## Abstract

We consider seriously the analogy between interpolation of nonlinear functions and manifold learning from samples, and examine the results of transferring ideas from each of these domains to the other. Illustrative examples are given in approximation theory, variational calculus (closed geodesics), and quantitative cobordism theory.

This is a preview of subscription content, access via your institution.

## Notes

Or geometry; I shall not distinguish between these.

Alternatively, if a connected submanifold

*M*didn’t separate, one would construct a closed curve transverse to*M*lying in a neighborhood of*M*, and then argue that this would contradict the simple connectivity of the sphere (by homotopy invariance of intersection numbers).In some cases the connection between the homotopy theory and the geometric problem is sufficiently tight that

*any*new geometric complexity statement would readily imply some, although perhaps not optimal, result about the complexity of homotopies.Many classical problems in analysis are also solved by homotopy methods. Typically, one starts, though, with an explicit homotopy to apply them to. I hope that the ideas discussed in later sections can be of use in that setting.

Unless something extraordinary is happening—say

*f*is locally constant.The first level for \(E = \int \langle \nabla f, \nabla f \rangle / \int \langle f, f \rangle \) where

*i*-dimensional mod 2 homology appears in the projective space of the Sobolev space \(H^{1,2}(M)\) is in the*i*th positive eigenvalue of the Laplacian.The reader could well want to know what can distinguish manifolds which mutually approach one another. The simplest invariants are odd primary characteristic classes of the topological tangent bundle. However, there are additional subtle secondary invariants that are torsion analogues of the invariants used by Atiyah and Bott [1] to distinguish lens spaces from one another.

The process described, though, is useful even when \(B\Gamma \) does not exist as a finite complex. The homology described can be used as a substitute that occasionally has more useful properties. See [20] for an example of this.

We shall not discuss here what to do about noise or when populations have overlapping images.

Of course, in the noncompact case, Morse theory in its ordinary sense requires a properness condition, or a Palais–Smale condition in the infinite-dimensional setting.

Other applications to approximation theory can be found in [49].

Except the global maximum, which doesn’t close any \(H_0\) bar. Thought of as being the circle, it begins the nontrivial \(H_1\) bar.

See the proof for a definition, or see, e.g., the Wikipedia page on Dehn functions of groups.

Here I mean a large scale idea of texture that is analogous to the differences that one would notice on the small scale if one took the graphs of (Baire generic) Hölder functions for different exponents. If you consider the persistence homology, the Hölder exponent would be apparent in the “length spectrum of the bars” as will be explained in detail in a future paper with Yuliy Baryshnikov (and is not hard to see by wrinkling a map by putting in as many bumps at small scale as permitted by dimension and the Hölder condition).

This linearity conjecture implies an estimate for \(\eta \)-invariants of manifolds (spectral invariants of odd-dimensional manifolds defined by Atiyah–Patodi–Singer) in terms of volume; Cheeger and Gromov proved these directly [13]. While the quantitative cobordism work we discuss is not strong enough to give the Cheeger–Gromov inequality, a different approach—with very interesting complexity applications—was given by [14].

Actually, there is a beautiful proof of Thom’s theorem [6] that avoids this. This method actually directly leads to a polynomial estimate (of degree \(2^n\)) for

*k*(*W*). Unlike what we are about to discuss, this does not extend to the oriented case.To see the issue, the number of point inverses for a Lipschitz map from \(S^1\) to itself can we be infinite, even with \(L = 2\). However, the degree of a Lipschitz map is bounded by

*L*and one can \(C^0\)-approximate such a map by one where the number inverse images grows linearly with*L*.The construction we are about to explain is called the Whitehead product in homotopy theory. (See, e.g., [54].)

However, for \({\text {Lip}}(S^1:M)\) for

*M*a 3-dimensional Sol manifold the diameter is \(\exp (L)\), as suggested by a worst case analysis based on covering numbers.

## References

M. Atiyah and R. Bott. A Lefschetz fixed point formula for elliptic complexes: II. Applications.

*Ann. of Math.*, 88:451–491, 1968.N. Amenta and M. Bern. Surface reconstruction by voronoi filtering.

*Discrete Comput. Geom.*, 22:481–504, 1999.Y. Barzdin. On the realization of networks in three-dimensional space. In A. N. Shiryayev, editor,

*Selected Works of A. N. Kolmogorov, Volume III*, volume 27 of*Mathematics and its Applications (Soviet Series)*, pages 194–202. Springer, Dordrecht, 1993.J. Boissonnat, F. Chazal, and M. Yvinec. Computational geometry and topology for data analysis. To appear.

J. Boissonnat, L. Guibas, and S. Oudot. Manifold reconstruction in arbitrary dimensions using witness complexes.

*Discrete Comput. Geom.*, 42(1):37–70, 2009.S. Buoncristiano and D. Hacon. An elementary geometric proof of two theorems of Thom.

