphi_{1})0 such that P_{V}(x,\delta)\subseteq P_{V}(x)+W , where P_{V}(x,\delta)=\{v\in V:\|v-x\|0 , \delta-\textrm{Cent}_{V}^{f}(F) is always non-empty. For V\in\mathcal{CC}(X) and F\in\mathcal{CB}(X) , when f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} is of the form f(a)=\underset{t\in F}{\sup}\rho(t)a(t) , where \rho=(\rho(t))_{t\in F}\in\ell_{\infty}^{+}(F) , then the quantities r_{f}(x,F) , \textrm{rad}_{V}^{f}(F) and \textrm{Cent}_{V}^{f}(F) are denoted by r_{\rho}(x,F) , \textrm{rad}_{V}^{\rho}(F) and \textrm{Cent}_{V}^{\rho}(F) respectively. When \rho(t)=1 for all t\in F , then the above quantites are denoted by r(x,F) , \textrm{rad}_{V}(F) and \textrm{Cent}_{V}(F) respectively, the so-called farthest distance from x to F , the restricted Chebyshev radius of F in V and the restricted Chebyshev center of F in V respectively. We refer to [ 3 , 4 , 5 ] , a few previous works of the authors in this line of investigation. Definition 1.3 . Let V be a \tau -closed subset of X , \mathfrak{F}\subseteq\mathcal{CB}(X) and \mathscr{F} be a collection of functions f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} for all F\in\mathfrak{F} . (a) The triplet (X,V,\mathfrak{F}) is said to have restricted \mathscr{F} -center property ( \mathscr{F} -rcp in short) if for all F\in\mathfrak{F} and functions f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} , \textrm{Cent}_{V}^{f}(F)\neq\emptyset . (b) The triplet (X,V,\mathfrak{F}) is said to have \tau - \mathscr{F} -property- (P_{1}) if, for all F\in\mathfrak{F} and all functions f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} in \mathscr{F} , \textrm{Cent}_{V}^{f}(F)\neq\emptyset and for all \tau -neighborhoods W of 0 , there exists \delta>0 such that \delta-\textrm{Cent}_{V}^{f}(F)\subseteq\textrm{Cent}_{V}^{f}(F)+W . (c) The triplet (X,V,\mathfrak{F}) is said to have \mathscr{F} -simultaneous approximative \tau -compactness property ( \tau - \mathscr{F} -SACP in short) if, for all F\in\mathfrak{F} , all functions f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} in \mathscr{F} and any sequence (v_{n})\subseteq V such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) implies that (v_{n}) has a \tau -convergent subsequence. (d) The triplet (X,V,\mathfrak{F}) is said to have weighted simultaneous approximative \tau -compactness property (weighted- \tau -SACP in short) if, for all F\in\mathfrak{F} , all bounded weights \rho=(\rho(t))_{t\in F} , and any sequence (v_{n})\subseteq V such that r_{\rho}(v_{n},F)\rightarrow\textrm{rad}_{V}^{\rho}(F) implies that (v_{n}) has a \tau -convergent subsequence. When \tau is the norm topology on X , we simply say that (X,V,\mathfrak{F}) has \mathscr{F} -property- (P_{1}) (for (b) ), \mathscr{F} -SACP ( for (c) ) or weighted-SACP (for (d) ). It is clear from definition 1.3 that, if \mathcal{F}_{s} is the collection of all singletons in X and f is the identity map on \mathbb{R}_{\geq 0} , then a \tau -closed subset V of X is approximatively \tau -compact ( \tau -strongly proximinal) in X if and only if (X,V,\mathcal{F}_{s}) has \tau - \{f\} -SACP ( \tau - \{f\} -property- (P_{1}) ). Let us observe that for a \tau -closed subset V of X , if the triplet (X,V,\mathcal{F}(X)) has \tau - \mathscr{F}_{cm} -SACP, then from [ 17 , Proposition 1.4 (i), (iv)] , it follows that (X,V\mathcal{F}(X)) has \mathscr{F}_{cm} -rcp. It has been observed in [ 2 , Exmaple 2.1] that c_{0} is not approximatively \tau -compact in \ell_{\infty} , however, it is well-known that c_{0} is strongly proximinal in \ell_{\infty} . Thus, when \mathcal{F}_{s} is the collection of all singletons in \ell_{\infty} and f is the identity map on \mathbb{R}_{\geq 0} , the triplet (\ell_{\infty},c_{0},\mathcal{F}_{s}) has \{f\} -property- (P_{1}) but does not have \tau - \{f\} -SACP. We refer to example 2.1 for an example of a subspace Y of X such that Y is approximatively compact in X but (X,Y,\mathcal{F}(X)) does not have \tau - \mathscr{F} -SACP. Let (\Omega,\Sigma,\mu) be a finite measure space. For 1\leq p<\infty , L_{p}(\Sigma,X) denotes the space of all \mu -strongly measurable X -valued p -Bochner integrable functions f:\Omega\rightarrow X with \|f\|_{p}:=\left(\int_{\Omega}\|f(t)\|^{p}d\mu(t)\right)^{\frac{1}{p}} . Let us recall the notions of L -projections, L -summands, and L -embedded spaces from [ 10 ] . Definition 1.4 . (a) A linear projection P:X\rightarrow X is said to be an L -projection if \|x\|=\|Px\|+\|x-Px\| for all x\in X . (b) A subspace Y of X is said to be an L -summand if Y is the range of an L -projection. (c) X is said to be L -embedded if X , under its canonical image in X^{**} , is the range of an L -projection on X^{**} . [ 10 , Ch. 4] serves as a reference for examples and other properties of L -embedded spaces. It is shown in [ 10 , Theorem IV.1.5] that, if Y is the range of a norm-1 projection of an L -embedded space X , Y is also L -embedded. Definition 1.5 . (a) [ 18 ] X is said to be \tau -compactly locally uniformly rotund ( \tau-CLUR in short) if, for every x\in S_{X} and every sequence (x_{n})\subseteq