[1] Alegre, C.Ferrer, J. andGregori, V.,Quasi-uniformities on real vector spaces,Indian J. Pure Appl. Math., 28, no. 7, pp. 929–937, 1997.
[2]___,On the Hahn-Banach theorem in certain linear quasi-uniform structures, ActaMath. Hungar.,82, no. 4, pp. 325–330, 1999.
[3] Alimov, A. R.,The Banach-Mazur theorem for spaces with nonsymmetric distance,Uspekhi Mat. Nauk, 58, no. 2, pp. 159–160, 2003.
[4] Babenko, V. F.,Nonsymmetric approximations in spaces of summable functions,Ukrain. Mat. Zh., 34, no. 4, pp. 409–416, 538, 1982.
[5]___,Nonsymmetric extremal problems of approximation theory, Dokl. Akad. NaukSSSR,269, no. 3, pp. 521–524, 1983.
[6]___,Duality theorems for certain problems of the theory of approximation, Currentproblems in real and complex analysis, Akad. Nauk Ukrain. SSR Inst. Mat., Kiev,pp. 3–13, 148, 1984.
[7] Borodin, P. A.,The Banach-Mazur theorem for spaces with an asymmetric norm andits applications in convex analysis, Mat. Zametki,69, no. 3, pp. 329–337, 2001.
[8] Cobzas, S.,Phelps type duality results in best approximation, Rev. Anal. Numer.Theor. Approx.,31, no. 1, pp. 29–43, 2002.
[9]___,Separation of convex sets and best approximation in spaces with asymmetricnorm, Quaest. Math.,27, no. 3, 275–296, 2004.
[10]___,Asymmetric locally convex spaces,Int. J. Math. Math. Sci., no. 16, 2585–2608,2005.
[11] Cobzas, S. and Mustata, C.,Extension of bilinear functionals and best approximationin 2-normed spaces, Studia Univ. Babes-Bolyai, Mathematica, 43, pp. 1–13, 1998.
[12]___,Extension of bounded linear functionals and best approximation in spaces withasymmetric norm, Rev. Anal. Numer. Theor. Approx.,33, no. 1, pp. 39–50, 2004.
[13] De Blasi, F. S. and Myjak, J., On a generalized best approximation problem, J. Approx.Theory,94, no. 1, pp. 54–72, 1998.
[14] Dolzhenko, E. P. andSevastyanov, E. A.,Approximations with a sign-sensitiveweight (existence and uniqueness theorems), Izv. Ross. Akad. Nauk Ser. Mat.,62, no. 6,pp. 59–102,1998.
[15]___,Sign-sensitive approximations, J. Math. Sci. (New York),91, no. 5, pp. 3205–3257, 1998.
[16]___,Approximation with a sign-sensitive weight (stability, applications to snake the-ory and Hausdorff approximations), Izv. Ross. Akad. Nauk Ser. Mat.,63, no. 3, pp.77–118, 1999.
[17] Ferrer, J.,Gregori, V. andAlegre, C.,Quasi-uniform structures in linear lattices,Rocky Mountain J. Math.,23, no. 3, pp. 877–884, 1993.
[18] Garcıa-Raffi, L. M.,Romaguera, S. and Sanchez Perez, E. A.,Extensions of asymmetric norms to linear spaces, Rend. Istit. Mat. Univ. Trieste,33, nos. 1–2, 113–125, 2001.
[19]___,The bicompletion of an asymmetric normed linear space, Acta Math. Hungar.,97, no. 3, pp. 183–191, 2002.
[20]___,Sequence spaces and asymmetric norms in the theory of computational com-plexity, Math. Comput. Modelling,36, nos. 1–2, pp. 1–11, 2002.
[21]___,The dual space of an asymmetric normed linear space, Quaest. Math.,26, no. 1,pp. 83–96, 2003.
[22]___,On Hausdorff asymmetric normed linear spaces, Houston J. Math.,29, no. 3,pp. 717–728 (electronic), 2003.
[23] Krein, M. G. and Nudelman, A. A., The Markov Moment Problem and ExtremumProblems, Nauka, Moscow, 1973 (in Russian), English translation: American Mathe-matical Society, Providence, R.I., 1977.
[24] Chong Li, On wel l posed generalized best approximation problems, J. Approx. Theory,107, no. 1, pp. 96–108, 2000.
[25] Chong Liand Renxing Ni, Derivatives of generalized distance functions and existenceof generalized nearest points, J. Approx. Theory,115, no. 1, pp. 44–55, 2002.
[26] Mohebi, H.,On quasi-Chebyshev subspaces of Banach spaces, J. Approx. Theory,107,no. 1, pp. 87–95, 2000.
[27]___,Pseudo-Chebyshev subspaces inL1, Korean J. Comput. Appl. Math.,7, no. 2,pp. 465–475, 2000.
[28]___,On pseudo-Chebyshev subspaces in normed linear spaces, Math. Sci. Res. Hot-Line,5, no. 9, pp. 29–45, 2001.
[29]___,Quasi-Chebyshev subspaces in dual spaces, J. Nat. Geom.,20, nos. 1-2, pp.33–44, 2001.
[30]___,On pseudo-Chebyshev subspaces in normed linear spaces, J. Nat. Geom.,24,nos. 1–2, pp. 37–56, 2003.
[31] Mohebi, H. andRezapour,Sh.,On weak compactness of the set of extensions of acontinuous linear functional, J. Nat. Geom.,22, nos. 1–2, pp. 91–102, 2002.
[32] Mustata, C.,Extensions of semi-Lipschitz functions on quasi-metric spaces, Rev.Anal. Numer. Theor. Approx.,30, no. 1, pp. 61–67, 2001.
[33]___,On the extremal semi-Lipschitz functions, Rev. Anal. Numer. Theor. Approx.,31, no. 1, pp. 103–108, 2002.
[34]___,A Phelps type theorem for spaces with asymmetric norms, Bul. Stiint. Univ.Baia Mare, Ser. B, Matematica-Informatica,18, no. 2, pp. 275–280, 2002.
[35]___,On the uniqueness of the extension and unique best approximation in the dual ofan asymmetric linear space, Rev. Anal. Numer. Theor. Approx.,32, no. 2, pp. 187–192,2003.
[36] Renxing Ni,Existence of generalized nearest points, Taiwanese J. Math.,7, no. 1, pp.115–128, 2003.
[37] Phelps, R. R.,Uniqueness of Hahn-Banach extensions and best approximations, Trans.Amer. Marth Soc.,95, pp. 238–255, 1960.
[38] Rezapour, Sh.,∈-weakly Chebyshev subspaces of Banach spaces, Anal. Theory Appl.,19, no. 2, pp. 130–135, 2003.
[39] Romaguera, S. and Sanchis, M.,Semi-Lipschitz functions and best approximation inquasi-metric spaces, J. Approx. Theory,103, no. 2, pp. 292–301, 2000.
[40] Romaguera, S. and Schellekens, M.,Duality and quasi-normability for complexityspaces, Appl. Gen. Topol.,3, no. 1, pp. 91–112, 2002.