*Topology*, 20(1):97–99, 1981.J. Bourgain. On Lipschitz embedding of finite metric spaces in Hilbert space.

*Israel J. Math.*, 52:46–52, 1985.E. Brown. Finite computability of Postnikov complexes.

*Ann. of Math.*, 65(1):1–20, 1957.W. Browder.

*Surgery on simply connected manifolds*. Springer Verlag, 1972.J. Block and S. Weinberger. Large scale homology theories and geometry. In

*Geometric Topology: 1993 Georgia International Topology Conference*, AMS/IP Stud. Adv. Math., pages 522–569, Providence, RI, 1997. Amer. Math. Soc.G. Chambers, D. Dotterer, F. Manin, and S. Weinberger. Quantitative null-cobordism.

*J. Amer. Math. Soc.*, 31(4):1165–1203, 2018.S. Cheng, T. Dey, and E. A. Ramos. Manifold reconstruction from point samples. In

*SODA ’05 Proceedings of the sixtieth annual ACM-SIAM symposium on discrete algorithms*, pages 1018–1027, Philadelphia, PA, 2005. Society for Industrial and Applied Mathematics.J. Cheeger and M. Gromov. On the characteristic numbers of complete manifolds of bounded curvature and finite volume. In I. Chavel and H. M. Farkas, editors,

*Differential Geometry and Complex Analysis*, pages 115–154. Springer, Berlin, Heidelberg, 1985.J. Cha. A topological approach to Cheeger-Gromov universal bounds for von Neumann \(\rho \)-invariants.

*Comm. Pure and Applied Math.*, 69:1154–1209, 2016.G. Carlsson and F. Memoli. Classifying clustering schemes.

*Found. Comput. Math.*, 13:221–252, 2013.G. Chambers, F. Manin, and S. Weinberger. Quantitative null homotopy and rational homotopy type.

*Geom. Funct. Anal.*, 28(3):563–588, 2018.D. Cohen-Steiner, H. Edelsbrunner, and J. Harer. Stability of persistence diagrams.

*Discrete Comput. Geom.*, 37(1):103–120, 2007.G. Carlsson and A. Zomorodian. Computing persistent homology.

*Discrete Comput. Geom.*, 33(2):249–274, 2005.F. Cucker and D. Zhou.

*Learning Theory: An Approximation Theory Viewpoint*. Cambridge Monographs on Applied and Computational Mathematics. Cambridge University Press, 2007.A. Dranishnikov, S. Ferry, and S. Weinberger. Large Riemannian manifolds which are flexible.

*Ann. of Math.*, 157:919–938, 2003.A. Dranishnikov, S. Ferry, and S. Weinberger. An infinite dimensional phenomenon in finite dimensional topology. Preprint, 2017.

M. DoCarmo.

*Riemannian Geometry*. Birkhäuser Verlag, 1992.H. Edelsbrunner and D. Grayson. Edgewise subdivision of a simplex.

*Discrete Comput. Geom.*, 24(4):707–719, 2000.H. Edelsbrunner, D. Letscher, and A. Zomorodian. Topological persistence and simplification.

*Discrete Comput. Geom.*, 28:511–533, 2002.Y. Eliashberg and N. Mishachev.

*Introduction to the h-Principle*, volume 48 of*Graduate Studies in Mathematics*. American Mathematical Society, Providence, RI, 2002.C. Fefferman. Whitney’s extension problem for \(c^m\).

*Ann. of Math.*, 164:313–359, 2006.C. Fefferman. \(c^m\)-extension by linear operators.

*Ann. of Math.*, 166:779–835, 2007.S. Ferry. Topological finiteness theorems for manifolds in Gromov-Hausdorff space.

*Duke Math. J.*, 74(1):95–106, 1994.C. Fefferman, S. Mitter, and H. Naryanan. Testing the manifold hypothesis.

*J. Amer. Math. Soc.*, 29:983–1049, 2016.S. Ferry and S. Weinberger. Quantitative algebraic topology and Lipschitz homotopy.

*Proc. Natl. Acad. Sci. U.S.A.*, 110:19246–19250, 2013.M. Gromov and L. Guth. Generalizations of the Kolmogorov-Barzdin embedding estimates.

*Duke Math. J.*, 161:2549–2603, 2012.M. Gromov. Homotopical effects of dilation.

*J. Diff. Geo.*, 13:313–310, 1978.M. Gromov. Groups of polynomial growth and expanding maps.

*Publ. Math. IHÉS*, 53:51–78, 1981.M. Gromov.

*Partial Differential Relations*. Ergebnisse der Mathematik und ihrer Grenzgebiete. 3. Folge. Springer Verlag, 1986.M. Gromov. Dimension, non-linear spectra, and width. In J. Lindenstrauss and V. D. Milman, editors,

*Geometric Aspects of Functional Analysis*, volume 1317 of*Lecture Notes in Mathematics*, pages 132–184. Springer, Berlin, Heidelberg, 1988.M. Gromov.

