Showing 1–50 of 144 results for author: Davis, D

The cohomology of the connective spectra for {K}-theory revisited

Authors: Donald M. Davis, W. Stephen Wilson

Abstract: The stable mod 2 cohomologies of the spectra for connective real and complex K-theories are well known and easy to work with. However, the known bases are in terms of the anti-automorphism of Milnor basis elements. We offer simple bases in terms of admissible sequences of Steenrod operations that come from the Adem relations. In particular, the basis for the complex case is that you don't use any… ▽ More The stable mod 2 cohomologies of the spectra for connective real and complex K-theories are well known and easy to work with. However, the known bases are in terms of the anti-automorphism of Milnor basis elements. We offer simple bases in terms of admissible sequences of Steenrod operations that come from the Adem relations. In particular, the basis for the complex case is that you don't use any Steenrod operations in degree one or $2^n+1$, $n > 0$. △ Less

Submitted 18 April, 2024; v1 submitted 12 January, 2024; originally announced January 2024.

Comments: Feedback welcome

MSC Class: 55S10; 55R45; 55N15

Journal ref: New York Journal of Mathematics, volume 30, pages 513-520, 2024

Unitary representations of real groups and localization theory for Hodge modules

Authors: Dougal Davis, Kari Vilonen

Abstract: We prove a conjecture of Schmid and the second named author that the unitarity of a representation of a real reductive Lie group with real infinitesimal character can be read off from a canonical filtration, the Hodge filtration. Our proof rests on two main ingredients. The first is a wall crossing theory for mixed Hodge modules: the key result is that, in certain natural families, the Hodge filtr… ▽ More We prove a conjecture of Schmid and the second named author that the unitarity of a representation of a real reductive Lie group with real infinitesimal character can be read off from a canonical filtration, the Hodge filtration. Our proof rests on two main ingredients. The first is a wall crossing theory for mixed Hodge modules: the key result is that, in certain natural families, the Hodge filtration varies semi-continuously with jumps controlled by extension functors. The second ingredient is a Hodge-theoretic refinement of Beilinson-Bernstein localization: we show that the Hodge filtration of a mixed Hodge module on the flag variety satisfies the usual cohomology vanishing and global generation properties enjoyed by the underlying $\mathcal{D}$-module. As byproducts of our work, we obtain a version of Saito's Kodaira vanishing for twisted mixed Hodge modules, a calculation of the Hodge filtration on a certain object in category $\mathcal{O}$, and a host of new vanishing results for, for example, homogeneous vector bundles on flag varieties. △ Less

Submitted 22 September, 2023; originally announced September 2023.

Comments: 71 pages

Geodesic complexity of a cube

Authors: Donald M. Davis

Abstract: The topological (resp. geodesic) complexity of a topological (resp. metric) space is roughly the smallest number of continuous rules required to choose paths (resp. shortest paths) between any points of the space. We prove that the geodesic complexity of a cube exceeds its topological complexity by exactly 2. The proof involves a careful analysis of cut loci of the cube. The topological (resp. geodesic) complexity of a topological (resp. metric) space is roughly the smallest number of continuous rules required to choose paths (resp. shortest paths) between any points of the space. We prove that the geodesic complexity of a cube exceeds its topological complexity by exactly 2. The proof involves a careful analysis of cut loci of the cube. △ Less

Submitted 8 August, 2023; originally announced August 2023.

MSC Class: 53C22; 52B10; 55M30

A continuous $p$-adic action on the $K(2)$-local algebraic $K$-theory of $p$-adic complex $K$-theory

Authors: Daniel G. Davis

Abstract: Let $p$ be a prime, let $KU_p$ be $p$-complete complex $K$-theory, and let $\mathbb{Z}_p^\times$ denote the group of units in the $p$-adic integers. The $p$-adic Adams operations induce an action of the profinite group $\mathbb{Z}_p^\times$ on $KU_p$, and hence, on the algebraic $K$-theory spectrum $K(KU_p)$. For $p \geq 5$, we give an elementary construction of the continuous homotopy fixed point… ▽ More Let $p$ be a prime, let $KU_p$ be $p$-complete complex $K$-theory, and let $\mathbb{Z}_p^\times$ denote the group of units in the $p$-adic integers. The $p$-adic Adams operations induce an action of the profinite group $\mathbb{Z}_p^\times$ on $KU_p$, and hence, on the algebraic $K$-theory spectrum $K(KU_p)$. For $p \geq 5$, we give an elementary construction of the continuous homotopy fixed point spectrum $(L_{K(2)}K(KU_p))^{hG}$, where $K(2)$ is the second Morava $K$-theory and $G$ is any closed subgroup of $\mathbb{Z}_p^\times$. Also, for each $G$, we show that there is an associated strongly convergent homotopy fixed point spectral sequence whose $E_2$-term is given by Jannsen's continuous group cohomology, with $E_2^{s,\ast} = 0$, for all $s > 2$. This work is related to a conjecture of Ausoni and Rognes. △ Less

Submitted 3 August, 2023; originally announced August 2023.

Comments: 9 pages

MSC Class: 19D55; 55N15; 55P42; 55T25

Geodesic complexity of a tetrahedron

Authors: Donald M. Davis

Abstract: We prove that the geodesic complexity of a regular tetrahedron exceeds its topological complexity by 1 or 2. The proof involves a careful analysis of minimal geodesics on the tetrahedron. We prove that the geodesic complexity of a regular tetrahedron exceeds its topological complexity by 1 or 2. The proof involves a careful analysis of minimal geodesics on the tetrahedron. △ Less

Submitted 19 June, 2023; originally announced June 2023.