[41] Simonov, B. V.,On the element of best approximation in spaces with nonsymmetricquasinorm, Mat. Zametki,74, no. 6, pp. 902–912, 2003.
[42] Singer, I.,Best Approximation in Normed Linear Spaces by Elements of Linear Sub-spaces, Editura Academiei Romane and Springer-Verlag, Bucharest-New York-Berlin,1970.
[43] Zanco, C. and Zucchi, A.,Moduli of rotundity and smoothness for convex bodies,Bolletino U. M. I., (7),7-B, pp. 833–855, 1993
Paper (preprint) in HTML form
2006-Mustata-Best approximation in spaces with asymmetric norm-Jnaat
BEST APPROXIMATION IN SPACES WITH ASYMMETRIC NORM
S. COBZAŞ* and C. MUSTĂŢA ^(†){ }^{\dagger}
Abstract
In this paper we shall present some results on spaces with asymmetric seminorms, with emphasis on best approximation problems in such spaces.
MSC 2000. 41A65.
Keywords. Spaces with asymmetric norm, best approximation, Hahn-Banach theorem, characterization of best approximation.
1. INTRODUCTION
Let XX be a real vector space. An asymmetric seminorm on XX is a positive sublinear functional p:X rarr[0,oo)p: X \rightarrow[0, \infty), i.e. pp satisfies the conditions:
(AN1) p(x) >= 0p(x) \geq 0;
(AN2) p(tx)=tp(x)p(t x)=t p(x);
(AN3) p(x+y) <= p(x)+p(y)p(x+y) \leq p(x)+p(y),
for all x,y in Xx, y \in X and t >= 0t \geq 0.
The function bar(p):X rarr[0,oo)\bar{p}: X \rightarrow[0, \infty) defined by bar(p)(x)=p(-x),x in X\bar{p}(x)=p(-x), x \in X, is another positive sublinear functional on XX, called the conjugate of pp, and
{:(1.1)p^(s)(x)=max{p(x)","p(-x)}","x in X:}\begin{equation*}
p^{s}(x)=\max \{p(x), p(-x)\}, x \in X \tag{1.1}
\end{equation*}
hold for all x,y in Xx, y \in X. If the seminorm p^(s)p^{s} is a norm on XX then we say that pp is an asymmetric norm on XX. This means that, beside (AN1)-(AN3), it satisfies also the condition
(AN4) p(x)=0p(x)=0 and p(-x)=0p(-x)=0 imply x=0x=0.
The pair ( X,pX, p ), where XX is a linear space and pp is an asymmetric seminorm on XX is called a space with asymmetric seminorm, respectively a space with asymmetric norm, if pp is an asymmetric norm.
The function rho:X xx X rarr[0;oo)\rho: X \times X \rightarrow[0 ; \infty) defined by rho(x,y)=p(y-x),x,y in X\rho(x, y)=p(y-x), x, y \in X, is an asymmetric semimetric on XX. Denote by B_(p)^(')(x,r)={x^(')in X:p(x^(')-x) < r}B_{p}^{\prime}(x, r)=\left\{x^{\prime} \in X: p\left(x^{\prime}-x\right)<r\right\} and B_(p)(x,r)={x^(')in X:p(x^(')-x) <= r}B_{p}(x, r)=\left\{x^{\prime} \in X: p\left(x^{\prime}-x\right) \leq r\right\}, the open, respectively closed, ball in XX of center xx and radius r > 0r>0. Denoting by
B_(p)^(')=B_(p)^(')(0,1)" and "B_(p)=B_(p)(0,1),B_{p}^{\prime}=B_{p}^{\prime}(0,1) \text { and } B_{p}=B_{p}(0,1),
the corresponding unit balls then
B_(p)^(')(x,r)=x+rB_(p)^(')" and "B_(p)(x,r)=x+rB_(p)B_{p}^{\prime}(x, r)=x+r B_{p}^{\prime} \text { and } B_{p}(x, r)=x+r B_{p}
The unit balls B_(p)^(')B_{p}^{\prime} and B_(p)B_{p} are convex absorbing subsets of the space XX and pp agrees with the Minkowski functional associated to any of them. Recall that for an absorbing subset CC of XX the Minkowski functional p_(C):X rarr[0;oo)p_{C}: X \rightarrow[0 ; \infty) is defined by
p_(C)(x)=i n f{t > 0:x in tC}p_{C}(x)=\inf \{t>0: x \in t C\}
If CC is absorbing and convex, then p_(C)p_{C} is a positive sublinear functional, and
{x in X:p_(C)(x) < 1}sub C sub{x in X:p_(C)(x) <= 1}\left\{x \in X: p_{C}(x)<1\right\} \subset C \subset\left\{x \in X: p_{C}(x) \leq 1\right\}
An asymmetric seminorm pp generates a topology tau_(p)\tau_{p} on XX, having as basis of neighborhoods of a point x in Xx \in X the family {B_(p)^(')(x,r):r > 0}\left\{B_{p}^{\prime}(x, r): r>0\right\} of open pp-balls. The family {B_(p)(x,r):r > 0}\left\{B_{p}(x, r): r>0\right\} of closed pp-balls is also a neighborhood basis at xx for tau_(p)\tau_{p}.
The topology tau_(p)\tau_{p} is translation invariant, i.e. the addition +:X xx X rarr X+: X \times X \rightarrow X is continuous, but the multiplication by scalars *:Rxx X rarr X\cdot: \mathbb{R} \times X \rightarrow X need not be continuous, as it is shown by some examples (see [7]).
The ball B_(p)^(')(x,r)B_{p}^{\prime}(x, r) is tau_(p)\tau_{p}-open but the ball B_(p)(x,r)B_{p}(x, r) need not to be tau_(p)\tau_{p}-closed, as can be seen from the following typical example.
Example 1.1. Consider on R\mathbb{R} the asymmetric seminorm u(alpha)=max{alpha,0}u(\alpha)=\max \{\alpha, 0\}, alpha inR\alpha \in \mathbb{R}, and denote by R_(u)\mathbb{R}_{u} the space R\mathbb{R} equipped with the topology tau_(u)\tau_{u} generated by uu. The conjugate seminorm is bar(u)(alpha)=-min{alpha,0}\bar{u}(\alpha)=-\min \{\alpha, 0\}, and u^(s)(alpha)=max{u(alpha), bar(u)(alpha)}=|alpha|u^{s}(\alpha)= \max \{u(\alpha), \bar{u}(\alpha)\}=|\alpha|. The topology tau_(u)\tau_{u}, called the upper topology of R\mathbb{R}, is generated by the intervals of the form (-oo;a),a inR(-\infty ; a), a \in \mathbb{R}, and the family {(-oo;alpha+epsilon):epsilon > 0}\{(-\infty ; \alpha+\epsilon): \epsilon>0\} is a neighborhood basis of every point alpha inR\alpha \in \mathbb{R}. The set (-oo;1)=B_(u)^(')(0,1)(-\infty ; 1)=B_{u}^{\prime}(0,1) is tau_(u)\tau_{u}-open, and the ball B_(u)(0,1)=(-oo;1]B_{u}(0,1)=(-\infty ; 1] is not tau_(u)\tau_{u}-closed because R\\B_(u)(0,1)=(1;oo)\mathbb{R} \backslash B_{u}(0,1)=(1 ; \infty) is not tau_(u)\tau_{u}-open.