B_{X} satisfying \|\frac{x_{n}+x}{2}\|\rightarrow 1 , (x_{n}) has a \tau -convergent subsequence. (b) [ 1 ] X is said to be \tau -almost locally uniformly rotund ( \tau-ALUR in short) if, for every x,x_{n}\in S_{X} and h_{m}\in S_{X^{*}} such that \lim_{m}\lim_{n}h_{m}(\frac{x_{n}+x}{2})=1 , x_{n}\xrightarrow{\tau}x . (c) X is said to have the Kadec-Klee property if x,x_{n}\in S_{X} with x_{n}\xrightarrow{w}x implies x_{n}\rightarrow x in the norm. It follows from [ 1 , Proposition 4.4 and Corollary 4.6] that X is \tau-ALUR if and only if for every f\in NA(X) and (x_{n})\subseteq B_{X} such that f(x_{n})\rightarrow 1 implies (x_{n}) is \tau -convergent. Let us recall property (**) from [ 15 ] which later became known as the Namioka-Phelps property. Definition 1.6 . X is said to have the Namioka-Phelps property if w^{*} and norm convergence coincide on S_{X^{*}} . Let us recall the following definitions from [ 13 ] . Definition 1.7 . Let X be a Banach space. (a) The norm of X is said to be differentiable at x\in S_{X} if \underset{t\rightarrow 0}{\lim}\frac{\|x+ty\|-1}{t} exists for y\in X . (b) The norm of X is said to be Fréchet differentiable at x\in S_{X} if the limit in (a) exists and the convergence is uniform over y\in S_{X} . A Banach space X is said to be smooth ( Fréchet smooth ) if the norm is differentiable (Fréchet differentiable) at each x\in S_{X} . 1.2. Observations: In section 2, we study the relationship between \tau - \mathscr{F} -property- (P_{1}) and \tau - \mathscr{F} -SACP for a triplet (X,V,\mathfrak{F}) . We study reflexivity and the Kadec-Klee property in relation to \tau - \mathscr{F} -SACP for a triplet (X,V,\mathfrak{F}) . In section 3, we derive that Fréchet smooth spaces are precisely those spaces in which for every w^{*} -closed convex subset V of X^{*} , (X^{*},V,\mathcal{F}(X^{*})) has \mathscr{F} -SACP, where \mathscr{F} is the collection of monotone smooth norms and \textrm{Cent}_{V}^{\gamma}(F) is singleton for all F\in\mathcal{F}(X^{*}) , and for all monotone norms \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} where card(F)=N . We further investigate several rotundity properties in relation to \tau - \mathscr{F} -SACP. In section 4, we study the continuity properties of certain set-valued mappings. 2. Various characterizations of \tau - \mathscr{F} -SACP Let us recall that a norm \gamma on \mathbb{R}^{3} is symmetric if \gamma=\gamma\circ\pi for every permutation \pi of the three variables. Example 2.1 . Let Y be a non-reflexive Banach space. Then from [ 17 , Theorem 2.4] , there exists a three-point set F\subseteq X such that for any monotone symmetric norm \gamma:\mathbb{R}^{3}\rightarrow\mathbb{R}_{\geq 0} , Y admits an equivalent norm |||\penalty\ .\penalty\ ||| such that \emph{Cent}_{Y}^{\gamma}(F)=\emptyset in (Y,|||\penalty\ .\penalty\ |||) . Let Y^{\prime} denote the Banach space Y equipped with the norm |||\penalty\ .\penalty\ ||| , and let X=Y^{\prime}\oplus_{\ell_{1}}\mathbb{R} . It is evident that Y^{\prime} , being an L -summand, is approximatively compact in X . However, the triplet (X,Y^{\prime},\mathcal{F}(X)) does not have \tau - \mathscr{F}_{cm} - \bf{\emph{SACP}} as \emph{Cent}_{Y^{\prime}}^{\gamma}(F)=\emptyset . The following result demonstrates the relationship between \tau - \mathscr{F}_{cm} -property- (P_{1}) and \tau - \mathscr{F}_{cm} -SACP for a triplet (X,V,\mathcal{A}) , where V is a \tau -closed subset of X and \mathcal{A}\subseteq\mathcal{F}(X) . Theorem 2.2 . Let V be a \tau -closed subset of X and \mathcal{A}\subseteq\mathcal{F}(X) . Then, (X,V,\mathcal{A}) has \tau - \mathscr{F}_{cm} - \bf{\emph{SACP}} if and only if (X,V,\mathcal{A}) has the \tau - \mathscr{F}_{cm} -property- (P_{1}) and \emph{Cent}_{V}^{f}(F) is \tau -compact for all F\in\mathcal{A} and f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} in \mathscr{F}_{cm} . Proof. The necessity follows from the proof of [ 2 , Thereom 2.2] . We now derive the converse. Let F\in\mathcal{A} , f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} be a function in \mathscr{F}_{cm} and (v_{n})\subseteq V be such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) . Let W be a \tau -neighbourhood of 0 and W_{1},W_{2} be two \tau -open neighbourhoods of 0 such that W_{1}+W_{2}\subseteq W . As (X,V,\mathcal{A}) has the \tau - \mathscr{F}_{cm} -property- (P_{1}) , there exists \delta>0 such that \delta-\textrm{Cent}_{V}^{f}(F)\subseteq\textrm{Cent}_{V}^{f}(F)+W_{1} . Now, there exists N_{1}\in\mathbb{N} such that r_{f}(v_{n},F)\leq\textrm{rad}_{V}^{f}(F)+\delta for all n\geq N_{1} , i.e., v_{n}\in\delta-\textrm{Cent}_{V}^{f}(F) for all n\geq N_{1} . Thus, there exists a sequence (w_{n})\subseteq\textrm{Cent}_{V}^{f}(F) such that v_{n}-w_{n}\in W_{1} for all n\geq N_{1} . As \textrm{Cent}_{V}^{f}(F) is \tau -compact, (w_{n}) has a \tau -convergent subsequence (w_{n_{k}}) converging to some w\in W . Thus there exists N_{2}\in\mathbb{N} such that w_{n_{k}}\in\{w\}+W_{2} for all k\geq N_{2} and so, v_{n_{k}}=(v_{n_{k}}-w_{n_{k}})+w_{n_{k}}\in\{w\}+W for all k\geq\max\{N_{1},N_{2}\} . This