MSC Class: 53C22; 52B10; 55M30

Aiming towards the minimizers: fast convergence of SGD for overparametrized problems

Authors: Chaoyue Liu, Dmitriy Drusvyatskiy, Mikhail Belkin, Damek Davis, Yi-An Ma

Abstract: Modern machine learning paradigms, such as deep learning, occur in or close to the interpolation regime, wherein the number of model parameters is much larger than the number of data samples. In this work, we propose a regularity condition within the interpolation regime which endows the stochastic gradient method with the same worst-case iteration complexity as the deterministic gradient method,… ▽ More Modern machine learning paradigms, such as deep learning, occur in or close to the interpolation regime, wherein the number of model parameters is much larger than the number of data samples. In this work, we propose a regularity condition within the interpolation regime which endows the stochastic gradient method with the same worst-case iteration complexity as the deterministic gradient method, while using only a single sampled gradient (or a minibatch) in each iteration. In contrast, all existing guarantees require the stochastic gradient method to take small steps, thereby resulting in a much slower linear rate of convergence. Finally, we demonstrate that our condition holds when training sufficiently wide feedforward neural networks with a linear output layer. △ Less

Submitted 5 June, 2023; originally announced June 2023.

The horizontal chord set: to CIRM and back

Authors: Diana Davis, Serge Troubetzkoy

Abstract: We study the set of lengths of the horizontal chords of a continuous function. We give a new proof of Hopf's characterization of this set, and show that it implies that no matter which function we choose, at least half of the possible lengths occur. We prove several results about functions for which all the possible lengths occur. We study the set of lengths of the horizontal chords of a continuous function. We give a new proof of Hopf's characterization of this set, and show that it implies that no matter which function we choose, at least half of the possible lengths occur. We prove several results about functions for which all the possible lengths occur. △ Less

Submitted 24 May, 2023; v1 submitted 22 March, 2023; originally announced March 2023.

Comments: 7 pages, 4 figures, plus appendix. Comments welcome!

MSC Class: 26A15

Several homotopy fixed point spectral sequences in telescopically localized algebraic $K$-theory

Authors: Daniel G. Davis

Abstract: Let $n \geq 1$, $p$ a prime, and $T(n)$ any representative of the Bousfield class of the telescope $v_n^{-1}F(n)$ of a finite type $n$ complex. Also, let $E_n$ be the Lubin-Tate spectrum, $K(E_n)$ its algebraic $K$-theory spectrum, and $G_n$ the extended Morava stabilizer group, a profinite group. Motivated by an Ausoni-Rognes conjecture, we show that there are two spectral sequences \[{^{I}}\mspa… ▽ More Let $n \geq 1$, $p$ a prime, and $T(n)$ any representative of the Bousfield class of the telescope $v_n^{-1}F(n)$ of a finite type $n$ complex. Also, let $E_n$ be the Lubin-Tate spectrum, $K(E_n)$ its algebraic $K$-theory spectrum, and $G_n$ the extended Morava stabilizer group, a profinite group. Motivated by an Ausoni-Rognes conjecture, we show that there are two spectral sequences \[{^{I}}\mspace{-3mu}E_2^{s,t} \Longrightarrow π_{t-s}((L_{T(n+1)}K(E_n))^{hG_n}) \Longleftarrow {^{II}}\mspace{-2mu}E_2^{s,t}\] with common abutment $π_\ast(-)$ of the continuous homotopy fixed points of $L_{T(n+1)}K(E_n)$, where ${^{I}}\mspace{-3mu}E_2^{s,t}$ is continuous cohomology with coefficients in a certain tower of discrete $G_n$-modules. If the tower satisfies the Mittag-Leffler condition, then there are continuous cochain cohomology groups \[{^{I}}\mspace{-3mu}E_2^{\ast,\ast} \cong H^\ast_\mathrm{cts}(G_n, π_\ast(L_{T(n+1)}K(E_n))) \cong {^{II}}\mspace{-2mu}E_2^{\ast,\ast}.\] We isolate two hypotheses, the first of which is true when $(n,p) = (1,2)$, that imply $(L_{T(n+1)}K(E_n))^{hG_n} \simeq L_{T(n+1)}K(L_{K(n)}S^0)$. Also, we show that there is a spectral sequence \[H^s_\mathrm{cts}(G_n, π_t(K(E_n) \otimes T(n+1))) \Longrightarrow π_{t-s}((K(E_n) \otimes T(n+1))^{hG_n}).\] △ Less

Submitted 27 February, 2023; originally announced February 2023.

Comments: 18 pages, submitted for publication

MSC Class: 19D55 (Primary) 55P42; 55P43; 55T99 (Secondary)

Isomorphism classes of cut loci for a cube

Authors: Donald M Davis, Manyi Guo

Abstract: We prove that a face of a cube can be optimally partitioned into connected 193 sets on which the cut locus, or ridge tree, is constant up to isomorphism as a labeled graph. These are 60 connected open sets, curves bounding them, and intersection points of curves. Polynomial equations for the curves are provided. Sixteen pairs of sets support the same cut locus class. We present the 177 distinct cu… ▽ More We prove that a face of a cube can be optimally partitioned into connected 193 sets on which the cut locus, or ridge tree, is constant up to isomorphism as a labeled graph. These are 60 connected open sets, curves bounding them, and intersection points of curves. Polynomial equations for the curves are provided. Sixteen pairs of sets support the same cut locus class. We present the 177 distinct cut locus classes. △ Less

Submitted 10 February, 2023; v1 submitted 26 January, 2023; originally announced January 2023.

Comments: Minor corrections. This version submitted for publication

MSC Class: 52B10; 53C22; 52C30

Asymptotic normality and optimality in nonsmooth stochastic approximation

Authors: Damek Davis, Dmitriy Drusvyatskiy, Liwei Jiang

Abstract: In their seminal work, Polyak and Juditsky showed that stochastic approximation algorithms for solving smooth equations enjoy a central limit theorem. Moreover, it has since been argued that the asymptotic covariance of the method is best possible among any estimation procedure in a local minimax sense of Hájek and Le Cam. A long-standing open question in this line of work is whether similar guara… ▽ More In their seminal work, Polyak and Juditsky showed that stochastic approximation algorithms for solving smooth equations enjoy a central limit theorem. Moreover, it has since been argued that the asymptotic covariance of the method is best possible among any estimation procedure in a local minimax sense of Hájek and Le Cam. A long-standing open question in this line of work is whether similar guarantees hold for important non-smooth problems, such as stochastic nonlinear programming or stochastic variational inequalities. In this work, we show that this is indeed the case. △ Less

Submitted 16 January, 2023; originally announced January 2023.