The topology tau_(p)\tau_{p} could not be Hausdorff even if pp is an asymmetric norm on XX. A necessary and sufficient condition in order that tau_(p)\tau_{p} be Hausdorff was given in [22]. Putting
{:(1.3) tilde(p)(x)=i n f{p(x^('))+p(x^(')-x):x^(')in X}","x in X",":}\begin{equation*}
\tilde{p}(x)=\inf \left\{p\left(x^{\prime}\right)+p\left(x^{\prime}-x\right): x^{\prime} \in X\right\}, x \in X, \tag{1.3}
\end{equation*}
it follows that tilde(p)\tilde{p} is the greatest (symmetric) seminorm majorized by pp and the topology tau_(p)\tau_{p} is Hausdorff if and only if tilde(p)(x) > 0\tilde{p}(x)>0 for every x!=0x \neq 0. Changing xx
to -x-x and taking x^(')=0x^{\prime}=0 it follows that, in this case, p(x) > 0p(x)>0 for every x!=0x \neq 0, but this condition is not sufficient for tau_(p)\tau_{p} to be Hausdorff, see [22].
Spaces with asymmetric seminorms were investigated in a series of papers, emphasizing similarities with seminormed spaces as well as differences, see [1, 2, 3, 7, 17, 18, 19, 21, 22, and the references quoted therein. Among the differences we mention the fact that the dual of a space with asymmetric seminorm is not a linear space but merely a convex cone in the algebraic dual X^(#)X^{\#} of XX. This is due to the fact that the continuity of a linear functional varphi\varphi on ( X,pX, p ) does not imply the continuity of -varphi-\varphi. For instance, varphi(u)=u\varphi(u)=u is continuous on ( R,u\mathbb{R}, u ) but psi(u)=-u\psi(u)=-u is not continuous. For an other example see [7]. The study of spaces with asymmetric norm was motivated and stimulated also by their applications in the complexity of algorithms, see [20, 40].
Some continuity properties of linear functionals in the symmetric case have their analogs in the asymmetric one.
Proposition 1.2. [21] Let ( X,pX, p ) be a space with asymmetric seminorm and varphi:X rarrR\varphi: X \rightarrow \mathbb{R} a linear functional. Then the following are equivalent.
(1) varphi\varphi is tau_(p)-tau_(u)\tau_{p}-\tau_{u}-continuous at 0in X0 \in X.
(2) varphi\varphi is tau_(p)-tau_(u)\tau_{p}-\tau_{u}-continuous on XX.
(3) There exists L >= 0L \geq 0 such that
{:(1.4)AA x in X","quad varphi(x) <= Lp(x).:}\begin{equation*}
\forall x \in X, \quad \varphi(x) \leq L p(x) . \tag{1.4}
\end{equation*}
(4) varphi\varphi is upper semi-continuous from (X,tau_(p))\left(X, \tau_{p}\right) to (R,||)(\mathbb{R},| |).
A linear functional satisfying (1.4) is called semi-Lipschitz (or p-bounded) and LL a semi-Lipschitz constant. Denote by X_(p)^(b)X_{p}^{b} the set of all bounded linear functionals on the space with asymmetric seminorm ( X,pX, p ). As we did mention, X_(p)^(b)X_{p}^{b} is a convex cone in X^(#)X^{\#}.
One can define a norm |||_(p)\|\left.\right|_{p} on X_(p)^(b)X_{p}^{b} by
{:(1.5)|| varphi|_(p)=s u p{varphi(x):x inB_(p)}","varphi inX_(p)^(b):}\begin{equation*}
\|\left.\varphi\right|_{p}=\sup \left\{\varphi(x): x \in B_{p}\right\}, \varphi \in X_{p}^{b} \tag{1.5}
\end{equation*}
Some useful properties of this norm, whose proofs can be found in [9, 12], are collected in the following proposition. We agree to call a linear functional varphi\varphi on ( X,pX, p ), ( p, bar(p)p, \bar{p} )-bounded if it is both pp - and bar(p)\bar{p}-bounded, where bar(p)\bar{p} is the seminorm conjugate to pp.
Proposition 1.3. If varphi\varphi is a bounded linear functional on a space with asymmetric seminorm ( X,pX, p ), p!=0p \neq 0, then the following assertions hold.
(1) || varphi|_(p)\|\left.\varphi\right|_{p} is the smallest of the numbers L >= 0L \geq 0 for which the inequality (1.4) holds.
(2) We have
{:[(1.6)|| varphi|_(p)=s u p{varphi(x)//p(x):x in X","p(x) > 0}],[(1.7)=s u p{varphi(x):x in X","p(x) < 1}],[(1.8)=s u p{varphi(x):x in X","p(x)=1}.]:}\begin{align*}
\|\left.\varphi\right|_{p} & =\sup \{\varphi(x) / p(x): x \in X, p(x)>0\} \tag{1.6}\\
& =\sup \{\varphi(x): x \in X, p(x)<1\} \tag{1.7}\\
& =\sup \{\varphi(x): x \in X, p(x)=1\} . \tag{1.8}
\end{align*}
(3) If varphi!=0\varphi \neq 0, then || varphi|_(p) > 0\|\left.\varphi\right|_{p}>0. Also, if varphi!=0\varphi \neq 0 and varphi(x_(0))=|| varphi|_(p)\varphi\left(x_{0}\right)=\|\left.\varphi\right|_{p} for some x_(0)inB_(p)x_{0} \in B_{p}, then p(x_(0))=1p\left(x_{0}\right)=1.
(4) If varphi\varphi is ( p, bar(p)p, \bar{p} )-bounded, then
{:[varphi(rB_(p)^('))=(-r|| varphi|_( bar(p)),r||varphi|_(p))quad" and "quad varphi(rB_( bar(p))^('))=(-r|| varphi|_(p),r||varphi|_( bar(p)))],[" where "B_(p)^(')={x in X:p(x) < 1}","B_( bar(p))^(')={x in X: bar(p)(x) < 1}" and "r > 0.]:}\begin{gathered}
\varphi\left(r B_{p}^{\prime}\right)=\left(-\left.r\left\|\left.\varphi\right|_{\bar{p}}, r\right\| \varphi\right|_{p}\right) \quad \text { and } \quad \varphi\left(r B_{\bar{p}}^{\prime}\right)=\left(-\left.r\left\|\left.\varphi\right|_{p}, r\right\| \varphi\right|_{\bar{p}}\right) \\
\text { where } B_{p}^{\prime}=\{x \in X: p(x)<1\}, B_{\bar{p}}^{\prime}=\{x \in X: \bar{p}(x)<1\} \text { and } r>0 .