*Metric Structures for Riemannian and non-Riemannian Spaces*. Modern Birkhäuser Classics. Birkhäuser Verlag, 1999.M. Gromov. Quantitative homotopy theory. In H. Rossi, editor,

*Prospects in Mathematics (Princeton, NJ, 1996)*, pages 45–49. Amer. Math. Soc., Providence, RI, 1999.J. Kleinberg. An impossibility theorm for clustering. In S. Becker, K. Obermayer, and S. Thrun, editors,

*Advances in Neural Information Processing Systems 15 (NIPS 2002)*, pages 463–470. MIT Press, Cambridge, MA, 2002.F. Manin. Plato’s cave and differential forms. Preprint, 2018.

J. Matousek. Lecture notes on metric embeddings. Preprint, 2013.

J. Milnor.

*Morse Theory*, volume 51 of*Annals of Mathematics Studies*. Princeton University Press, Princeton, NJ, 1963.J. Milnor. A note on curvature and fundamental group.

*J. Differ. Geom.*, 2:1–7, 1968.F. Manin and S. Weinberger. The Gromov-Guth embedding theorem. Appendix to [11].

A. Nabutovsky. Non-recursive functions, knots “with thick ropes” and self-clenching “thick” hyperspheres.

*Commun. Pure Appl. Math.*, 48(4):1–50, 1995.A. Nabutovsky. Morse landscapes of Riemannian functionals and related problems. In

*Proceedings of the International Congress of Mathematicians: Hyderabad, India*, pages 862–881, 2010.P. Niyogi, S. Smale, and S. Weinberger. Finding the homology of submanifolds with high confidence from random samples.

*Discrete Comput. Geom.*, 39:419–441, 2008.A. Nabutovsky and S. Weinberger. Variational problems for Riemannian functionals and arithmetic groups.

*Publ. Math. IHÉS*, 92:5–62, 2000.S. Oudot.

*Persistence Theory: From Quiver Representations to Data Analysis*, volume 209 of*Mathematical Surveys and Monographs*. American Mathematical Society, Providence, RI, 2015.I. Polterovich, L. Polterovich, and V. Stojisavljević. Persistence barcodes and Laplace eigenfunctions on surfaces. Preprint, 2017.

L. Polterovich, D. Rosen, K. Samvelyan, and J. Zhang. Persistent homology for symplectic topologists. Preprint, 2018.

V. Robins. Toward computing homology from finite approximations.

*Topology Proceedings*, 24:503–532, 1999.J. Roe.

*Index theory, coarse geometry, and topology of manifolds*, volume 90 of*CMBS Regional Conference Series in Mathematics*. American Mathematical Society, Providence, RI, 1996.S. Smale. The classification of immersions of spheres in Euclidean spaces.

*Ann. of Math.*, 69:327–344, 1959.E. Spanier.

*Algebraic Topology*. McGraw-Hill, 1956.R. Strong.

*Notes on cobordism theory*. Mathematical Notes. Princeton University Press, Princeton, NJ, 1968.D. Sullivan. Infinitesimal computations in topology.

*Publ. Math. IHÉS*, 47:269–331, 1977.R. Thom. Quelques propriétés globales des variétés différentiables.

*Comment. Math. Helv.*, 28:17–86, 1954.J. Traub and A. Werschulz.

*Complexity and information*. Lezioni Lincee. Cambridge University Press, Cambridge, United Kingdom, 1999.L. Valiant. A theory of the learnable.

*Commun. ACM*, 27(11):1134–1142, 1984.C. T. C. Wall.

*Surgery on Compact Manifolds*. London Mathematical Society Monographs. Academic Press, Cambridge, MA, 1969.S. Weinberger.

*Computers, Rigidity, and Moduli*. Princeton University Press, Princeton, NJ, 2004.S. Weinberger. What is... persistent homology.

*Notices Amer. Math. Soc.*, 58(1), 2011.Y. Yomdin and G. Comte.

*Tame Geometry with Applications in Smooth Analysis*, volume 1834 of*Lecture Notes in Mathematics*. Springer Verlag Berlin Heidelberg, 2004.

## Author information

### Authors and Affiliations

### Corresponding author

## Additional information

Communicated by Martin Sombra.

### Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Shmuel Weinberger: Partially supported by an NSF Grant.

## Rights and permissions

## About this article

### Cite this article

Weinberger, S. Interpolation, the Rudimentary Geometry of Spaces of Lipschitz Functions, and Geometric Complexity.
*Found Comput Math* **19**, 991–1011 (2019). https://doi.org/10.1007/s10208-019-09416-0

Received:

Revised:

Accepted:

Published:

Issue Date:

DOI: https://doi.org/10.1007/s10208-019-09416-0

### Keywords

- Interpolation
- Function space
- Persistent homology
- Homotopy
- Quantitative cobordism
- Embedding

### Mathematics Subject Classification

- Primary 68U05
- 57R75
- 41A10
- Secondary 55Q05
- 68T10