establishes our assertion. ∎ The following theorem characterizes reflexive spaces based on triplets satisfying weak- \mathscr{F}_{cmc} -SACP. Using the fact that for a F\in\mathcal{CB}(X) and a certain f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} in \mathscr{F}_{cmc} , due to the coercivity of f , any sequence (v_{n})\subseteq V such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) implies that (v_{n}) is bounded. Thus we have the following. Theorem 2.3 . If X is reflexive, then for every V\in\mathcal{CC}(X) , the triplet (X,V,\mathcal{CB}(X)) has weak- \mathscr{F}_{cmc} - \bf{\emph{SACP}} . Theorem 2.3 leads to the following observation. Corollary 2.4 . If C is a weakly compact subset of X , then (X,C,\mathcal{CB}(X)) has weak- \mathscr{F}_{cmc} - \bf{\emph{SACP}} . Corollary 2.5 . Let C be a weakly compact convex subset of X and 1\leq p<\infty . Then (L_{p}(\Sigma,X),L_{p}(\Sigma,C),\mathcal{CB}(L_{p}(\Sigma,X)) has weak- \mathscr{F}_{cmc} - \bf{\emph{SACP}} . Proof. Let us recall from [ 8 , Theorem 2] that if C is a weakly compact convex subset of X , then for 1\leq p<\infty , L_{p}(\Sigma,C) is weakly compact in L_{p}(\Sigma,X) . Thus, from proposition 2.4 , our assertion follows. ∎ We now characterize reflexive spaces with the Kadec-Klee property. Theorem 2.6 . For a Banach space X , the following are equivalent. (a) X is reflexive and has the Kadec-Klee property. (b) For every V\in\mathcal{CC}(X) , (X,V,\mathcal{F}(X)) has \mathscr{F}_{cwmc} - \bf{\emph{SACP}} . Proof. It remains to prove (a)\Rightarrow(b) . Let F=\{x_{1},\cdots,x_{N}\}\in\mathcal{F}(X) , f:\mathbb{R}^{N}_{\geq 0}\rightarrow\mathbb{R}_{\geq 0} be a function in \mathscr{F}_{cwmc} and (v_{n})\subseteq V be such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) . Since f is coercive, (v_{n}) is bounded. Now, by the reflexivity of X , (v_{n}) has a subsequence (v_{n_{k}}) , converging weakly to v_{0}\in V . By passing on to a subsequence, if necessary, let \mu_{i}=\underset{k}{\lim}\|v_{n_{k}}-x_{i}\| for all i=1,\cdots,N . Now, by the lower semicontinuity of the norm, we have \|v_{0}-x_{i}\|\leq\mu_{i} for all i=1,\cdots,N . From [ 17 , Proposition 1.4(iv)] , by the lower semicontinuity of the function r_{f}(.,F) , we obtain \textrm{rad}_{V}^{f}(F)\leq r_{f}(v_{0},F)\leq\liminf_{k}r_{f}(v_{n_{k}},F)=\textrm{rad}_{V}^{f}(F). Thus v_{0}\in\textrm{Cent}_{V}^{f}(F) . Claim: There exists i\in\{1,\cdots,N\} such that \|v_{0}-x_{i}\|=\mu_{i} . If not, suppose that \|v_{0}-x_{i}\|<\mu_{i} for all i=1,\cdots,N . Using the weakly strictly monotonicity of f and the continuity of f on bounded subsets of \mathbb{R}^{N}_{\geq 0} , we have \displaystyle f(\|v_{0}-x_{1}\|,\cdots,\|v_{0}-x_{N}\|) \displaystyle< \displaystyle f(\mu_{1},\cdots,\mu_{N}) \displaystyle= \displaystyle\lim_{k}f(\|v_{n_{k}}-x_{1}\|,\cdots,\|v_{n_{k}}-x_{N}\|) \displaystyle= \displaystyle\textrm{rad}_{V}^{f}(F). This contradiction establishes our claim. Also, as v_{n_{k}}-x_{i}\xrightarrow{w}v_{0}-x_{i} , by the Kadec-Klee property of X , we have v_{n_{k}}\rightarrow v_{0} . This establishes our assertion. ∎ Remark 2.7 . The authors in [ 19 , Example 2.9] provided an example of a reflexive Banach space X that does not have the Kadec-Klee property. Hence, from Theorems 2.3 and 2.6 , it follows that there exists a closed convex subset V of X such that (X,V,\mathcal{F}(X)) has weak- \mathscr{F}_{cwmc} - \bf{\emph{SACP}} but (X,V,\mathcal{F}(X)) does not have \mathscr{F}_{cwmc} - \bf{\emph{SACP}} . Our next result shows that for Banach spaces X with the Kadec-Klee property, for a closed convex subset V of X , weak- \mathscr{F}_{cwm} -SACP for a triplet (X,V,\mathcal{F}(X)) implies \mathscr{F}_{cwm} -SACP for (X,V,\mathcal{F}(X)) . Proposition 2.8 . For a Banach space X , the following are equivalent. (a) X has the Kadec-Klee property. (b) For V\in\mathcal{CC}(X) , if (X,V,\mathcal{F}(X)) has weak- \mathscr{F}_{cwm} - \bf{\emph{SACP}} , then (X,V,\mathcal{F}(X)) has \mathscr{F}_{cwm} - \bf{\emph{SACP}} . Proof. (a)\Rightarrow(b) . Let F\in\mathcal{F}(X) , f:\ell_{\infty}^{+}(F)\rightarrow\mathbb{R}_{\geq 0} be a function in \mathcal{F}_{cwm} and (v_{n})\subseteq V be such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) . Then, by our assumption, (v_{n}) has a subsequence (v_{n_{k}}) converging weakly to v_{0}\in V . Then, as in the proof of (a)\Rightarrow(b) of Theorem 2.6 , by passing on to a subsequence, if necessary, we can show that v_{n_{k}}\rightarrow v_{0} , and so, (a)\Rightarrow(b) . (b)\Rightarrow(a) Let V\in\mathcal{CC}(X) be such that (X,V,\mathcal{F}(X)) has weak- \mathscr{F}_{cwm} -SACP. Then V is weakly approximatively compact in X and hence, by our assumption, V is approximatively compact in X . Our assertion now follows from [ 14 , Page 396, Theorem 3] . ∎ Proceeding in the same way as in the proof of [ 17 , Theorem 3.4] , we obtain the following. Theorem 2.9 . Let X be a separable Banach space with the Namioka-Phelps property. Then, for any w^{*} -closed subspace Y of X^{*} , (X^{*},Y,\mathcal{F}(X^{*})) has \mathscr{F}_{cwmc} - \bf{\emph{SACP}} . 