\end{gathered}
(5) If varphi\varphi is pp-bounded but not bar(p)\bar{p}-bounded, then
varphi(rB_(p)^('))=(-oo,r|| varphi|_(p))\varphi\left(r B_{p}^{\prime}\right)=\left(-\infty, r \|\left.\varphi\right|_{p}\right)
Remark 1.4. A linear functional varphi:X rarrR\varphi: X \rightarrow \mathbb{R} is (p, bar(p))(p, \bar{p})-bounded if and only if
{:(1.9)AA x in X","quad|varphi(x)| <= Lp(x):}\begin{equation*}
\forall x \in X, \quad|\varphi(x)| \leq L p(x) \tag{1.9}
\end{equation*}
for some L >= 0L \geq 0.
Indeed, if L_(1),L_(2) >= 0L_{1}, L_{2} \geq 0 are such that
for all x in Xx \in X, then -varphi(x)=varphi(-x) <= L_(2)p(x),x in X-\varphi(x)=\varphi(-x) \leq L_{2} p(x), x \in X, so (1.9) holds with L=max{L_(1),L_(2)}L=\max \left\{L_{1}, L_{2}\right\}.
Denote by X_( bar(p))^(b)X_{\bar{p}}^{b} the dual cone to ( X, bar(p)X, \bar{p} ) and let X^(**)X^{*} be the conjugate of the seminormed space (X,p^(s))\left(X, p^{s}\right), where p^(s)p^{s} is the symmetric seminorm associated to pp and bar(p)\bar{p} (see (1.1)).
Since
varphi(x) <= Lp(x) <= Lp^(s)(x),x in X,\varphi(x) \leq L p(x) \leq L p^{s}(x), x \in X,
implies |varphi(x)| <= Lp(x),x in X|\varphi(x)| \leq L p(x), x \in X, it follows that X_(p)^(b)X_{p}^{b} is contained in the dual X^(**)X^{*} of ( X,p^(s)X, p^{s} ). Similarly, X_( bar(p))^(b)X_{\bar{p}}^{b} is contained in X^(**)X^{*} too.
For x^(**)inX^(**)x^{*} \in X^{*} put
||x^(**)||=s u p{x^(**)(x):x in X,p^(s)(x) <= 1}.\left\|x^{*}\right\|=\sup \left\{x^{*}(x): x \in X, p^{s}(x) \leq 1\right\} .
Then ||||:}\left\|\|\right. is a norm on X^(**)X^{*} and X^(**)X^{*} is complete with respect to this norm, i.e. is a Banach space (even if p^(s)p^{s} is not a norm, see [11]).
Proposition 1.5. Let ( X,pX, p ) be a space with asymmetric seminorm.
(1) The cones X_(p)^(b)X_{p}^{b} and X_( bar(p))^(b)X_{\bar{p}}^{b} are contained in X^(**)X^{*} and
(2) We have || varphi|_(p)=||- varphi|_( bar(p))\left\|\left.\varphi\right|_{p}=\right\|-\left.\varphi\right|_{\bar{p}}, so that
varphi inX_(p)^(b)" and "|| varphi|_(p) <= r Longleftrightarrow-varphi inX_( bar(p))^(b)" and "||- varphi|_( bar(p)) <= r". "\varphi \in X_{p}^{b} \text { and } \|\left.\varphi\right|_{p} \leq r \Longleftrightarrow-\varphi \in X_{\bar{p}}^{b} \text { and } \|-\left.\varphi\right|_{\bar{p}} \leq r \text {. }
The properties of the dual space X_(p)^(b)X_{p}^{b} were investigated in [21] where, among other things, the analog of the weak* topology of XX was defined. This is denoted by w^(b)w^{b} and has a neighborhood basis at a point varphi inX_(p)^(b)\varphi \in X_{p}^{b}, the family
for n inN,x_(1),dots,x_(n)in Xn \in \mathbb{N}, x_{1}, \ldots, x_{n} \in X and epsilon > 0\epsilon>0. The w^(b)w^{b}-convergence of a net ( varphi_(i):i in I\varphi_{i}: i \in I ) in X_(p)^(b)X_{p}^{b} to varphi inX_(p)^(b)\varphi \in X_{p}^{b} can be characterized in the following way
varphi_(i)rarr"w^(b)"varphi Longleftrightarrow AA x in X,varphi_(i)(x)rarr varphi(x)" in "(R,u)". "\varphi_{i} \xrightarrow{w^{b}} \varphi \Longleftrightarrow \forall x \in X, \varphi_{i}(x) \rightarrow \varphi(x) \text { in }(\mathbb{R}, u) \text {. }
It was shown that w^(b)w^{b} is the restriction of the topology w^(**)=sigma(X^(**),X)w^{*}=\sigma\left(X^{*}, X\right) on X^(**)X^{*} to X_(p)^(b)X_{p}^{b} (see [21]). This study was continued in [9] where separation theorems for convex sets and a Krein-Milman type theorem were proved. In 10 asymmetric locally convex spaces were introduced and their basic properties were studied.
Another direction of investigation is that of best approximation in spaces with asymmetric seminorm. Due to the asymmetry of the seminorm we have two distances. For a nonempty subset YY of a space with asymmetric seminorm ( X,pX, p ) and x in Xx \in X put
{:(1.10)d_(p)(x","Y)=i n f{p(y-x):y in Y}:}\begin{equation*}
d_{p}(x, Y)=\inf \{p(y-x): y \in Y\} \tag{1.10}
\end{equation*}
and
{:(1.11)d_(p)(Y","x)=i n f{p(x-y):y in Y}.:}\begin{equation*}
d_{p}(Y, x)=\inf \{p(x-y): y \in Y\} . \tag{1.11}
\end{equation*}
Note that d_(p)(Y,x)=d_( bar(p))(x,Y)d_{p}(Y, x)=d_{\bar{p}}(x, Y).
Duality formulae and characterization results for best approximation in spaces with asymmetric norm were obtained in [5, 6, 9, 12, 34, 35, The papers [32, 33, 39] are concerned with best approximation in spaces of semiLipschitz functions defined on asymmetric metric spaces (called quasi-metric spaces) in connection with the extension properties of these functions. In the papers [13, 24, 25, 36, supposing that pp is the Minkowski functional p_(C)p_{C} of a bounded convex body CC in a normed space ( X,||||X,\| \| ), some generic existence results for best approximation with respect to the asymmetric norm p_(C)p_{C} were proved, extending similar results from the normed case. As in the symmetric case, the geometric properties of the body CC (or, equivalently, of the functional p_(C)p_{C} ) are essential. A study of the moduli of convexity and smoothness corresponding to p_(C)p_{C} is done in [43].
Best approximation with respect to some asymmetric norms in concrete function spaces of continuous or of integrable functions, called sign-sensitive approximation, was also studied in a series of papers, see [14, 15, 16, 41, the references quoted therein, and the monograph by Krein and Nudelman [23, Ch. 9, §5]).
The present paper, which can be viewed as a sequel to 12 and [9], is concerned mainly with characterizations of the elements of best approximation in a subspace YY of a space with asymmetric norm ( X,pX, p ) and duality results for best approximation. As in the case of (symmetric) normed spaces the characterizations will be done in terms of some linear bounded functionals vanishing on YY. The duality results will involve the annihilator in X_(p)^(b)X_{p}^{b} of the subspace YY. For this reason we start by recalling some extension results for bounded linear functionals on spaces with asymmetric seminorm. For proofs, all resorting to the classical Hahn-Banach extension theorem, see [9, 12].
