Surjective Mappings in the Hyers–Ulam Theorem and the Gromov–Hausdorff Distance The work of A.A. Tuzhilin was supported by grant No. 25-21-00152 of the Russian Science Foundation, by National Key R&D Program of China (Grant No. 2020YFE0204200), as well as by the Sino-Russian Mathematical Center at P
S. A. Bogatyĭ, E. A. Reznichenko, A. A. Tuzhilin · 2025
Read Surjective Mappings in the Hyers–Ulam Theorem and the Gromov–Hausdorff Distance The work of A.A. Tuzhilin was supported by grant No. 25-21-00152 of the Russian Science Foundation, by National Key R&D Program of China (Grant No. 2020YFE0204200), as well as by the Sino-Russian Mathematical Center at P by S. A. Bogatyĭ, E. A. Reznichenko, A. A. Tuzhilin (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-2512-22776.txt.
Introduction
S. A. Bogatyi, E. A. Reznichenko, and A. A. Tuzhilin prove a stability result about maps between metric spaces that almost preserve distances and almost cover their targets. The setting uses cardinality homogeneous spaces, where every nonempty open piece has as many points as the whole space, and both spaces have the same total cardinality. A map that is nearly an isometry and nearly surjective is shown to be close to a perfect one.
Precisely, a near-isometry with a small surjectivity defect between such spaces guarantees a genuine bijective map whose distance error is only slightly larger. That bridge lets the authors reduce the Dilworth-Tabor theorem to the earlier Gevirtz-Omladič-Šemrl approximation theorem. In particular, hard questions about when Banach spaces must be isometric can be rerouted through approximation results, connecting two lines of work in the geometry of metric and normed spaces.
Catalog record
| CREATOR | S. A. Bogatyĭ, E. A. Reznichenko, A. A. Tuzhilin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-22776
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Surjective Mappings in the Hyers–Ulam Theorem and the Gromov–Hausdorff Distance The work of A.A. Tuzhilin was supported by grant No. 25-21-00152 of the Russian Science Foundation, by National Key R&D Program of China (Grant No. 2020YFE0204200), as well as by the Sino-Russian Mathematical Center at Peking University. Partial of the work by A.A. Tuzhilin were done in Sino-Russian Math. center, and he thanks the Math. Center for the invitation and the hospitality. S. A. Bogatyĭ, E. A. Reznichenko, A. A. Tuzhilin Abstract A topological space is said to be cardinality homogeneous if every nonempty open subset has the same cardinality as the space itself. Let X X and Y Y be cardinality homogeneous metric spaces of the same cardinality. If there exists a δ \delta -surjective d d -isometry between such equicardinal cardinality homogeneous metric spaces X X and Y Y , then there exists a bijective ( d + 2 δ ) (d+2\delta) -isometry between X X and Y Y . This result allows us to reduce the Dilworth–Tabor theorem to the Gevirtz–Omladič–Šemrl theorem on approximation by isometries and, in particular, to questions concerning the isometry of Banach spaces. Keywords : metric space, Gromov–Hausdorff distance, δ \delta -surjective mapping, d d -isometry, isometry. 