3. Various implications of smoothness and rotundity in relation to \tau - \mathscr{F} -SACP For a \tau -closed subset V of X , and a monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} we define the following. Notation . (1) \ell_{\gamma}^{N}(X) denotes the space X^{N} with norm |||\penalty\ .\penalty\ ||| defined by |||(x_{1},\cdots,x_{N})|||=\gamma(\|x_{1}\|,\cdots,\|x_{N}\|) . (2) \Delta_{\gamma}(V,N)=\{(v,\cdots,v)\in X^{N}:v\in V\} . We further define the following. Notation . (1) \mathscr{F}_{mn} denotes the class of all monotone norms \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} for all N\in\mathbb{N} . (2) \mathscr{F}_{msn} denotes the class of all monotone smooth norms \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} for all N\in\mathbb{N} . (3) \mathscr{F}_{mscn} denotes the class of all monotone strictly convex norms \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} for all N\in\mathbb{N} . For x\in X , a \tau -closed subset V of X , F=\{x_{1},\cdots,x_{N}\}\in\mathcal{F}(X) , and a monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , we have r_{\gamma}(x,F)=\gamma(\|x-x_{1}\|,\cdots,\|x-x_{N}\|)=|||(x,\cdots,x)-(x_{1},\cdots,x_{N})||| and \textrm{rad}_{V}^{\gamma}(F)=\underset{v\in V}{\inf}\gamma(\|v-x_{1}\|,\cdots,\|v-x_{N}\|)=\underset{v\in V}{\inf}|||(v,\cdots,v)-(x_{1},\cdots,x_{N})|||=d((x_{1},\cdots,x_{N}),\Delta_{\gamma}(V,N)) . Now, it is easy to observe that v\in\textrm{Cent}_{V}^{\gamma}(F) if and only if (v,\cdots,v)\in P_{\Delta_{\gamma}(V,N)}((x_{1},\cdots,x_{N})) . We identify the dual space \ell_{\gamma}^{N}(X)^{*} by \ell_{\gamma^{\circ}}^{N}(X^{*}) , with the norm |||\penalty\ .\penalty\ |||^{\circ} defined by |||(f_{1},\cdots,f_{N})|||^{\circ}=\gamma^{\circ}(\|f_{1}\|,\cdots,\|f_{N}\|) , where \gamma^{\circ} is the dual norm of \gamma . The pairing between \ell_{\gamma}^{N}(X) and \ell_{\gamma^{\circ}}^{N}(X^{*}) is given as, for f=(f_{1},\cdots,f_{N})\in\ell_{\gamma^{\circ}}^{N}(X^{*}) and x=(x_{1},\cdots,x_{N})\in\ell_{\gamma}^{N}(X) , f(x)=\sum_{i=1}^{N}f_{i}(x_{i}) . From the above discussion, we immediately have the following. Theorem 3.1 . Let V be a \tau -closed subset of X and \mathscr{G}\subseteq\mathscr{F}_{mn} . Then (X,V,\mathcal{F}(X)) has \tau - \mathscr{G} - \bf{\emph{SACP}} if and only if \Delta_{\gamma}(V,N) is approximatively \tau -compact in \ell_{\gamma}^{N}(X) for all monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} in \mathscr{G} for all N\in\mathbb{N} . Also for F\in\mathcal{F}(X) with card(F)=N , and a monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} in \mathscr{G} , \emph{Cent}_{V}^{\gamma}(F) is singleton if and only if P_{\Delta_{\gamma}(V,N)}((x_{1},\cdots,x_{N})) is singleton. Our next result shows that for every N\in\mathbb{N} and a monotone smooth norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , \ell_{\gamma}^{N}(X) is Fréchet smooth, when X is Fréchet smooth. Theorem 3.2 . Suppose X is Fréchet smooth. Then for every N\in\mathbb{N} and monotone smooth norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , \ell_{\gamma}^{N}(X) is Fréchet smooth. Proof. We prove the case for N=2 , since the proof for other values of N follows similarly. Let \gamma:\mathbb{R}^{2}\rightarrow\mathbb{R}_{\geq 0} be a monotone smooth norm, and \gamma^{\circ} be the dual norm of \gamma . Let (x_{1},x_{2})\in S_{\ell_{\gamma}^{2}(X)} , and ((f_{n}^{1},f_{n}^{2}))\subseteq S_{\ell_{\gamma^{\circ}}^{2}(X^{*})} be such that f_{n}^{1}(x_{1})+f_{n}^{2}(x_{2})\rightarrow 1 . Let (f_{1},f_{2})\in S_{\ell_{\gamma^{\circ}}^{2}(X^{*})} be such that f_{1}(x_{1})+f_{2}(x_{2})=1 . Then 1=f_{1}(x_{1})+f_{2}(x_{2})\leq\|f_{1}\|\|x_{1}\|+\|f_{2}\|\|x_{2}\|\leq\gamma^{\circ}(\|f_{1}\|,\|f_{2}\|)\gamma(\|x_{1}\|,\|x_{2}\|)=1 . Thus \|f_{1}\|\|x_{1}\|+\|f_{2}\|\|x_{2}\|=1 . Since \gamma is a smooth norm, J_{\mathbb{R}^{2}}((\|x_{1}\|,\|x_{2}\|))=\{(\|f_{1}\|,\|f_{2}\|)\} . Claim: (\|f_{n}^{1}\|),(\|f_{n}^{2}\|) converges to \|f_{1}\| and \|f_{2}\| respectively. Let (\|f_{n_{k}}^{1}\|) be a subsequence of (\|f_{n}^{1}\|) . Since (\|f_{n_{k}}^{1}\|) is bounded, it has a convergent subsequence, also denoted by (\|f_{n_{k}}^{1}\|) , converging to some \alpha_{1}\in\mathbb{R} . Similarly, (\|f_{n_{k}}^{2}\|) has a convergent subsequence, also denoted by (\|f_{n_{k}}^{2}\|) , converging to some \alpha_{2}\in\mathbb{R} . Now \alpha_{1}\|x_{1}\|+\alpha_{2}\|x_{2}\|=\underset{k}{\lim}(\|f_{n_{k}}^{1}\|\|x_{1}\|+\|f_{n_{k}}^{2}\|\|x_{2}\|)=1 . Since \gamma is a smooth norm on \mathbb{R}^{2} , we have \alpha_{1}=\|f_{1}\| and \alpha_{2}=\|f_{2}\| . Thus every subsequence of (\|f_{n}^{1}\|) has a further subsequence converging to \|f_{1}\| and hence (\|f_{n}^{1}\|) is convergent to \|f_{1}\| . Proceeding in the same way, we can show that (\|f_{n}^{2}\|) is convergent to \|f_{2}\| . Case 1: Suppose x_{1},x_{2}\neq 0 . Then \displaystyle 1=\underset{n}{\lim}(f_{n}^{1}(x_{1})+f_{n}^{2}(x_{2})) \displaystyle\leq \displaystyle\underset{n}{\lim\inf}(\|f_{n}^{1}\|\|x_{1}\|+\|f_{n}^{2}\|\|x_{2}\|) \displaystyle\leq \displaystyle\underset{n}{\lim\sup}(\|f_{n}^{1}\|\|x_{1}\|+\|f_{n}^{2}\|\|x_{2}\|) \displaystyle\leq \displaystyle\underset{n}{\lim\sup}\gamma^{\circ}(\|f_{n}^{1}\|,\|f_{n}^{2}\|)\gamma(\|x_{1}\|,\|x_{2}\|)=1. Hence \underset{n}{\lim}(f_{n}^{1}(x_{1})+f_{n}^{2}(x_{2}))=\underset{n}{\lim}(\|f_{n}^{1}\|\|x_{1}\|+\|f_{n}^{2}\|\|x_{2}\|)=1 . Now it is easy to see that \underset{n}{\lim}(\|f_{n}^{1}\|\|x_{1}\|-f_{n}^{1}(x_{1}))=0 and \underset{n}{\lim}(\|f_{n}^{2}\|\|x_{2}\|-f_{n}^{2}(x_{2}))=0 . Hence \frac{f_{n}^{1}}{\|f_{n}^{1}\|}(\frac{x_{1}}{\|x_{1}\|})\rightarrow 1 and \frac{f_{n}^{2}}{\|f_{n}^{2}\|}(\frac{x_{2}}{\|x_{2}\|})\rightarrow 1 . Now, due to the Fréchet differentiability of X , we have (\frac{f_{n}^{1}}{\|f_{n}^{1}\|}) and (\frac{f_{n}^{2}}{\|f_{n}^{2}\|}) are convergent. Hence (f_{n}^{1}) and (f_{n}^{2}) are convergent, and thus ((f_{n}^{1},f_{n}^{2})) is convergent. Case 2: Suppose x_{1}\neq 0 and x_{2}=0 . Then we have 1=\|f_{1}\|\|x_{1}\|\leq\gamma^{\circ}(\|f_{1}\|,0)\gamma(\|x_{1}\|,\|x_{2}\|)=\gamma^{\circ}(\|f_{1}\|,0)\leq\gamma^{\circ}(\|f_{1}\|,\|f_{2}\|)=1 . Thus \gamma^{\circ}(\|f_{1}\|,0)=1 and as J_{\mathbb{R}^{2}}((\|x_{1}\|,\|x_{2}\|))=\{(\|f_{1}\|,\|f_{2}\|)\} , we have f_{2}=0 . Hence (f_{n}^{2}) converges to 0. Now proceeding in a similar way, as in Case 1, we can show that (f_{n}^{1}) is convergent. Thus ((f_{n}^{1},f_{n}^{2})) is convergent. This concludes our assertion. ∎ We now characterize Banach spaces that are Fréchet smooth, based on triplets having \mathscr{F} -SACP. Theorem 3.3 . For a Banach space X , the following are equivalent. (a) X is Fréchet smooth. (b) For any w^{*} -closed convex subset V of X^{*} , (X^{*},V,\mathcal{F}(X^{*})) has \mathscr{F}_{msn} - \bf{\emph{SACP}} and for all F\in\mathcal{F}(X^{*}) and monotone smooth norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , where card(F)=N , \emph{Cent}_{V}^{\gamma}(F) is singleton. (c) Every w^{*} -closed convex subset of X^{*} is approximatively compact and Chebyshev in X^{*} . Proof. (a)\Rightarrow(c) Let V be a w^{*} -closed convex subset of X^{*} and f\in X^{*} . Without loss of generality, we may assume that f=0 and d(0,V)=1 . Now, as d(0,V)=1 , the open unit ball of X^{*} is disjoint from the convex set V , and so, by [ 13 , Theorem 2.2.26] , there exists x\in S_{X} such that f(x)\geq 1 for all f\in V . Let (f_{n})\subseteq V be such that \|f_{n}\|\rightarrow d(0,V)=1 . Now, 1\leq f_{n}(x)\leq\|f_{n}\|\rightarrow 1 and hence by the Fréchet differentiability of the norm in X at x , (\frac{f_{n}}{\|f_{n}\|}) is convergent and hence (f_{n}) is convergent. Since any minimizing sequence for f in V is convergent, V is Chebyshev in X^{*} . This concludes our assertion. (c)\Rightarrow(a) follows from [ 2 , Theorem 2.10] . We only need to prove (a)\Rightarrow(b) as (b)\Rightarrow(c) is obvious. (a)\Rightarrow(b) Let V be a w^{*} -closed convex subset of X^{*} , \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} be a monotone smooth norm and \gamma^{\circ} be the dual norm of \gamma . Since X is Fréchet smooth, by Theorem 3.2 , \ell_{\gamma}^{N}(X) is also Fréchet smooth. Now \Delta_{\gamma^{\circ}}(V,N) is a w^{*} -closed convex subset of \ell_{\gamma^{\circ}}^{N}(X^{*}) and hence from (a)\Rightarrow(c) , \Delta_{\gamma^{\circ}}(V,N) is approximatively compact and Chebyshev in \ell_{\gamma^{\circ}}^{N}(X^{*}) . Now, from Theorem 3.1 , our assertion follows. ∎ Proposition 3.4 . For an N\in\mathbb{N} and monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , if (f_{1},\cdots,f_{N})\in NA(\ell_{\gamma}^{N}(X)) , then \frac{f_{i}}{\|f_{i}\|}\in NA(X) for all i=1,\cdots,N such that f_{i}\neq 0 . Proof. We prove the result for N=2 as no new ideas are involved for higher values of N . Let \gamma:\mathbb{R}^{2}\rightarrow\mathbb{R}_{\geq 0} be a monotone norm and \gamma^{\circ} be the dual norm of \gamma . Let (f_{1},f_{2})\in NA(\ell_{\gamma}^{2}(X)) and (x_{1},x_{2})\in S_{\ell_{\gamma}^{2}(X)} be such that f_{1}(x_{1})+f_{2}(x_{2})=1 . Then \displaystyle 1=f_{1}(x_{1})+f_{2}(x_{2}) \displaystyle\leq \displaystyle\|f_{1}\|\|x_{1}\|+\|f_{2}\|\|x_{2}\| \displaystyle\leq \displaystyle\gamma^{\circ}(\|f_{1}\|,\|f_{2}\|)\gamma(\|x_{1}\|,\|x_{2}\|)=1. Thus \|f_{1}\|\|x_{1}\|+\|f_{2}\|\|x_{2}\|=f_{1}(x_{1})+f_{2}(x_{2}) . Since f_{i}(x_{i})\leq\|f_{i}\|\|x_{i}\| , we have f_{i}(\frac{x_{i}}{\|x_{i}\|})=\|f_{i}\| for i=1,2 . Thus if f_{i}\neq 0 , then \frac{f_{i}}{\|f_{i}\|}\in NA(X) for i=1,2 . ∎ Our next result demostrates the fact that for every N\in\mathbb{N} , and a strictly convex monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , \ell_{\gamma}^{N}(X) is a \tau-ALUR space, when X is a \tau-ALUR space. Theorem 3.5 . Let X be a \tau -ALUR space. Then for every N\in\mathbb{N} , and strictly convex monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , \ell_{\gamma}^{N}(X) is a \tau -ALUR space. Proof. We consider the case when N=2 because the proof for other values of N is similar. Let \gamma:\mathbb{R}^{2}\rightarrow\mathbb{R}_{\geq 0} be a strictly convex monotone norm and \gamma^{\circ} be the dual norm of \gamma . Let (f_{1},f_{2})\in NA(\ell_{\gamma}^{2}(X)) and ((x_{n},y_{n}))\subseteq S_{\ell_{\gamma}^{2}(X)} be such that f_{1}(x_{n})+f_{2}(y_{n})\rightarrow 1 . Let (z_{1},z_{2})\in S_{\ell_{\gamma}^{2}(X)} be such that f_{1}(z_{1})+f_{2}(z_{2})=1 . Then \displaystyle 1=f_{1}(z_{1})+f_{2}(z_{2}) \displaystyle\leq \displaystyle\|f_{1}\|\|z_{1}\|+\|f_{2}\|\|z_{2}\| \displaystyle\leq \displaystyle\gamma^{\circ}(\|f_{1}\|,\|f_{2}\|)\gamma(\|z_{1}\|,\|z_{2}\|)=1. Thus \|f_{1}\|\|z_{1}\|+\|f_{2}\|\|z_{2}\|=1 . Thus (\|f_{1}\|,\|f_{2}\|) attains its norm at (\|z_{1}\|,\|z_{2}\|) . Claim: (\|x_{n}\|),(\|y_{n}\|) converges to \|z_{1}\| and \|z_{2}\| respectively. Let (\|x_{n_{k}}\|) be a subsequence of (\|x_{n}\|) . Since (\|x_{n_{k}}\|) is bounded, (\|x_{n_{k}}\|) has a convergent subsequence, also denoted by (\|x_{n_{k}}\|) , converging to some \alpha\in\mathbb{R} . In a similar way, we can show that (\|y_{n_{k}}\|) has a convergent subsequence, also denoted by (\|y_{n_{k}}\|) , converging to some \beta\in\mathbb{R} . Since the norm function is continuous, \gamma(\alpha,\beta)=\underset{k}{\lim}\gamma(\|x_{n_{k}}\|,\|y_{n_{k}}\|)=1 . Now \|f_{1}\|\alpha+\|f_{2}\|\beta=\underset{k}{\lim}(\|f_{1}\|\|x_{n_{k}}\|+\|f_{2}\|\|y_{n_{k}}\|)=1 . Now by the strict convexity of the norm \gamma , we have \alpha=\|z_{1}\| and \beta=\|z_{2}\| . Thus (\|x_{n}\|) is convergent to \|z_{1}\| and (\|y_{n}\|) is convergent to \|z_{2}\| . Case 1: Suppose f_{1},f_{2}\neq 0 . Now \displaystyle 1=\underset{n}{\lim}(f_{1}(x_{n})+f_{2}(y_{n})) \displaystyle\leq \displaystyle\underset{n}{\lim\inf}(\|f_{1}\|\|x_{n}\|+\|f_{2}\|\|y_{n}\|) \displaystyle\leq \displaystyle\underset{n}{\lim\sup}(\|f_{1}\|\|x_{n}\|+\|f_{2}\|\|y_{n}\|) \displaystyle\leq \displaystyle\underset{n}{\lim\sup}\gamma^{\circ}(\|f_{1}\|,\|f_{2}\|)\gamma(\|x_{n}\|,\|y_{n}\|)\leq 1. Hence \underset{n}{\lim}(\|f_{1}\|\|x_{n}\|+\|f_{2}\|\|y_{n}\|)=\underset{n}{\lim}(f_{1}(x_{n})+f_{2}(y_{n}))=1 . Now it is easy to see that \underset{n}{\lim}(\|f_{1}\|\|x_{n}\|-f_{1}(x_{n}))=0 and \underset{n}{\lim}(\|f_{2}\|\|y_{n}\|-f_{2}(y_{n}))=0 . Thus \frac{f_{1}}{\|f_{1}\|}(\frac{x_{n}}{\|x_{n}\|})\rightarrow 1 and \frac{f_{2}}{\|f_{2}\|}(\frac{y_{n}}{\|y_{n}\|})\rightarrow 1 . From lemma 3.4 , \frac{f_{1}}{\|f_{1}\|},\frac{f_{2}}{\|f_{2}\|}\in NA(X) , and so, by our assumption, (\frac{x_{n}}{\|x_{n}\|}),(\frac{y_{n}}{\|y_{n}\|}) are \tau -convergent in X and hence ((x_{n},y_{n})) is \tau -convergent in \ell_{\gamma}^{2}(X) . Case 2: Suppose f_{1}\neq 0 and f_{2}=0 . Now 1=\|f_{1}\|\|x_{1}\|=\gamma^{\circ}(\|f_{1}\|,0)\gamma(\|x_{1}\|,0)\leq\gamma(\|x_{1}\|,\|x_{2}\|)=1 and hence \gamma(\|x_{1}\|,0)=1 . Since (\|f_{1}\|,0) attains its norm on (\|x_{1}\|,0) and \gamma is strictly convex, we have \|x_{2}\|=0 and thus x_{2}=0 . Hence (y_{n}) converges to 0. Also, in a similar way, as in Case 1, (x_{n}) is \tau -convergent. This concludes our assertion. ∎ We now characterize \tau-ALUR spaces, based on triplets having \tau - \mathscr{F} -SACP. Theorem 3.6 . The following are equivalent. (a) X is a \tau -ALUR space. (b) For every V\in\mathcal{CC}(X) , such that (X,V,\mathcal{F}(X)) has \mathscr{F}_{scmn} - \bf{\emph{rcp}} , then (X,V,\mathcal{F}(X)) has \tau - \mathscr{F}_{scmn} - \bf{\emph{SACP}} and for every F\in\mathcal{F}(X) and strictly convex monotone norm \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} , where card(F)=N , \emph{Cent}_{V}^{\gamma}(F) is singleton. (c) Every proximinal closed convex subset of X is approximatively compact and Chebyshev in X . Proof. (a)\Leftrightarrow(c) follows from [ 2 , Theorem 2.11] . We only need to prove (a)\Rightarrow(b) as (b)\Rightarrow(c) is obvious. (a)\Rightarrow(b) Let V\in\mathcal{CC}(X) be such that (X,V,\mathcal{F}(X)) has \mathscr{F}_{scmn} -rcp. Let F\in\mathcal{F}(X) , card(F)=N , and \gamma:\mathbb{R}^{N}\rightarrow\mathbb{R}_{\geq 0} be a strictly convex monotone norm. Then from Theorem 3.5 , \ell_{\gamma}^{N}(X) is a \tau-ALUR space. Now by our assumption, \Delta_{\gamma}(V,N) is proximinal in \ell_{\gamma}^{N}(X) and hence from (a)\Rightarrow(c) , \Delta_{\gamma}(V,N) is approximatively compact and Chebyshev in \ell_{\gamma}^{N}(X) . This concludes our assertion. ∎ When X is a \tau-CLUR space, we have the following. Theorem 3.7 . Let X be a \tau-CLUR space, V\in\mathcal{CC}(X) , \mathcal{G}=\{F\in\mathcal{F}(X):\emph{Cent}_{V}^{\rho}(F)\neq\emptyset\penalty\ \textrm{for\penalty\ all}\penalty\ \rho=(\rho_{1},\cdots,\rho_{card(F)})\} . Then, (X,V,\mathcal{G}) has weighted \tau - \bf{\emph{SACP}} . Proof. Let F=\{x_{1},\cdots,x_{N}\}\in\mathcal{G} and \rho=(\rho_{1},\cdots,\rho_{N}) be positive weights. Let (v_{n})\subseteq V be such that r_{\rho}(v_{n},F)\rightarrow\textrm{rad}_{V}^{\rho}(F) . Then, it is easy to observe that (v_{n}) is bounded. Let v_{0}\in\textrm{Cent}_{V}^{\rho}(F) and w_{n}=\frac{v_{n}+v_{0}}{2} for all n\in\mathbb{N} . By passing on to a subsequence, if necessary, let r_{\rho}(w_{n},F)=\rho_{n}\|w_{n}-x_{n}\| , where \rho_{n}\in(\rho_{1},\cdots,\rho_{N}) and x_{n}\in\{x_{1},\cdots,x_{N}\} for all n\in\mathbb{N} . Then, there