Theorem 1.6. Let ( X,pX, p ) be a space with asymmetric seminorm and YY a linear subspace of XX. If varphi_(0):Y rarrR\varphi_{0}: Y \rightarrow \mathbb{R} is a linear pp-bounded functional on YY then there exists a pp-bounded linear functional varphi\varphi defined on the whole XX such that
varphi|_(Y)=varphi_(0)quad" and " quad|| varphi|_(p)=||varphi_(0)|_(p).\left.\varphi\right|_{Y}=\varphi_{0} \quad \text { and }\left.\quad\left\|\left.\varphi\right|_{p}=\right\| \varphi_{0}\right|_{p} .
We agree to call a functional varphi\varphi satisfying the conclusions of the above theorem a norm preserving extension of varphi_(0)\varphi_{0}.
Based on this extension result one can prove the following existence result.
Proposition 1.7. Let ( X,pX, p ) be a space with asymmetric seminorm and x_(0)in Xx_{0} \in X such that p(x_(0)) > 0p\left(x_{0}\right)>0. Then there exists a pp-bounded linear functional varphi:X rarrR\varphi: X \rightarrow \mathbb{R} such that
|| varphi|_(p)=1quad" and "quad varphi(x_(0))=p(x_(0)).\|\left.\varphi\right|_{p}=1 \quad \text { and } \quad \varphi\left(x_{0}\right)=p\left(x_{0}\right) .
In its turn, this proposition has the following corollary.
Corollary 1.8. If p(x_(0)) > 0p\left(x_{0}\right)>0 then
Moreover, there exists varphi_(0)inX_(p)^(b),||varphi_(0)|_(p)=1\varphi_{0} \in X_{p}^{b}, \|\left.\varphi_{0}\right|_{p}=1, such that varphi_(0)(x_(0))=p(x_(0))\varphi_{0}\left(x_{0}\right)=p\left(x_{0}\right).
The following proposition is the asymmetric analog of a well known result of Hahn.
Proposition 1.9. ([12]) Let YY be a subspace of a space with asymmetric seminorm ( X,pX, p ) and x_(0)in Xx_{0} \in X.
(1) If d:=d_(p)(x_(0),Y) > 0d:=d_{p}\left(x_{0}, Y\right)>0, then there exists varphi inX_(p)^(b)\varphi \in X_{p}^{b} such that
(i) varphi|_(Y)=0\left.\varphi\right|_{Y}=0,
(ii) || varphi|_(p)=1,quad\|\left.\varphi\right|_{p}=1, \quad and
(iii) varphi(-x_(0))=d\varphi\left(-x_{0}\right)=d.
(2) If bar(d):=d_(p)(Y,x_(0)) > 0\bar{d}:=d_{p}\left(Y, x_{0}\right)>0, then there exists psi inX_(p)^(b)\psi \in X_{p}^{b} such that
(j) psi|_(Y)=0\left.\psi\right|_{Y}=0, (jj)|| psi|_(p)=1,quad(\mathrm{jj}) \|\left.\psi\right|_{p}=1, \quad and (jjj)psi(x_(0))= bar(d)(\mathrm{jjj}) \psi\left(x_{0}\right)=\bar{d}.
2. BEST APPROXIMATION IN SPACES WITH ASYMMETRIC SEMINORM
Let ( X,pX, p ) be a space with asymmetric seminorm, bar(p)\bar{p} the seminorm conjugate to pp and YY a nonempty subset of XX. The distances d_(p)(x,Y)d_{p}(x, Y) and d_(p)(Y,x)d_{p}(Y, x) from an element x in Xx \in X to YY are defined by the formulae (1.10) and (1.11). An element y_(0)in Yy_{0} \in Y such that p(y_(0)-x)=d_(p)(x,Y)p\left(y_{0}-x\right)=d_{p}(x, Y) will be called a pp-nearest point to xx in YY, and an element y_(1)in Yy_{1} \in Y such that p(x-y_(1))= bar(p)(y_(1)-x)=d_( bar(p))(x,Y)p\left(x-y_{1}\right)=\bar{p}\left(y_{1}-x\right)=d_{\bar{p}}(x, Y) is called a bar(p)\bar{p}-nearest point to xx in YY.
Denote by
{:[P_(Y)(x)={y in Y:p(y-x)=d_(p)(x,Y)}","" and "],[(2.1) bar(P)_(Y)(x)={y in Y:p(x-y)=d_(p)(Y,x)}","]:}\begin{align*}
& P_{Y}(x)=\left\{y \in Y: p(y-x)=d_{p}(x, Y)\right\}, \text { and } \\
& \bar{P}_{Y}(x)=\left\{y \in Y: p(x-y)=d_{p}(Y, x)\right\}, \tag{2.1}
\end{align*}
the possibly empty sets of pp-nearest points, respectively bar(p)\bar{p}-nearest points, to xx in YY. The set YY is called pp-proximinal, pp-semi-Chebyshev, pp-Chebyshev if
for every x in Xx \in X the set P_(Y)(x)P_{Y}(x) is nonempty, contains at most one element, contains exactly one element, respectively. Similar definitions are given in the case of bar(p)\bar{p}-nearest points. A semi-Chebyshev set is called also a uniqueness set.
For a nonempty subset YY of a space with asymmetric seminorm ( X,pX, p ), denote by Y_(p)^(_|_)Y_{p}^{\perp} the annihilator of YY in X_(p)^(b)X_{p}^{b}, i.e.
We start by a characterization of nearest points given in [12] we shall need in the sequel.
Proposition 2.1 ( [12][12] ) Let ( X,pX, p ) be a space with asymmetric seminorm, YY a subspace of XX and x_(0)x_{0} a point in XX.
(1) Suppose that d:=d_(p)(x_(0),Y) > 0d:=d_{p}\left(x_{0}, Y\right)>0. An element y_(0)in Yy_{0} \in Y is a pp-nearest point to x_(0)x_{0} in YY if and only if there exists a bounded linear functional varphi:X rarrR\varphi: X \rightarrow \mathbb{R} such that
(i) varphi|_(Y)=0\left.\varphi\right|_{Y}=0,
(ii) || varphi|_(p)=1\|\left.\varphi\right|_{p}=1,
(iii) varphi(-x_(0))=p(y_(0)-x_(0))\varphi\left(-x_{0}\right)=p\left(y_{0}-x_{0}\right).
(2) Suppose that bar(d):=d_(p)(Y,x_(0)) > 0\bar{d}:=d_{p}\left(Y, x_{0}\right)>0. An element y_(0)in Yy_{0} \in Y is a bar(p)\bar{p}-nearest point to x_(0)x_{0} in YY if and only if there exists a bounded linear functional psi:X rarrR\psi: X \rightarrow \mathbb{R} such that
(j) psi|_(Y)=0\left.\psi\right|_{Y}=0, (jj)|| psi|_(p)=1(\mathrm{jj}) \|\left.\psi\right|_{p}=1,
(jjj) psi(x_(0))=p(x_(0)-y_(0))\psi\left(x_{0}\right)=p\left(x_{0}-y_{0}\right).