Comments: The arxiv report arXiv:2108.11832 has been split into two parts. This is Part 2 of the original submission, augmented by a some new results and a reworked exposition

MSC Class: 65K05; 65K10; 90C15; 90C30; 90C06

A linearly convergent Gauss-Newton subgradient method for ill-conditioned problems

Authors: Damek Davis, Tao Jiang

Abstract: We analyze a preconditioned subgradient method for optimizing composite functions $h \circ c$, where $h$ is a locally Lipschitz function and $c$ is a smooth nonlinear map**. We prove that when $c$ satisfies a constant rank property and $h$ is semismooth and sharp on the image of $c$, the method converges linearly. In contrast to standard subgradient methods, its oracle complexity is invariant un… ▽ More We analyze a preconditioned subgradient method for optimizing composite functions $h \circ c$, where $h$ is a locally Lipschitz function and $c$ is a smooth nonlinear map**. We prove that when $c$ satisfies a constant rank property and $h$ is semismooth and sharp on the image of $c$, the method converges linearly. In contrast to standard subgradient methods, its oracle complexity is invariant under reparameterizations of $c$. △ Less

Submitted 26 December, 2022; originally announced December 2022.

The connective K-theory of the Eilenberg-MacLane space K(Z/p,2)

Authors: Donald M. Davis, W. Stephen Wilson

Abstract: We compute ku^*(K(Z/p,2)) and ku_*(K(Z/p,2)), the connective KU-cohomology and connective KU-homology groups of the mod-p Eilenberg-MacLane space K(Z/p,2), using the Adams spectral sequence. We obtain a striking interaction between h_0-extensions and exotic extensions. The mod-p connective KU-cohomology groups, computed elsewhere, are needed in order to establish higher differentials and exotic ex… ▽ More We compute ku^*(K(Z/p,2)) and ku_*(K(Z/p,2)), the connective KU-cohomology and connective KU-homology groups of the mod-p Eilenberg-MacLane space K(Z/p,2), using the Adams spectral sequence. We obtain a striking interaction between h_0-extensions and exotic extensions. The mod-p connective KU-cohomology groups, computed elsewhere, are needed in order to establish higher differentials and exotic extensions in the integral groups. △ Less

Submitted 20 February, 2023; v1 submitted 10 October, 2022; originally announced October 2022.

Comments: minor changes, submitted for publication

MSC Class: 55T15; 55N20; 55N15

The connective Morava K-theory of the second mod p Eilenberg-MacLane space

Authors: Donald M. Davis, Douglas C. Ravenel, W. Stephen Wilson

Abstract: We develop tools for computing the connective n-th Morava K-theory of spaces. Starting with a Universal Coefficient Theorem that computes the cohomology version from the homology version, we show that every step in the process of computing one is mirrored in the other and that this can be used to make computations. As our example, we compute the connective n-th Morava K-theory of the second mod p… ▽ More We develop tools for computing the connective n-th Morava K-theory of spaces. Starting with a Universal Coefficient Theorem that computes the cohomology version from the homology version, we show that every step in the process of computing one is mirrored in the other and that this can be used to make computations. As our example, we compute the connective n-th Morava K-theory of the second mod p Eilenberg-MacLane space. △ Less

Submitted 23 July, 2022; v1 submitted 28 June, 2022; originally announced June 2022.

Comments: minor change of title and one sentence added

MSC Class: 55N20; 55N35; 55P20; 55P43; 55Q51; 55T15; 55U20

Gorenstein Duality and Universal Coefficient Theorems

Authors: Donald M. Davis, J. P. C. Greenlees

Abstract: The paper describes a duality phenomenon for cohomology theories with the character of Gorenstein rings. For a connective cohomology theory with the p-local integers in degree 0, and coefficient ring R_* Gorenstein of shift 0, this states that for X with R_*(X) torsion, we have R^*(X)=Σ^a Hom( R_*(X), Z/p^{\infty}). A corresponding statement for modules over a commutative Gorenstein ring spectrum… ▽ More The paper describes a duality phenomenon for cohomology theories with the character of Gorenstein rings. For a connective cohomology theory with the p-local integers in degree 0, and coefficient ring R_* Gorenstein of shift 0, this states that for X with R_*(X) torsion, we have R^*(X)=Σ^a Hom( R_*(X), Z/p^{\infty}). A corresponding statement for modules over a commutative Gorenstein ring spectrum is also proved. [Minor typographical and bibliographic changes to the last version.] △ Less

Submitted 3 October, 2022; v1 submitted 22 June, 2022; originally announced June 2022.

MSC Class: 55U20; 55U30; 55N20; 55P43; 18G15; 13H10

Hodge Filtrations on Tempered Hodge Modules

Authors: Dougal Davis, Kari Vilonen

Abstract: We show that the Hodge filtration of a tempered Hodge module is generated by the lowest piece of its Hodge filtration. As a consequence, we prove the main conjecture of [SV] in the special case of tempered representations of real reductive Lie groups. We show that the Hodge filtration of a tempered Hodge module is generated by the lowest piece of its Hodge filtration. As a consequence, we prove the main conjecture of [SV] in the special case of tempered representations of real reductive Lie groups. △ Less

Submitted 8 March, 2023; v1 submitted 17 June, 2022; originally announced June 2022.

Comments: 9 pages. Minor changes to improve the clarity of the exposition (v2). Added a conceptual overview to the introduction (v3)

MSC Class: 14F10; 22E46; 32S35

Duality in BP<n> (co)homology

Authors: Donald M. Davis

Abstract: Let E=BP<n> denote the Johnson-Wilson spectrum, localized at p. It is proved that if E_*(X) is locally finite, then there is an isomorphism of right E_*-modules E^*(X) = (E_*(Sigma^{D+n+1}X))^V, where D=Sum |v_i| and M^V=Hom(M,Q/Z) is the Pontryagin dual. This result was motivated by work of the author and W.S.Wilson regarding the 2-local ku-homology and -cohomology of the Eilenberg-MacLane space… ▽ More Let E=BP<n> denote the Johnson-Wilson spectrum, localized at p. It is proved that if E_*(X) is locally finite, then there is an isomorphism of right E_*-modules E^*(X) = (E_*(Sigma^{D+n+1}X))^V, where D=Sum |v_i| and M^V=Hom(M,Q/Z) is the Pontryagin dual. This result was motivated by work of the author and W.S.Wilson regarding the 2-local ku-homology and -cohomology of the Eilenberg-MacLane space K(Z/2,2). △ Less

Submitted 30 May, 2022; v1 submitted 12 May, 2022; originally announced May 2022.