exists a subsequence (x_{n_{k}}) of (x_{n}) such that x_{n_{k}}=x_{j} for all k\in\mathbb{N} and for some j\in\{1,\cdots,N\} . Thus, r_{\rho}(w_{n_{k}},F)=\rho_{j}\|w_{n_{k}}-x_{j}\| for all k\in\mathbb{N} . Let u_{0}=\rho_{j}(v_{0}-x_{j}) and u_{k}=\rho_{j}(v_{n_{k}}-x_{j}) for all k\in\mathbb{N} . Then it is easy to observe that \|u_{0}\|\leq\textrm{rad}_{V}^{\rho}(F) and \underset{k}{\lim}\|u_{k}\|\leq\textrm{rad}_{V}^{\rho}(F) . Now, for all k\in\mathbb{N} , \|\frac{u_{k}+u_{0}}{2}\|=\rho_{j}\|w_{n_{k}}-x_{j}\|=r_{\rho}(w_{n_{k}},F)\geq\textrm{rad}_{V}^{\rho}(F). Since X is \tau-CLUR , (u_{k}) has a \tau -convergent subsequence, and so (v_{n}) has a \tau -convergent subsequence. Hence, our conclusion follows. ∎ Theorem 3.8 . Let X be a Banach space, which is an L -summand in its bidual, and Y be a subspace of X , which is also an L -summand in its bidual. Then, (X,Y,\mathcal{F}(X)) has weak- \mathscr{F}_{wcmc} - \bf{\emph{SACP}} . Proof. Let F=\{x_{1},\cdots,x_{N}\}\in\mathcal{F}(X) , f:\mathbb{R}^{N}_{\geq 0}\rightarrow\mathbb{R}_{\geq 0} be a function in \mathscr{F}_{wcmc} and (y_{n})\subseteq Y be such that r_{f}(y_{n},F)\rightarrow\textrm{rad}_{Y}^{f}(F) . Let P:X^{**}\rightarrow X be the L -projecton. Then, from [ 10 , Theorem 1.2] , we have P(\overline{Y}^{w^{*}})=Y . Since f is coercive, (y_{n}) is bounded. Let \Lambda be a w^{*} -accumulation point of (y_{n}) . Then, as in the proof of [ 5 , Theorem 3.5(a)] , we have \textrm{rad}_{Y}^{f}(F)=\textrm{rad}_{\overline{Y}^{w^{*}}}^{f}(F) . Since by [ 17 , Thorem 1.4(iv)] , r_{f}(.,F):X^{**}\rightarrow\mathbb{R}_{\geq 0} is w^{*} -lower semicontinuous, we have \textrm{rad}_{\overline{Y}^{w^{*}}}^{f}(F)\leq r_{f}(P(\Lambda),F)\leq r_{f}(\Lambda,F)\leq\textrm{rad}_{Y}^{f}(F). Thus, r_{f}(\Lambda,F)=r_{f}(P(\Lambda),F) . Now, if \Lambda\neq P(\Lambda) , then for all i=1,\cdots,N , we have \|\Lambda-x_{i}\|=\|P(\Lambda)-x_{i}\|+\|\Lambda-P(\Lambda)\|>\|P(\Lambda)-x_{i}\|. Since f is weakly strictly monotone, we have r_{f}(P(\Lambda),F)0 , there exists \delta>0 such that diam(\delta-\emph{Cent}_{V}^{f}(F))<\varepsilon . Proof. Suppose for all (v_{n})\subseteq V such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) implies that (v_{n}) is convergent. Then, it is easy to observe that \textrm{Cent}_{V}^{f}(F) is singleton. Let \textrm{Cent}_{V}^{f}(F)=\{v_{0}\} . Suppose that there exists \varepsilon>0 such that for all n\in\mathbb{N} , v_{n}\in\frac{1}{n}-\textrm{Cent}_{V}^{f}(F) but \|v_{n}-v_{0}\|\geq\varepsilon . Then r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) but (v_{n}) does not converge to v_{0} , which is a contraction. Conversely, assume that, for all \varepsilon>0 , there exists \delta>0 such that diam(\delta-\textrm{Cent}_{V}^{f}(F))<\varepsilon . Then, by the Cantor Intersection Theorem, \textrm{Cent}_{V}^{f}(F) is singleton \{v_{0}\} (say). Let (v_{n})\subseteq V be such that r_{f}(v_{n},F)\rightarrow\textrm{rad}_{V}^{f}(F) . Then, there exists N\in\mathbb{N} , v_{n}\in\delta-\textrm{Cent}_{V}^{f}(F) for all n\geq N . Thus, \|v_{n}-v_{0}\|\leq diam(\delta-\textrm{Cent}_{V}^{f}(F))<\varepsilon for all n\geq N . Thus, v_{n}\rightarrow v_{0} and this establishes our assertion. ∎ 4. Continuity properties of some set-valued maps in relation to \tau - \mathscr{F} -SACP Throughout this section, by B[x,r] we denote the closed ball centered at x and radius r . In the following propositions, we present some basic observations. Proposition 4.1 . Let F=\{x_{1},\cdots,x_{N}\}\in\mathcal{F}(X) and f:\mathbb{R}^{N}_{\geq 0}\rightarrow\mathbb{R}_{\geq 0} be a convex monotone function. Then, the map r_{f}(.,F):X\rightarrow\mathbb{R}_{\geq 0} is Lipschitz continuous on bounded subsets of X . Proposition 4.2 . Let Y be a subspace of X , F=\{x_{1},\cdots,x_{N}\}\in\mathcal{F}(X) , f:\mathbb{R}^{N}_{\geq 0}\rightarrow\mathbb{R}_{\geq 0} be a monotone function, and \varepsilon>0 . Then, (a) \emph{rad}_{Y}^{f}(F)=\inf\{r_{f}(y,F):y\in\cup_{i=1}^{N}B[x_{i},\emph{rad}_{Y}(F)+\varepsilon]\cap Y\}=\inf\{r_{f}(y,F):y\in Y\penalty\ \textrm{and}\penalty\ \|y\|\leq\underset{1\leq i\leq N}{\max}\|x_{i}\|+\emph{rad}_{Y}(F)+\varepsilon\} . (b) \emph{Cent}_{Y}^{f}(F)\subseteq\cup_{i=1}^{N}B[x_{i},\emph{rad}_{Y}(F)+\varepsilon]\cap Y if f is weakly strictly monotone. (c) \emph{Cent}_{Y}^{f}(F)\subseteq\cap_{i=1}^{N}B[x_{i},\emph{rad}_{Y}(F)+diam(F)+\varepsilon]\cap Y if f is weakly strictly monotone. Proof. (a) . Let y\in Y\setminus(\cup_{i=1}^{N}B[x_{i},\textrm{rad}_{Y}(F)+\varepsilon]\cap Y) . Then, \|y-x_{i}\|>\textrm{rad}_{Y}(F)+\varepsilon for all i=1,\cdots,N . We select y_{0}\in Y such that \underset{1\leq i\leq N}{\max}\|y_{0}-x_{i}\|\leq\textrm{rad}_{Y}(F)+\varepsilon . Then, we have \|y_{0}-x_{i}\|\leq\textrm{rad}_{Y}(F)+\varepsilon<\|y-x_{i}\| for all i=1,\cdots,N . As f is monotone, r_{f}(y_{0},F)\leq r_{f}(y,F) , and hence this establishes our first equality in (a) . Let y\in\cup_{i=1}^{N}B[x_{i},\textrm{rad}_{Y}(F)+\varepsilon]\cap Y . Then, there exists j\in\{1,\cdots,N\} such that \|y-x_{j}\|\leq\textrm{rad}_{Y}(F)+\varepsilon . Thus, we