From this theorem one can obtain characterizations of sets of nearest points.
Corollary 2.2. Let ( X,pX, p ) be a space with asymmetric seminorm, YY a subspace of X,x in XX, x \in X, and ZZ a nonempty subset of YY.
(1) If d=d_(p)(x_(0),Y) > 0d=d_{p}\left(x_{0}, Y\right)>0 then Z subP_(Y)(x)Z \subset P_{Y}(x) if and only if there exists a functional varphi inX_(p)^(b)\varphi \in X_{p}^{b} such that
(i) varphi|_(Y)=0\left.\varphi\right|_{Y}=0,
(ii) || varphi|_(p)=1\|\left.\varphi\right|_{p}=1,
(iii) AA y in Z,varphi(-x_(0))=p(y-x_(0))\forall y \in Z, \varphi\left(-x_{0}\right)=p\left(y-x_{0}\right).
(2) If bar(d)=d_(p)(Y,x_(0)) > 0\bar{d}=d_{p}\left(Y, x_{0}\right)>0 then Z sub bar(P)_(Y)(x)Z \subset \bar{P}_{Y}(x) if and only if there exists a functional psi inX_(p)^(b)\psi \in X_{p}^{b} such that
(j) psi|_(Y)=0\left.\psi\right|_{Y}=0, (jj)|| psi|_(p)=1(\mathrm{jj}) \|\left.\psi\right|_{p}=1, (jjj)AA y in Z,psi(x_(0))=p(x_(0)-y)(\mathrm{jjj}) \forall y \in Z, \psi\left(x_{0}\right)=p\left(x_{0}-y\right).
In the next proposition we extend to the asymmetric case some characterization results for semi-Chebyshev subspaces (see 42, Chapter I, Theorem 3.2]).
Theorem 2.3. Let YY be a subspace of a space with asymmetric norm ( X,pX, p ) such that p(x) > 0p(x)>0 for every x!=0x \neq 0. Then the following assertions are equivalent.
(1) YY is a pp-semi-Chebyshev subspace of XX.
(2) There are no varphi inY_(p)^(_|_)\varphi \in Y_{p}^{\perp} and x_(1),x_(2)in Xx_{1}, x_{2} \in X with x_(1)-x_(2)in Y\\{0}x_{1}-x_{2} \in Y \backslash\{0\}, such that
(i) || varphi|_(p)=1quad\|\left.\varphi\right|_{p}=1 \quad and (ii) varphi(-x_(i))=p(-x_(i)),i=1,2\varphi\left(-x_{i}\right)=p\left(-x_{i}\right), i=1,2.
(3) There are no psi inY_(p)^(_|_),x in X\psi \in Y_{p}^{\perp}, x \in X, and y_(0)in Y\\{0}y_{0} \in Y \backslash\{0\} such that
Proof. (1) =>\Rightarrow (2) Suppose that (2) does not hold. Let varphi inY_(p)^(_|_)\varphi \in Y_{p}^{\perp} and x_(1),x_(2)in Xx_{1}, x_{2} \in X with x_(1)-x_(2)in Y\\{0}x_{1}-x_{2} \in Y \backslash\{0\}, such that the conditions (i) and (ii) of the assertion (2) are satisfied, and put y_(0)=x_(1)-x_(2)y_{0}=x_{1}-x_{2}. Then
By Proposition 2.1, it follows that 0 and y_(0)y_{0} are pp-nearest points to x_(1)x_{1} in YY. (2)=>(3)(2) \Rightarrow(3) Suppose that (3) does not hold. Then there exist psi inY_(p)^(_|_),x in X\psi \in Y_{p}^{\perp}, x \in X, and y_(0)in Y\\{0}y_{0} \in Y \backslash\{0\} such that the conditions ( j ) and ( jj ) of the assertion (3) are fulfilled. It follows that the conditions (i) and (ii) of the assertion (2) are satisfied by varphi=psi,x_(1)=x\varphi=\psi, x_{1}=x and x_(2)=y_(0)-xx_{2}=y_{0}-x, i.e. (2) does not hold. (3)=>(1)(3) \Rightarrow(1) Supposing that (1) does not hold, there exist z in X\\Yz \in X \backslash Y and y_(1),y_(2)in Y,y_(1)!=y_(2)y_{1}, y_{2} \in Y, y_{1} \neq y_{2}, such that
showing that (3) does not hold.
Remark 2.4. Obviously that a similar characterization result holds for bar(p)\bar{p} -semi-Chebyshev subspaces.
Using Corollary 2.2, one can extend Theorem 2.3 to obtain characterizations of pseudo-Chebyshev subspaces, a notion introduced by Mohebi [28] in the case of normed spaces. Concerning other weaker notions of Chebyshev spaces - quasi-Chebyshev subspaces, weak-Chebyshev subspaces, as well as for their behaviour in concrete function spaces, see the papers [26, 27, 29, 31. For a subset ZZ of a vector space XX denote by aff( ZZ ) the affine hull of the set ZZ, i.e. aff(Z)={x in X:EE n inN,EEz_(1),dots,z_(n)in Z,EEa_(1),dots,a_(n)inR,a_(1)+dots+a_(n)=:}\operatorname{aff}(Z)=\left\{x \in X: \exists n \in \mathbb{N}, \exists z_{1}, \ldots, z_{n} \in Z, \exists a_{1}, \ldots, a_{n} \in \mathbb{R}, a_{1}+\ldots+a_{n}=\right. 1 such that x=a_(1)z_(1)+dots+a_(n)z_(n)x=a_{1} z_{1}+\ldots+a_{n} z_{n} }. There exists a unique subspace YY of XX such that aff(Z)=z+Y\operatorname{aff}(Z)=z+Y, for an arbitrary z in Zz \in Z. By definition, the affine dimension of the set ZZ is the dimension of this subspace YY of XX.
A subspace YY of a space with asymmetric norm ( X,pX, p ) is called pp-pseudoChebyshev if it is pp-proximinal and the set P_(Y)(x)P_{Y}(x) has finite affine dimension for every x in Xx \in X.
The following theorem extends a result proved by Mohebi [28] in normed spaces.
Theorem 2.5. Let YY be a subspace of an asymmetric normed space ( X,pX, p ) such that p(x) > 0p(x)>0 for every x!=0x \neq 0. The following assertions are equivalent.
(1) The subspace YY is p-pseudo-Chebyshev.
(2) There do not exist varphi inY_(p)^(_|_),x_(0)in X\varphi \in Y_{p}^{\perp}, x_{0} \in X, and infinitely many linearly independent elements x_(n)in X,n inNx_{n} \in X, n \in \mathbb{N}, with x_(0)-x_(n)in Y,n inNx_{0}-x_{n} \in Y, n \in \mathbb{N}, such that
(i) || varphi|_(p)=1\|\left.\varphi\right|_{p}=1 and
(ii) varphi(-x_(n))=p(-x_(n)),n=0,1,dots\varphi\left(-x_{n}\right)=p\left(-x_{n}\right), n=0,1, \ldots.