Comments: Minor modifications

MSC Class: 55U20; 55U30; 55N20; 18G15

A nearly linearly convergent first-order method for nonsmooth functions with quadratic growth

Authors: Damek Davis, Liwei Jiang

Abstract: Classical results show that gradient descent converges linearly to minimizers of smooth strongly convex functions. A natural question is whether there exists a locally nearly linearly convergent method for nonsmooth functions with quadratic growth. This work designs such a method for a wide class of nonsmooth and nonconvex locally Lipschitz functions, including max-of-smooth, Shapiro's decomposabl… ▽ More Classical results show that gradient descent converges linearly to minimizers of smooth strongly convex functions. A natural question is whether there exists a locally nearly linearly convergent method for nonsmooth functions with quadratic growth. This work designs such a method for a wide class of nonsmooth and nonconvex locally Lipschitz functions, including max-of-smooth, Shapiro's decomposable class, and generic semialgebraic functions. The algorithm is parameter-free and derives from Goldstein's conceptual subgradient method. △ Less

Submitted 17 July, 2023; v1 submitted 29 April, 2022; originally announced May 2022.

Comments: 75 pages, 8 figures. Revised intro. New experiments. Simplified proofs. PyTorch code available here: https://github.com/COR-OPT/ntd.py/

MSC Class: 65K05; 65K10; 90C15; 90C30

Mixed Hodge modules and real groups

Authors: Dougal Davis, Kari Vilonen

Abstract: Let $G$ be a complex reductive group, $θ\colon G \to G$ an involution, and $K = G^θ$. In arXiv:1206.5547, W. Schmid and the second named author proposed a program to study unitary representations of the corresponding real form $G_\mathbb{R}$ using $K$-equivariant twisted mixed Hodge modules on the flag variety of $G$ and their polarizations. In this paper, we make the first significant steps towar… ▽ More Let $G$ be a complex reductive group, $θ\colon G \to G$ an involution, and $K = G^θ$. In arXiv:1206.5547, W. Schmid and the second named author proposed a program to study unitary representations of the corresponding real form $G_\mathbb{R}$ using $K$-equivariant twisted mixed Hodge modules on the flag variety of $G$ and their polarizations. In this paper, we make the first significant steps towards implementing this program. Our first main result gives an explicit combinatorial formula for the Hodge numbers appearing in the composition series of a standard module in terms of the Lusztig-Vogan polynomials. Our second main result is a polarized version of the Jantzen conjecture, stating that the Jantzen forms on the composition factors are polarizations of the underlying Hodge modules. Our third main result states that, for regular Beilinson-Bernstein data, the minimal $K$-type of an irreducible Harish-Chandra module lies in the lowest piece of the Hodge filtration of the corresponding Hodge module. An immediate consequence of our results is a Hodge-theoretic proof of the signature multiplicity formula of arXiv:1212.2192, which was the inspiration for this work. △ Less

Submitted 7 June, 2023; v1 submitted 17 February, 2022; originally announced February 2022.

Comments: 58 pages, including one appendix. v2: Added some references, minor updates to the introduction

MSC Class: 14F10; 22E46; 32S35

A superlinearly convergent subgradient method for sharp semismooth problems

Authors: Vasileios Charisopoulos, Damek Davis

Abstract: Subgradient methods comprise a fundamental class of nonsmooth optimization algorithms. Classical results show that certain subgradient methods converge sublinearly for general Lipschitz convex functions and converge linearly for convex functions that grow sharply away from solutions. Recent work has moreover extended these results to certain nonconvex problems. In this work we seek to improve the… ▽ More Subgradient methods comprise a fundamental class of nonsmooth optimization algorithms. Classical results show that certain subgradient methods converge sublinearly for general Lipschitz convex functions and converge linearly for convex functions that grow sharply away from solutions. Recent work has moreover extended these results to certain nonconvex problems. In this work we seek to improve the complexity of these algorithms, asking: is it possible to design a superlinearly convergent subgradient method? We provide a positive answer to this question for a broad class of sharp semismooth functions. △ Less

Submitted 12 January, 2022; originally announced January 2022.

Comments: 48 pages, 7 figures

MSC Class: 65K05; 65K10; 90C06; 90C30

A gradient sampling method with complexity guarantees for Lipschitz functions in high and low dimensions

Authors: Damek Davis, Dmitriy Drusvyatskiy, Yin Tat Lee, Swati Padmanabhan, Guanghao Ye

Abstract: Zhang et al. introduced a novel modification of Goldstein's classical subgradient method, with an efficiency guarantee of $O(\varepsilon^{-4})$ for minimizing Lipschitz functions. Their work, however, makes use of a nonstandard subgradient oracle model and requires the function to be directionally differentiable. In this paper, we show that both of these assumptions can be dropped by simply adding… ▽ More Zhang et al. introduced a novel modification of Goldstein's classical subgradient method, with an efficiency guarantee of $O(\varepsilon^{-4})$ for minimizing Lipschitz functions. Their work, however, makes use of a nonstandard subgradient oracle model and requires the function to be directionally differentiable. In this paper, we show that both of these assumptions can be dropped by simply adding a small random perturbation in each step of their algorithm. The resulting method works on any Lipschitz function whose value and gradient can be evaluated at points of differentiability. We additionally present a new cutting plane algorithm that achieves better efficiency in low dimensions: $O(d\varepsilon^{-3})$ for Lipschitz functions and $O(d\varepsilon^{-2})$ for those that are weakly convex. △ Less

Submitted 15 February, 2022; v1 submitted 13 December, 2021; originally announced December 2021.