1 Introduction The celebrated Mazur–Ulam theorem of 1932 states that every surjective isometry f\colon V\to W between real normed spaces is affine [ 1 ] . One direction for strengthening this theorem arose from the 1945 work of Hyers and Ulam [ 2 ] . D. H. Hyers and S. M. Ulam posed the following question: Is there a constant K , depending only on the Banach spaces V and W , such that for every d>0 and every surjective d -isometry f\colon V\to W there exists a surjective isometry U\colon V\to W satisfying \|f-U\|=\sup\bigl\{\|f(x)-U(x)\|:x\in V\bigr\}\leq K\,d? They noted that the surjectivity assumption for f is essential and provided an affirmative answer in the case V=W=\ell_{2} , the Hilbert space (with K=10 ). The efforts of numerous mathematicians over many years culminated in a complete affirmative resolution by J. Gevirtz in 1983 [ 3 ] . In 1995, M. Omladič and P. Šemrl [ 4 ] established the sharp constant K=2 . S. J. Dilworth [ 5 ] and J. Tabor [ 6 ] relaxed the surjectivity condition: If d,\delta\geq 0 and f\colon V\to W is a d -isometry between Banach spaces with f(0)=0 that maps V \delta -surjectively onto a closed subspace L\subset W ( i.e., d_{H}(L,f(V))\leq\delta ) , then there exists a bijective linear isometry U\colon V\to L such that (1) \|f-U\|\leq M=12d+5\delta\quad\text{(Dilworth)};\qquad M=2d+35\delta\quad\text{($L=W$, Tabor)}. In 2003, P. Šemrl and J. Väisälä [ 7 ] showed that the right-hand side of (\ref{eq:d1}) can be replaced by M=2d+2\delta . They also proved that when L=W , the universal bound M=2d holds . Inspired by the proof technique of the Cantor–Bernstein theorem, the present paper approximates two mappings f\colon X\to Y and g\colon Y\to X between sets by a bijection (Theorem 3.1 ). Two applications of this approximation are presented. First, we show that for cardinality homogeneous metric spaces of the same cardinality, the Gromov–Hausdorff distance can be characterised using bijective mappings (Theorem 4.6 ). Theorem 3.1 and its corollaries also permit a reduction of the approximation problem for d -isometric \delta -surjective mappings of Banach spaces (onto a closed linear subspace) to the approximation of (d+2\delta) -isometric (or (d+6\delta) -isometric) surjective mappings (see the general Theorems 5.5 and 5.8 and Propositions 6.4 and 6.6 for Banach spaces). This yields a simpler reduction, in our view, of the Dilworth–Tabor theorem to the theorem of J. Gevirtz (see Corollaries 6.5 and 6.7 ). A further application of the Šemrl–Väisälä theorem shows that the coefficient d+2\delta arising in this reduction is non-essential. 2 Preliminaries and Notation Let (X,d) be a metric space (distances may be infinite). For \varepsilon>0 , x,y\in X , and M\subset X , we denote by \displaystyle|xy| \displaystyle=d(x,y) \displaystyle\textit{the distance between $x$ and $y$}; \displaystyle U_{\varepsilon}(x) \displaystyle=\{y\in X:|xy|<\varepsilon\} \displaystyle\textit{the open $\varepsilon$-ball centered at $x$}; \displaystyle U_{\varepsilon}(M) \displaystyle=\bigcup_{x\in M}U_{\varepsilon}(x) \displaystyle\textit{the $\varepsilon$-neighborhood of $M$}; \displaystyle\operatorname{diam}M \displaystyle=\sup\{|xy|:x,y\in M\} \displaystyle\textit{the diameter of $M$}. For subsets A,B\subset X , the Hausdorff distance is d_{H}(A,B)=\inf\{\varepsilon>0:A\subset U_{\varepsilon}(B)\ \text{and}\ B\subset U_{\varepsilon}(A)\}. The Gromov–Hausdorff distance d_{GH}(X,Y) between metric spaces X and Y is the infimum of d_{H}(X,Y) over all isometric embeddings of X and Y into a common metric space Z ; see [ 8 , 9 , 10 ] for details. A correspondence between sets X and Y is a relation R\subset X\times Y such that every x\in X is related to at least one y\in Y , and vice versa. For nonempty R\subset X\times Y , its distortion is \operatorname{dis}R=\sup\bigl\{\bigl||xx^{\prime}|-|yy^{\prime}|\bigr|:(x,y),(x^{\prime},y^{\prime})\in R\bigr\}. An equivalent expression for the Gromov–Hausdorff distance is [ 11 , Theorem 7.3.25] d_{GH}(X,Y)=\frac{1}{2}\inf\{\operatorname{dis}R:R\subset X\times Y\ \text{is a correspondence}\}. The co-distortion of nonempty R,R^{\prime}\subset X\times