(3) There do not exist psi inY_(p)^(_|_),x_(0)in X\psi \in Y_{p}^{\perp}, x_{0} \in X, and infinitely many linearly independent elements y_(n)in Y,n inNy_{n} \in Y, n \in \mathbb{N}, such that
(j) || psi|_(p)=1\|\left.\psi\right|_{p}=1 and
(jj) psi(-x_(0))=p(-x_(0))=p(y_(n)-x_(0)),n=1,2,dots\psi\left(-x_{0}\right)=p\left(-x_{0}\right)=p\left(y_{n}-x_{0}\right), n=1,2, \ldots.
Proof. (1) =>\Rightarrow (2) Suppose that (2) does not hold. Then there exist varphi inY_(p)^(_|_),x_(0)in X\varphi \in Y_{p}^{\perp}, x_{0} \in X, and infinitely many linearly independent elements x_(n)in Xx_{n} \in X, with x_(0)-x_(n)in Y,n inNx_{0}-x_{n} \in Y, n \in \mathbb{N}, satisfying the conditions (i) and (ii). The elements y_(n):=x_(0)-x_(n),n inNy_{n}:=x_{0}-x_{n}, n \in \mathbb{N}, all belong to YY, are linearly independent, and
so that, by Corollary 2.2, they are all contained in P_(Y)(x_(0))P_{Y}\left(x_{0}\right), showing that YY is not pp-pseudo-Chebyshev.
(2) =>\Rightarrow (3) Suppose again that (3) does not hold, and let psi inY_(p)^(_|_),x_(0)in X\psi \in Y_{p}^{\perp}, x_{0} \in X, and the linearly independent elements {y_(n):n=1,2,dots}sub Y\left\{y_{n}: n=1,2, \ldots\right\} \subset Y fulfilling the conditions ( j ) and ( jj ).
Then x_(n):=x_(0)-y_(n),n=1,2,dotsx_{n}:=x_{0}-y_{n}, n=1,2, \ldots, are linearly independent elements in XX and
showing that (2) does not hold.
(3) =>\Rightarrow (1) Supposing that (1) does not hold, there exist an element z in Xz \in X and an infinite set {y_(n):n=1,2,dots}\left\{y_{n}: n=1,2, \ldots\right\} of linearly independent elements contained in P_(Y)(z)P_{Y}(z).
By Corollary 2.2, there exists varphi inY_(p)^(_|_),|| varphi|_(p)=1\varphi \in Y_{p}^{\perp}, \|\left.\varphi\right|_{p}=1, such that
{:[d_(p)(x","Y)=i n f{p(y+y_(1)-z):y in Y}=i n f{p(y^(')-z):y^(')in Y}=],[=d_(p)(z","Y)=p(y_(n)-z)=p(y_(n)-y_(1)-x)","n=2","3","dots","]:}\begin{aligned}
d_{p}(x, Y) & =\inf \left\{p\left(y+y_{1}-z\right): y \in Y\right\}=\inf \left\{p\left(y^{\prime}-z\right): y^{\prime} \in Y\right\}= \\
& =d_{p}(z, Y)=p\left(y_{n}-z\right)=p\left(y_{n}-y_{1}-x\right), n=2,3, \ldots,
\end{aligned}
showing that {y_(n)-y_(1):n=2,3,dots}subP_(Y)(x)\left\{y_{n}-y_{1}: n=2,3, \ldots\right\} \subset P_{Y}(x). By Corollary 2.2, there exists psi inY_(p)^(_|_)\psi \in Y_{p}^{\perp} with || psi|_(p)=1\|\left.\psi\right|_{p}=1 such that
showing that (3) does not hold.
Phelps [37] emphasized for the first time some close connections existing between the approximation properties of the annihilator Y^(_|_)Y^{\perp} of a subspace YY of a normed space XX and the extension properties of the subspace YY. A presentation of various situations in which such a connection occurs is done in [8]. The case of spaces with asymmetric norms was considered in 34, 35.
Let ( X,pX, p ) be a space with asymmetric seminorm and YY a subspace of XX. For a pp-bounded linear functional varphi:Y rarrR\varphi: Y \rightarrow \mathbb{R} denote by
the set of all norm-preserving extensions of the functional varphi\varphi. By the HahnBanach theorem (Theorem 1.6) the set E_(p)(varphi)E_{p}(\varphi) is always nonempty.
For varphi inX_(p)^(b)\varphi \in X_{p}^{b} consider the following minimization problem
A solution to this problem is an element psi_(0)inY_(p)^(_|_)\psi_{0} \in Y_{p}^{\perp} such that ||varphi+psi_(0)|_(p)=gamma(varphi,Y_(p)^(_|_))\| \varphi+\left.\psi_{0}\right|_{p}= \gamma\left(\varphi, Y_{p}^{\perp}\right). Denote by Pi_(Y_(p)^(_|_))(varphi)\Pi_{Y_{p}^{\perp}}(\varphi) the set of all these solutions.
Theorem 2.6. If the linear functional varphi:X rarrR\varphi: X \rightarrow \mathbb{R} is (p, bar(p))(p, \bar{p})-bounded then the minimization problem (2.2) has a solution and the following formulae hold
{:(2.3)gamma(varphi,Y_(p)^(_|_))=|| varphi|_(Y)|_(p)quad" and "quadPi_(Y_(p)^(_|_))(varphi)=E_(p)( varphi|_(Y))-varphi:}\begin{equation*}
\gamma\left(\varphi, Y_{p}^{\perp}\right)=\|\left.\left.\varphi\right|_{Y}\right|_{p} \quad \text { and } \quad \Pi_{Y_{p}^{\perp}}(\varphi)=E_{p}\left(\left.\varphi\right|_{Y}\right)-\varphi \tag{2.3}
\end{equation*}
Proof. Let varphi inX_(p)^(b)nnX_( bar(p))^(b)\varphi \in X_{p}^{b} \cap X_{\bar{p}}^{b} and psi inY_(p)^(_|_)\psi \in Y_{p}^{\perp}. Then
Conversely, if psi inPi_(Y_(p)^(_|_))(varphi)\psi \in \Pi_{Y_{p}^{\perp}}(\varphi), then Phi:=varphi+psi\Phi:=\varphi+\psi satisfies Phi|_(Y)= varphi|_(Y)\left.\Phi\right|_{Y}=\left.\varphi\right|_{Y} and || Phi|_(p)=||varphi+ psi|_(p)=gamma(varphi,Y_(p)^(_|_))=|| varphi|_(Y)|_(p)\left\|\left.\Phi\right|_{p}=\right\| \varphi+\left.\psi\right|_{p}=\gamma\left(\varphi, Y_{p}^{\perp}\right)=\|\left.\left.\varphi\right|_{Y}\right|_{p}, i.e. Phi inE_(p)( varphi|_(Y))\Phi \in E_{p}\left(\left.\varphi\right|_{Y}\right) and
the annihilator Y^(_|_)Y^{\perp} of a subspace YY of XX in the symmetric dual X^(**)X^{*} of the seminormed space (X,p^(s))^(**)\left(X, p^{s}\right)^{*}, it follows that Y^(_|_)Y^{\perp} is a subspace of X^(**)X^{*}. Consider on X^(**)X^{*} the asymmetric extended norm |||_(p)^(**):X^(**)rarr[0;oo]\|\left.\right|_{p} ^{*}: X^{*} \rightarrow[0 ; \infty] defined by
|| varphi|_(p)^(**)=s u p varphi(B_(p)).\|\left.\varphi\right|_{p} ^{*}=\sup \varphi\left(B_{p}\right) .
and || varphi|_(p)^(**)=||varphi|_(p)\left.\left\|\left.\varphi\right|_{p} ^{*}=\right\| \varphi\right|_{p} for varphi inX_(p)^(b)\varphi \in X_{p}^{b} (see Proposition 1.5).