Comments: 14 pages

MSC Class: 65K05; 65K10; 90C15; 90C30

Hyperbolic Staircases: Periodic Paths on $2g+1$-gons

Authors: Mei Rose Connor, Diana Davis, Paige Helms, Michael Kielstra, Samuel Lelièvre, Zachary Steinberg, Chenyang Sun

Abstract: The study of polygonal billiards, particularly those in the regular pentagon, has been the subject of two recent papers. One of these papers approaches the problem of discovering the periodic trajectories on the pentagon by identifying slopes of periodic directions with points in the Poincaré disk generated by hyperbolic isometric transformations. The other approach, coming from the other paper, t… ▽ More The study of polygonal billiards, particularly those in the regular pentagon, has been the subject of two recent papers. One of these papers approaches the problem of discovering the periodic trajectories on the pentagon by identifying slopes of periodic directions with points in the Poincaré disk generated by hyperbolic isometric transformations. The other approach, coming from the other paper, transforms the double pentagon into a rectilinear translation surface called the 'golden L', where periodic directions are generated by a set of matrices associated with this surface in a special way. We connect and unify these two approaches, and use our unification of these results to generalize them to arbitrary $2g+1$-sided regular polygons. △ Less

Submitted 29 November, 2021; v1 submitted 27 November, 2021; originally announced November 2021.

Comments: 16 pages, 12 figures

MSC Class: 37E35

Clustering a Mixture of Gaussians with Unknown Covariance

Authors: Damek Davis, Mateo Díaz, Kaizheng Wang

Abstract: We investigate a clustering problem with data from a mixture of Gaussians that share a common but unknown, and potentially ill-conditioned, covariance matrix. We start by considering Gaussian mixtures with two equally-sized components and derive a Max-Cut integer program based on maximum likelihood estimation. We prove its solutions achieve the optimal misclassification rate when the number of sam… ▽ More We investigate a clustering problem with data from a mixture of Gaussians that share a common but unknown, and potentially ill-conditioned, covariance matrix. We start by considering Gaussian mixtures with two equally-sized components and derive a Max-Cut integer program based on maximum likelihood estimation. We prove its solutions achieve the optimal misclassification rate when the number of samples grows linearly in the dimension, up to a logarithmic factor. However, solving the Max-cut problem appears to be computationally intractable. To overcome this, we develop an efficient spectral algorithm that attains the optimal rate but requires a quadratic sample size. Although this sample complexity is worse than that of the Max-cut problem, we conjecture that no polynomial-time method can perform better. Furthermore, we gather numerical and theoretical evidence that supports the existence of a statistical-computational gap. Finally, we generalize the Max-Cut program to a $k$-means program that handles multi-component mixtures with possibly unequal weights. It enjoys similar optimality guarantees for mixtures of distributions that satisfy a transportation-cost inequality, encompassing Gaussian and strongly log-concave distributions. △ Less

Submitted 29 November, 2021; v1 submitted 4 October, 2021; originally announced October 2021.

Comments: 89 pages

MSC Class: 62H30; 62H12; 62H05

Active manifolds, stratifications, and convergence to local minima in nonsmooth optimization

Authors: Damek Davis, Dmitriy Drusvyatskiy, Liwei Jiang

Abstract: We show that the subgradient method converges only to local minimizers when applied to generic Lipschitz continuous and subdifferentially regular functions that are definable in an o-minimal structure. At a high level, the argument we present is appealingly transparent: we interpret the nonsmooth dynamics as an approximate Riemannian gradient method on a certain distinguished submanifold that capt… ▽ More We show that the subgradient method converges only to local minimizers when applied to generic Lipschitz continuous and subdifferentially regular functions that are definable in an o-minimal structure. At a high level, the argument we present is appealingly transparent: we interpret the nonsmooth dynamics as an approximate Riemannian gradient method on a certain distinguished submanifold that captures the nonsmooth activity of the function. In the process, we develop new regularity conditions in nonsmooth analysis that parallel the stratification conditions of Whitney, Kuo, and Verdier and extend stochastic processes techniques of Pemantle. △ Less

Submitted 9 January, 2023; v1 submitted 26 August, 2021; originally announced August 2021.

Comments: Version 1 of the arxiv report has been split into two parts. Version 2 of the arxiv report is Part 1 of the original submission. Part 2 will appear as a separate arxiv submission

MSC Class: 65K05; 65K10; 90C15; 90C30; 90C06

The mod-2 connected KU-homology of the Eilenberg-MacLane space K(Z/2,2)

Authors: Donald M. Davis, W. Stephen Wilson

Abstract: We compute the mod-2 connected KU-homology of the Eilenberg-MacLane space K(Z/2,2), using a novel Adams spectral sequence analysis. We compute the mod-2 connected KU-homology of the Eilenberg-MacLane space K(Z/2,2), using a novel Adams spectral sequence analysis. △ Less

Submitted 24 June, 2021; originally announced June 2021.

MSC Class: 55T15; 55P20; 55N15

Esca** strict saddle points of the Moreau envelope in nonsmooth optimization

Authors: Damek Davis, Mateo Díaz, Dmitriy Drusvyatskiy

Abstract: Recent work has shown that stochastically perturbed gradient methods can efficiently escape strict saddle points of smooth functions. We extend this body of work to nonsmooth optimization, by analyzing an inexact analogue of a stochastically perturbed gradient method applied to the Moreau envelope. The main conclusion is that a variety of algorithms for nonsmooth optimization can escape strict sad… ▽ More Recent work has shown that stochastically perturbed gradient methods can efficiently escape strict saddle points of smooth functions. We extend this body of work to nonsmooth optimization, by analyzing an inexact analogue of a stochastically perturbed gradient method applied to the Moreau envelope. The main conclusion is that a variety of algorithms for nonsmooth optimization can escape strict saddle points of the Moreau envelope at a controlled rate. The main technical insight is that typical algorithms applied to the proximal subproblem yield directions that approximate the gradient of the Moreau envelope in relative terms. △ Less

Submitted 17 June, 2021; originally announced June 2021.