Y is \operatorname{codis}(R,R^{\prime})=\sup\bigl\{\bigl||xx^{\prime}|-|yy^{\prime}|\bigr|:(x,y)\in R,\ (x^{\prime},y^{\prime})\in R^{\prime}\bigr\}. For mappings f\colon X\to Y and g\colon Y\to X , the distortion of f is \operatorname{dis}f=\sup\bigl\{\bigl||f(x)f(x^{\prime})|-|xx^{\prime}|\bigr|:x,x^{\prime}\in X\bigr\}, and the co-distortion is \operatorname{codis}(f,g)=\sup\bigl\{\bigl||xg(y)|-|f(x)y|\bigr|:x\in X,\ y\in Y\bigr\}. The graph 1 1 1 In set theory, the graph of a mapping is identified with the mapping itself. We use \operatorname{Gr}f for clarity. of f is \operatorname{Gr}f=\{(x,f(x)):x\in X\}. For a relation Q\subset Y\times X , we denote Q^{-1}=\{(x,y)\in X\times Y:(y,x)\in Q\} the converse relation . We denote \operatorname{Gr}^{-1}g=(\operatorname{Gr}g)^{-1}=\{(g(y),y):y\in Y\}. The following facts are easily verified: • R\subset X\times Y is a correspondence if and only if there exist f\in Y^{X} and g\in X^{Y} with \operatorname{Gr}f\cup\operatorname{Gr}^{-1}g\subset R ; • \operatorname{dis}f=\operatorname{dis}(\operatorname{Gr}f) ; • \operatorname{codis}(f,g)=\operatorname{codis}(\operatorname{Gr}f,\operatorname{Gr}^{-1}g) ; • \operatorname{dis}(R\cup R^{\prime})=\max\{\operatorname{dis}R,\operatorname{codis}(R,R^{\prime}),\operatorname{dis}R^{\prime}\} . Thus, d_{GH}(X,Y)=\inf\{d_{GH}(f,g):f\in Y^{X},\ g\in X^{Y}\}, where d_{GH}(f,g)=\frac{1}{2}\max\{\operatorname{dis}f,\operatorname{codis}(f,g),\operatorname{dis}g\}. This form is noted in [ 12 , Proposition 1(5), Remark 2] for bounded spaces. The density of a topological space X is the cardinal d(X)=\min\{|D|:D\subset X=\overline{D}\}. 3 Approximation of Mappings by a Bijection Theorem 3.1 . Let X and Y be sets equipped with covers \mathcal{L}_{X} and \mathcal{L}_{Y} such that |X|=|Y|=|V| for every V\in\mathcal{L}_{X} and |U| for every U\in\mathcal{L}_{Y} . Then for any mappings f\colon X\to Y and g\colon Y\to X , there exists a bijection \tilde{f}\colon X\to Y such that for every x\in X at least one of the following holds : (1) \tilde{f}(x) and f(x) lie in the same member of \mathcal{L}_{Y} ; (2) g(\tilde{f}(x)) and x lie in the same member of \mathcal{L}_{X} . Proof. Choose functions U\colon X\to\mathcal{L}_{X} and V\colon Y\to\mathcal{L}_{Y} with x\in U(x) and y\in V(y) . Enumerate the points using the cardinal \tau=|X|=|Y| : X=\{x_{\alpha}:\alpha<\tau\} and Y=\{y_{\alpha}:\alpha<\tau\} . For \alpha\leq\tau , set X_{\alpha}=\{x_{\beta}:\beta<\alpha\} and Y_{\alpha}=\{y_{\beta}:\beta<\alpha\} . By transfinite induction on \alpha\leq\tau , construct injective mappings f_{\alpha}\colon X_{\alpha}\to Y and g_{\alpha}\colon Y_{\alpha}\to X satisfying, for every \beta<\alpha : • f_{\alpha}|_{X_{\beta}}=f_{\beta} and g_{\alpha}|_{Y_{\beta}}=g_{\beta} ; • if f_{\alpha}(x_{\beta})\in Y_{\alpha} , then x_{\beta}=g_{\alpha}(f_{\alpha}(x_{\beta})) ; • if g_{\alpha}(y_{\beta})\in X_{\alpha} , then y_{\beta}=f_{\alpha}(g_{\alpha}(y_{\beta})) ; • either f_{\alpha}(x_{\beta})\in V(f(x_{\beta})) or g_{\alpha}(y_{\beta})\in U(g(y_{\beta})) . If \alpha is limit, take unions of the previous mappings. For successor \alpha=\beta+1 , extend f_{\alpha} and g_{\alpha} as follows. To define f_{\alpha}(x_{\beta}) : Case 1: If x_{\beta}\notin g_{\beta}(Y_{\beta}) , choose y\in V(f(x_{\beta}))\setminus(Y_{\beta}\cup f_{\beta}(X_{\beta})) and set f_{\alpha}(x_{\beta})=y . Case 2: If x_{\beta}\in g_{\beta}(Y_{\beta}) , set f_{\alpha}(x_{\beta})=y where y\in Y_{\beta} and x_{\beta}=g_{\beta}(y) . To define g_{\alpha}(y_{\beta}) : Case 1: If y_{\beta}\notin f_{\alpha}(X_{\alpha}) , choose x\in U(g(y_{\beta}))\setminus(X_{\alpha}\cup g_{\beta}(Y_{\beta}))