For varphi inX_(p)^(b)\varphi \in X_{p}^{b} consider the distance from varphi\varphi to Y^(_|_)Y^{\perp} defined by
Because ||varphi-0|_(p)^(**)=||varphi|_(p) < oo\left.\left\|\varphi-\left.0\right|_{p} ^{*}=\right\| \varphi\right|_{p}<\infty this distance is always finite. Put
Therefore d_(p)(Y^(_|_),varphi)=|| varphi|_(Y)|_(p)d_{p}\left(Y^{\perp}, \varphi\right)=\|\left.\left.\varphi\right|_{Y}\right|_{p} and varphi-E_(p)( varphi|_(Y))subP_(Y^(_|_))(varphi)\varphi-E_{p}\left(\left.\varphi\right|_{Y}\right) \subset P_{Y^{\perp}}(\varphi).
If psi inP_(Y^(_|_))(varphi)\psi \in P_{Y^{\perp}}(\varphi) and Phi:=varphi-psi\Phi:=\varphi-\psi, then Phi|_(Y)= varphi|_(Y)\left.\Phi\right|_{Y}=\left.\varphi\right|_{Y} and || Phi|_(p)=||varphi- psi|_(p)=d_(p)(Y^(_|_),varphi)=|| varphi|_(Y)|_(p)\left\|\left.\Phi\right|_{p}=\right\| \varphi-\left.\psi\right|_{p}= d_{p}\left(Y^{\perp}, \varphi\right)=\|\left.\left.\varphi\right|_{Y}\right|_{p}, i.e. Phi inE_(p)( varphi|_(Y))\Phi \in E_{p}\left(\left.\varphi\right|_{Y}\right), showing that varphi-P_(Y^(_|_))(varphi)subE_(p)( varphi|_(Y))\varphi-P_{Y^{\perp}}(\varphi) \subset E_{p}\left(\left.\varphi\right|_{Y}\right), or equivalently, P_(Y^(_|_))(varphi)sub varphi-E_(p)( varphi|_(Y))P_{Y^{\perp}}(\varphi) \subset \varphi-E_{p}\left(\left.\varphi\right|_{Y}\right).
From these theorems we obtain some uniqueness conditions for the minimization problems we have considered, in terms of the uniqueness of normpreserving extensions.
Corollary 2.8. Let (X,p)(X, p) be a space with asymmetric seminorm and YY a subspace of XX.
(1) If every f inY_(p)^(b)f \in Y_{p}^{b} has a unique norm preserving extension F inX_(p)^(b)F \in X_{p}^{b}, then the minimization problem (2.2) has a unique solution for every varphi inX_(p)^(b)\varphi \in X_{p}^{b}.
(2) Every point varphi inX_(p)^(b)\varphi \in X_{p}^{b} has a unique bar(p)\bar{p}-nearest point in Y^(_|_)Y^{\perp} if and only if every f inY_(p)^(b)f \in Y_{p}^{b} has a unique norm-preserving extension F inX_(p)^(b)F \in X_{p}^{b}.
Proof. (1) If every f inY_(p)^(b)f \in Y_{p}^{b} has a unique norm-preserving extension F inX_(p)^(b)F \in X_{p}^{b}, then for every varphi inX_(p)^(b)\varphi \in X_{p}^{b} the set Pi_(Y_(p)^(_|_))(varphi)=varphi+E_(p)( varphi|_(Y))\Pi_{Y_{p}^{\perp}}(\varphi)=\varphi+E_{p}\left(\left.\varphi\right|_{Y}\right) contains exactly one element.
(2) Similarly, P_(Y^(_|_))(varphi)=varphi-E_(p)( varphi|_(Y))P_{Y^{\perp}}(\varphi)=\varphi-E_{p}\left(\left.\varphi\right|_{Y}\right) contains exactly one element, provided every f inY_(p)^(b)f \in Y_{p}^{b} has exactly one norm-preserving extension F inX_(p)^(b)F \in X_{p}^{b}.
Conversely, suppose that there exists f inY_(p)^(b)f \in Y_{p}^{b} having two distinct normpreserving extensions F_(1),F_(2)inX_(p)^(b)F_{1}, F_{2} \in X_{p}^{b}. Then
REMARK 2.9. We can not prove the reverse implication in the assertion (1) of the above corollary. To do this we would need an extension theorem for ( p, bar(p)p, \bar{p} )-bounded linear functionals, preserving both pp - and bar(p)\bar{p}-norm, and we are not aware of such a result.
Some results connecting the epsilon\epsilon-approximations and epsilon\epsilon-extensions were obtained by Rezapour [38]. In the next proposition we transpose these results to the asymmetric case.
Let ( X,pX, p ) be a space with asymmetric seminorm and YY a subspace of XX. For x in Xx \in X and epsilon > 0\epsilon>0 let
denote the nonempty sets of epsilon-p\epsilon-p-, respectively epsilon- bar(p)\epsilon-\bar{p}-nearest points to xx in YY. For varphi inX_(p)^(b)\varphi \in X_{p}^{b} consider the set of epsilon\epsilon-solutions of the minimization problem (2.2)
E_(p)^(epsilon)(f)={F inX_(p)^(b):F|_(Y)=f" and "||F|_(p) <= ||f|_(p)+epsilon},E_{p}^{\epsilon}(f)=\left\{F \in X_{p}^{b}:\left.F\right|_{Y}=f \text { and }\left.\left\|\left.F\right|_{p} \leq\right\| f\right|_{p}+\epsilon\right\},
the set of epsilon\epsilon-extensions of a functional f inY_(p)^(b)f \in Y_{p}^{b}.
These two sets are related in the following way.
Proposition 2.10. Let ( X,pX, p ) be a space with asymmetric seminorm, YY a subspace of XX and varphi inX_(p)^(b)\varphi \in X_{p}^{b}. Then
Working with the annihilator Y^(_|_)Y^{\perp} of the subspace YY in the symmetric dual X^(**)=(X,p^(s))^(**)X^{*}=\left(X, p^{s}\right)^{*} given by (2.4) and putting
we have
Proposition 2.11. Let YY be a subspace of a space with asymmetric seminorm (X,p),epsilon > 0(X, p), \epsilon>0, and varphi inX_(p)^(b)\varphi \in X_{p}^{b}. Then