Comments: 29 pages, 1 figure

MSC Class: 65K05; 65K10; 90C15; 90C30; 90C06

Graph-theoretic algorithms for Kolmogorov operators: Approximating solutions and their gradients in elliptic and parabolic problems on manifolds

Authors: Andrew D. Davis, Dimitrios Giannakis

Abstract: We employ kernel-based approaches that use samples from a probability distribution to approximate a Kolmogorov operator on a manifold. The self-tuning variable-bandwidth kernel method [Berry & Harlim, Appl. Comput. Harmon. Anal., 40(1):68--96, 2016] computes a large, sparse matrix that approximates the differential operator. Here, we use the eigendecomposition of the discretization to (i) invert t… ▽ More We employ kernel-based approaches that use samples from a probability distribution to approximate a Kolmogorov operator on a manifold. The self-tuning variable-bandwidth kernel method [Berry & Harlim, Appl. Comput. Harmon. Anal., 40(1):68--96, 2016] computes a large, sparse matrix that approximates the differential operator. Here, we use the eigendecomposition of the discretization to (i) invert the operator, solving a differential equation, and (ii) represent gradient vector fields on the manifold. These methods only require samples from the underlying distribution and, therefore, can be applied in high dimensions or on geometrically complex manifolds when spatial discretizations are not available. We also employ an efficient $k$-$d$ tree algorithm to compute the sparse kernel matrix, which is a computational bottleneck. △ Less

Submitted 19 April, 2022; v1 submitted 30 April, 2021; originally announced April 2021.

Conservative and semismooth derivatives are equivalent for semialgebraic maps

Authors: Damek Davis, Dmitriy Drusvyatskiy

Abstract: Subgradient and Newton algorithms for nonsmooth optimization require generalized derivatives to satisfy subtle approximation properties: conservativity for the former and semismoothness for the latter. Though these two properties originate in entirely different contexts, we show that in the semi-algebraic setting they are equivalent. Both properties for a generalized derivative simply require it t… ▽ More Subgradient and Newton algorithms for nonsmooth optimization require generalized derivatives to satisfy subtle approximation properties: conservativity for the former and semismoothness for the latter. Though these two properties originate in entirely different contexts, we show that in the semi-algebraic setting they are equivalent. Both properties for a generalized derivative simply require it to coincide with the standard directional derivative on the tangent spaces of some partition of the domain into smooth manifolds. An appealing byproduct is a new short proof that semi-algebraic maps are semismooth relative to the Clarke Jacobian. △ Less

Submitted 16 February, 2021; originally announced February 2021.

Comments: 12 pages

MSC Class: Primary: 49J53; 49J52; Secondary: 32B20; 14P15

A computational reduction for many base cases in profinite telescopic algebraic $K$-theory

Authors: Daniel G. Davis

Abstract: For primes $p\geq 5 $, $K(KU_p)$ -- the algebraic $K$-theory spectrum of $(KU)^{\wedge}_p$, Morava $K$-theory $K(1)$, and Smith-Toda complex $V(1)$, Ausoni and Rognes conjectured (alongside related conjectures) that $L_{K(1)}S^0 \mspace{-1.5mu}\xrightarrow{\mspace{-2mu}\text{unit} \, i}~\mspace{-7mu}(KU)^{\wedge}_p$ induces a map… ▽ More For primes $p\geq 5 $, $K(KU_p)$ -- the algebraic $K$-theory spectrum of $(KU)^{\wedge}_p$, Morava $K$-theory $K(1)$, and Smith-Toda complex $V(1)$, Ausoni and Rognes conjectured (alongside related conjectures) that $L_{K(1)}S^0 \mspace{-1.5mu}\xrightarrow{\mspace{-2mu}\text{unit} \, i}~\mspace{-7mu}(KU)^{\wedge}_p$ induces a map $K(L_{K(1)}S^0) \wedge v_2^{-1}V(1) \to K(KU_p)^{h\mathbb{Z}^\times_p} \wedge v_2^{-1}V(1)$ that is an equivalence. Since the definition of this map is not well understood, we consider $K(L_{K(1)}S^0) \wedge v_2^{-1}V(1) \to (K(KU_p) \wedge v_2^{-1}V(1))^{h\mathbb{Z}^\times_p}$, which is induced by $i$ and also should be an equivalence. We show that for any closed $G < \mathbb{Z}^\times_p$, $π_\ast((K(KU_p) \wedge v_2^{-1}V(1))^{hG})$ is a direct sum of two pieces given by (co)invariants and a coinduced module, for $K(KU_p)_\ast(V(1))[v_2^{-1}]$. When $G = \mathbb{Z}^\times_p$, the direct sum is, conjecturally, $K(L_{K(1)}S^0)_\ast(V(1))[v_2^{-1}]$ and, by using $K(L_p)_\ast(V(1))[v_2^{-1}]$, where $L_p = ((KU)^{\wedge}_p)^{h\mathbb{Z}/((p-1)\mathbb{Z})}$, the summands simplify. The Ausoni-Rognes conjecture suggests that in \[(-)^{h\mathbb{Z}^\times_p} \wedge v_2^{-1}V(1) \simeq (K(KU_p) \wedge v_2^{-1}V(1))^{h\mathbb{Z}^\times_p},\] $K(KU_p)$ fills in the blank; we show that for any $G$, the blank can be filled by $(K(KU_p))^\mathrm{dis}_\mathcal{O}$, a discrete $\mathbb{Z}^\times_p$-spectrum built out of $K(KU_p)$. △ Less

Submitted 9 April, 2021; v1 submitted 26 January, 2021; originally announced January 2021.

Comments: 20 pages; resubmitted for publication; Section 1.1 is new; updated several references

MSC Class: 19D55; 55N15; 55P42; 55T25

Assessing congressional districting in Maine and New Hampshire

Authors: Sara Asgari, Quinn Basewitz, Ethan Bergmann, Jackson Brogsol, Nathaniel Cox, Diana Davis, Martina Kampel, Becca Keating, Katie Knox, Angus Lam, Jorge Lopez-Nava, Jennifer Paige, Nathan Pitock, Victoria Song, Dylan Torrance

Abstract: We use voting precinct and election data to analyze the political geography of New Hampshire and Maine. We find that the location of dividing line between Congressional districts in both states are significantly different than what we would expect, which we argue is likely due to incumbent gerrymandering. We also discuss the limitations of classical fairness measures for plans with only two distri… ▽ More We use voting precinct and election data to analyze the political geography of New Hampshire and Maine. We find that the location of dividing line between Congressional districts in both states are significantly different than what we would expect, which we argue is likely due to incumbent gerrymandering. We also discuss the limitations of classical fairness measures for plans with only two districts. △ Less

Submitted 12 November, 2020; originally announced November 2020.