have \|y\|\leq\|y-x_{j}\|+\|x_{j}\|\leq\textrm{rad}_{Y}(F)+\underset{1\leq i\leq N}{\max}\|x_{i}\|+\varepsilon. This establishes our second equality in (a) . (b) . Let y\in Y\setminus(\cup_{i=1}^{N}B[x_{i},\textrm{rad}_{Y}(F)+\varepsilon]\cap Y) . Then, as shown in the proof of (a) , we obtain a y_{0}\in Y such that \|y_{0}-x_{i}\|\leq\textrm{rad}_{Y}(F)+\varepsilon<\|y-x_{i}\| for all i=1,\cdots,N . As f is weakly strictly monotone, we have r_{f}(y_{0},F)0 such that d(F,F^{\prime})\leq M for all F,F^{\prime}\in B . Fix a set G=\{z_{1},\cdots z_{N}\}\in B . Let R>0 be such that \underset{1\leq i\leq N}{\max}\|z_{i}\|\leq R . Claim: \textrm{rad}_{Y}^{f} is Lipschitz continuous on B . Let F=\{x_{1},\cdots,x_{N}\},F^{\prime}=\{x_{1}^{\prime},\cdots,x_{N}^{\prime}\}\in B , Now, \underset{1\leq i\leq N}{\max}\|x_{i}\|+\textrm{rad}_{Y}(F)\leq 2\underset{1\leq i\leq N}{\max}\|x_{i}\|\leq 2\underset{1\leq i\leq N}{\max}\|x_{i}-z_{i}\|+2\underset{1\leq i\leq N}{\max}\|z_{i}\|\leq 2M+2R , and from proposition 4.2 (a) , we have \textrm{rad}_{Y}^{f}(F)=\inf\{r_{f}(y,F):\|y\|\leq 2M+2R+\varepsilon\} for an \varepsilon>0 . Thus, for any H=\{s_{1},\cdots,s_{N}\}\in B and any y\in Y such that \|y\|\leq 2M+2R+\varepsilon , we have \|y-s_{i}\|\leq 3M+3R+\varepsilon for all i=1,\cdots,N . For \varepsilon>0 , let L be the Lipschitz constant of f on [0,3M+3R+\varepsilon]^{N} . Then, for y\in Y with \|y\|\leq 2M+2R+\varepsilon , we have (1) \displaystyle\begin{split}|f(\|y-x_{1}\|,\cdots,\|y-x_{N}\|)-f(\|y-x_{1}^{\prime}\|,\cdots,\|y-x_{N}^{\prime}\|)|\\ \leq L\underset{1\leq i\leq N}{\max}|\|y-x_{i}\|-\|y-x_{i}^{\prime}\||\leq L\underset{1\leq i\leq N}{\max}\|x_{i}-x_{i}^{\prime}\|.\end{split} Thus, \displaystyle\textrm{rad}_{Y}^{f}(F) \displaystyle\leq \displaystyle f(\|y-x_{1}\|,\cdots,\|y-x_{N}\|) \displaystyle\leq \displaystyle f(\|y-x_{1}^{\prime}\|,\cdots,\|y-x_{N}^{\prime}\|)+L\underset{1\leq i\leq N}{\max}\|x_{i}-x_{i}^{\prime}\|. Hence \textrm{rad}_{Y}^{f}(F)\leq\textrm{rad}_{Y}^{f}(F^{\prime})+L\underset{1\leq i\leq N}{\max}\|x_{i}-x_{i}^{\prime}\| . Similarly, by interchanging F and F^{\prime} in ( 1 ), we have |\textrm{rad}_{Y}^{f}(F)-\textrm{rad}_{Y}^{f}(F^{\prime})|\leq L\underset{1\leq i\leq N}{\max}\|x_{i}-x_{i}^{\prime}\|=Ld(F,F^{\prime}) . Hence, our assertion follows. ∎ We now recall the notion of upper semicontinuity for set-valued maps. Definition 4.4 . Let T be a topological space and F:T\rightarrow\mathcal{CB}(X) be a set-valued map. Then, F is said to be upper semicontinuous if, for any t_{0}\in T and any open set W containing F(t_{0}) , there exists an open set U containing t_{0} such that F(t)\subseteq W for all t\in U . Theorem 4.5 . Let Y be a subspace of X and (X,Y,\mathcal{F}(X)) has \tau - \mathscr{F}_{cwm} - \bf{\emph{SACP}} . Then, for N\in\mathbb{N} and f:\mathbb{R}^{N}_{\geq 0}\rightarrow\mathbb{R}_{\geq 0} in \mathscr{F}_{cwm} , the map \emph{Cent}_{Y}^{f}:(\mathcal{F}_{N}(X),d)\rightarrow\mathcal{CB}(Y) is metric- \tau upper semicontinuous. Proof. Suppose that \textrm{Cent}_{Y}^{f} is not metric- \tau upper semicontinuous at F=\{x_{1},\cdots,x_{N}\}\in\mathcal{F}_{N}(X) . Then, there exists a \tau -open set W containing \textrm{Cent}_{Y}^{f}(F) and a sequence (F_{n})\subseteq\mathcal{F}_{N}(X) such that F_{n}\rightarrow F but \textrm{Cent}_{Y}^{f}(F_{n})\setminus W\neq\emptyset for all n\in\mathbb{N} . Let F_{n}=\{x_{1}^{n},\cdots,x_{N}^{n}\} for all n\in\mathbb{N} . Let r=\textrm{rad}_{Y}^{f}(F),r_{n}=\textrm{rad}_{Y}^{f}(F_{n}),\varphi(x)=r_{f}(x,F) and \varphi_{n}(x)=r_{f}(x,F_{n}) for all x\in X and n\in\mathbb{N} . Let us choose M>0 such that for all n\in\mathbb{N} , \underset{1\leq i\leq N}{\max}\|x_{i}^{n}\|\leq M and \underset{1\leq i\leq N}{\max}\|x_{i}\|\leq M . Let y_{n}\in\textrm{Cent}_{Y}^{f}(F_{n})\setminus W for all n\in\mathbb{N} and \varepsilon>0 . Then, from proposition 4.2 (c) , we have \|y_{n}-x_{i}^{n}\|\leq\textrm{rad}_{Y}(F_{n})+diam(F_{n})+\varepsilon\leq 3M+\varepsilon , and hence, \|y_{n}-x_{i}\|\leq\|y_{n}-x_{i}^{n}\|+\|x_{i}^{n}\|+\|x_{i}\|\leq 5M+\varepsilon for all i=1,\cdots,N . Let L be the Lipschitz constant for f on [0,5M+\varepsilon]^{N} and that for \textrm{rad}_{Y}^{f} on \{\{y_{1},\cdots,y_{N}\}\in\mathcal{F}_{N}(X):\underset{1\leq i\leq N}{\max}\|y_{i}\|\leq M\} . Then, we have r\leq\varphi(y_{n})=r+(r_{n}-r)+(\varphi(y_{n})-\varphi_{n}(y_{n}))\leq r+2L\sum_{i=1}^{N}\|x_{i}^{n}-x_{i}\|. Thus, \varphi(y_{n})\rightarrow r . By our assumption, (y_{n}) has a \tau -convergent subsequence (y_{n_{k}}) converging to y\in Y in \tau -topology. Clearly, y\in\textrm{Cent}_{Y}^{f}(F)\subseteq W but W is a \tau -open set, and y_{n}\notin W for all n\in\mathbb{N} . This contradiction proves our assertion. ∎ Remark 4.6 . It has been observed in [ 6 ] that there exists a one dimensional subspace Y of a finite dimensional Banach space X such that the metric projection x\mapsto P_{Y}(x) is not continuous. Since Y is finite dimensional, (X,Y,\mathcal{F}(X)) has \mathscr{F}_{cmc} - \bf{\emph{SACP}} . However, the map x\mapsto\emph{Cent}_{Y}(\{x\})=P_{Y}(x) is not continuous.