Comments: 19 pages, 12 figures; comments and further research questions welcome

MSC Class: 05C90

Stiefel-Whitney classes and immersions of orientable and Spin manifolds

Authors: Donald M. Davis, W. Stephen Wilson

Abstract: We determine a nice simple formula for the largest Euclidean space for which there is an orientable n-manifold with a nonimmersion detected by Stiefel-Whitney classes. For Spin manifolds, we prove the analogue of the upper bound and establish the complete answer for n<24 and n=33,34. Results similar to many of these were obtained some 50 years ago, but in a much less tractable form. The sharp resu… ▽ More We determine a nice simple formula for the largest Euclidean space for which there is an orientable n-manifold with a nonimmersion detected by Stiefel-Whitney classes. For Spin manifolds, we prove the analogue of the upper bound and establish the complete answer for n<24 and n=33,34. Results similar to many of these were obtained some 50 years ago, but in a much less tractable form. The sharp results for Spin manifolds require detailed calculations of ko-homology groups of mod-2 Eilenberg-MacLane spaces. △ Less

Submitted 22 March, 2021; v1 submitted 2 November, 2020; originally announced November 2020.

Comments: Some new results. Some minor corrections. Addition of coauthor

MSC Class: 57R42; 57R20; 55N22

Geodesic complexity for non-geodesic spaces

Authors: Donald M. Davis

Abstract: We define the notion of near geodesic between points of a metric space when no geodesic exists, and use this to extend Recio-Mitter's notion of geodesic complexity to non-geodesic spaces. This has potential application to topological robotics. We determine explicit near geodesics and geodesic complexity in a variety of cases. We define the notion of near geodesic between points of a metric space when no geodesic exists, and use this to extend Recio-Mitter's notion of geodesic complexity to non-geodesic spaces. This has potential application to topological robotics. We determine explicit near geodesics and geodesic complexity in a variety of cases. △ Less

Submitted 27 May, 2021; v1 submitted 24 September, 2020; originally announced September 2020.

Comments: One major correction

MSC Class: 53C22; 55R80; 55M30; 68T40

Two robots moving geodesically on a tree

Authors: Donald M. Davis, Michael Harrison, David Recio-Mitter

Abstract: We study the geodesic complexity of the ordered and unordered configuration spaces of graphs in both the $\ell_1$ and $\ell_2$ metrics. We determine the geodesic complexity of the ordered two-point $\varepsilon$-configuration space of any star graph in both the $\ell_1$ and $\ell_2$ metrics and of the unordered two-point configuration space of any tree in the $\ell_1$ metric, by finding explicit g… ▽ More We study the geodesic complexity of the ordered and unordered configuration spaces of graphs in both the $\ell_1$ and $\ell_2$ metrics. We determine the geodesic complexity of the ordered two-point $\varepsilon$-configuration space of any star graph in both the $\ell_1$ and $\ell_2$ metrics and of the unordered two-point configuration space of any tree in the $\ell_1$ metric, by finding explicit geodesics from any pair to any other pair, and arranging them into a minimal number of continuously-varying families. In each case the geodesic complexity matches the known value of the topological complexity. △ Less

Submitted 25 June, 2020; originally announced June 2020.

Comments: 20 pages, 17 figures

MSC Class: 53C22; 55R80; 55M30; 68T40

Journal ref: Algebr. Geom. Topol. 22 (2022) 785-814

Rate-optimal refinement strategies for local approximation MCMC

Authors: Andrew D. Davis, Youssef Marzouk, Aaron Smith, Natesh Pillai

Abstract: Many Bayesian inference problems involve target distributions whose density functions are computationally expensive to evaluate. Replacing the target density with a local approximation based on a small number of carefully chosen density evaluations can significantly reduce the computational expense of Markov chain Monte Carlo (MCMC) sampling. Moreover, continual refinement of the local approximati… ▽ More Many Bayesian inference problems involve target distributions whose density functions are computationally expensive to evaluate. Replacing the target density with a local approximation based on a small number of carefully chosen density evaluations can significantly reduce the computational expense of Markov chain Monte Carlo (MCMC) sampling. Moreover, continual refinement of the local approximation can guarantee asymptotically exact sampling. We devise a new strategy for balancing the decay rate of the bias due to the approximation with that of the MCMC variance. We prove that the error of the resulting local approximation MCMC (LA-MCMC) algorithm decays at roughly the expected $1/\sqrt{T}$ rate, and we demonstrate this rate numerically. We also introduce an algorithmic parameter that guarantees convergence given very weak tail bounds, significantly strengthening previous convergence results. Finally, we apply LA-MCMC to a computationally intensive Bayesian inverse problem arising in groundwater hydrology. △ Less

Submitted 11 July, 2022; v1 submitted 29 May, 2020; originally announced June 2020.

Comments: 32 pages, 17 figures

MSC Class: 62C10; 62F15 ACM Class: F.2.1; G.3; G.4

A construction of some objects in many base cases of an Ausoni-Rognes conjecture

Authors: Daniel G. Davis

Abstract: Let $p$ be a prime, $n \geq 1$, $K(n)$ the $n$th Morava $K$-theory spectrum, $\mathbb{G}_n$ the extended Morava stabilizer group, and $K(A)$ the algebraic $K$-theory spectrum of a commutative $S$-algebra $A$. For a type $n+1$ complex $V_n$, Ausoni and Rognes conjectured that (a) the unit map $i_n: L_{K(n)}(S^0) \to E_n$ from the $K(n)$-local sphere to the Lubin-Tate spectrum induces a map \[K(L_{K… ▽ More Let $p$ be a prime, $n \geq 1$, $K(n)$ the $n$th Morava $K$-theory spectrum, $\mathbb{G}_n$ the extended Morava stabilizer group, and $K(A)$ the algebraic $K$-theory spectrum of a commutative $S$-algebra $A$. For a type $n+1$ complex $V_n$, Ausoni and Rognes conjectured that (a) the unit map $i_n: L_{K(n)}(S^0) \to E_n$ from the $K(n)$-local sphere to the Lubin-Tate spectrum induces a map \[K(L_{K(n)}(S^0)) \wedge v_{n+1}^{-1}V_n \to (K(E_n))^{h\mathbb{G}_n} \wedge v_{n+1}^{-1}V_n\] that is a weak equivalence, where (b) since $\mathbb{G}_n$ is profinite, $(K(E_n))^{h\mathbb{G}_n}$ denotes a continuous homotopy fixed point spectrum, and (c) $π_\ast(-)$ of the target of the above map is the abutment of a homotopy fixed point spectral sequence. For $n = 1$, $p \geq 5$, and $V_1 = V(1)$, we give a way to realize the above map and (c), by proving that $i_1$ induces a map \[K(L_{K(1)}(S^0)) \wedge v_{2}^{-1}V_1 \to (K(E_1) \wedge v_{2}^{-1}V_1)^{h\mathbb{G}_1},\] where the target of this map is a continuous homotopy fixed point spectrum, with an associated homotopy fixed point spectral sequence. Also, we prove that there is an equivalence \[(K(E_1) \wedge v_{2}^{-1}V_1)^{h\mathbb{G}_1} \simeq (K(E_1))^{\widetilde{h}\mathbb{G}_1} \wedge v_2^{-1}V_1,\] where $(K(E_1))^{\widetilde{h}\mathbb{G}_1}$ is the homotopy fixed points with $\mathbb{G}_1$ regarded as a discrete group. △ Less

Submitted 11 December, 2020; v1 submitted 8 May, 2020; originally announced May 2020.

Comments: 32 pages; submitted for publication; updated description of status of $(K(E_n))^{h\mathbb{G}_n}$ by adding Remark 1.5 and modifying the paragraph that precedes it (and removing description of status from abstract); sharpened the writing in various places

MSC Class: 19D55; 55N15; 55P42; 55T99

On subregular slices of the elliptic Grothendieck-Springer resolution

Authors: Dougal Davis

Abstract: In \cite{davis19}, the author constructed an elliptic version of the Grothendieck-Springer resolution for the stack $\mathrm{Bun}_G$ of principal bundles under a simply connected simple group $G$ on an elliptic curve $E$. This is a simultaneous log resolution of a map from $\mathrm{Bun}_G$ to the union of the coarse moduli space of semistable $G$-bundles and a single stacky point. In this paper, w… ▽ More In \cite{davis19}, the author constructed an elliptic version of the Grothendieck-Springer resolution for the stack $\mathrm{Bun}_G$ of principal bundles under a simply connected simple group $G$ on an elliptic curve $E$. This is a simultaneous log resolution of a map from $\mathrm{Bun}_G$ to the union of the coarse moduli space of semistable $G$-bundles and a single stacky point. In this paper, we study singularities, resolutions and deformations coming from subregular slices of this elliptic Grothendieck-Springer resolution. More precisely, we construct explicit slices of $\mathrm{Bun}_G$ through all subregular unstable bundles, for every $G$. For $G \neq SL_2$, we describe the pullbacks of the elliptic Grothendieck-Springer resolution to these slices as concrete varieties, extending and refining earlier work of I. Grojnowski and N. Shepherd-Barron, who related these varieties to del Pezzo surfaces in type $E$. We use the resolutions to identify the singularities of the unstable locus of the subregular slices, and prove that that the extended coarse moduli space map gives deformations that are miniversal among torus-equivariant deformations with appropriate weights. △ Less

Submitted 27 April, 2020; originally announced April 2020.

Comments: 50 pages, 4 tables. Comments welcome!

MSC Class: 14H60 (primary); 14J17; 14L35; 14L40 (secondary)

Stochastic optimization over proximally smooth sets

Authors: Damek Davis, Dmitriy Drusvyatskiy, Zhan Shi

Abstract: We introduce a class of stochastic algorithms for minimizing weakly convex functions over proximally smooth sets. As their main building blocks, the algorithms use simplified models of the objective function and the constraint set, along with a retraction operation to restore feasibility. All the proposed methods come equipped with a finite time efficiency guarantee in terms of a natural stationar… ▽ More We introduce a class of stochastic algorithms for minimizing weakly convex functions over proximally smooth sets. As their main building blocks, the algorithms use simplified models of the objective function and the constraint set, along with a retraction operation to restore feasibility. All the proposed methods come equipped with a finite time efficiency guarantee in terms of a natural stationarity measure. We discuss consequences for nonsmooth optimization over smooth manifolds and over sets cut out by weakly-convex inequalities. △ Less

Submitted 14 February, 2020; originally announced February 2020.

MSC Class: 65K05; 65K10; 90C15; 90C30

Geodesics in the configuration spaces of two points in R^n

Authors: Donald M Davis

Abstract: We determine explicit formulas for geodesics (in the Euclidean metric) in the configuration space of ordered pairs (x,x') of points in R^n which satisfy d(x,x')>=epsilon. We interpret this as two or three (depending on the parity of n) geodesic motion-planning rules for this configuration space. In the associated unordered configuration space, we need not prescribe that the points stay apart by ep… ▽ More We determine explicit formulas for geodesics (in the Euclidean metric) in the configuration space of ordered pairs (x,x') of points in R^n which satisfy d(x,x')>=epsilon. We interpret this as two or three (depending on the parity of n) geodesic motion-planning rules for this configuration space. In the associated unordered configuration space, we need not prescribe that the points stay apart by epsilon. For this space, with a Euclidean metric, we show that geodesic motion-planning rules correspond to ordinary motion-planning rules on RP^{n-1}. △ Less

Submitted 4 July, 2020; v1 submitted 3 January, 2020; originally announced January 2020.

Comments: Replacement of paper which just did the ordered case

MSC Class: 53C22; 55R80; 55M30; 68T40

The geodesic complexity of n-dimensional Klein bottles

Authors: Donald M. Davis, David Recio-Mitter

Abstract: The geodesic complexity of a metric space X is the smallest k for which there is a partition of X x X into ENRs E_0,...,E_k on each of which there is a continuous choice of minimal geodesic sigma(x_0,x_1) from x_0 to x_1. We prove that the geodesic complexity of an n-dimensional Klein bottle equals 2n. Its topological complexity remains unknown for n>2. The geodesic complexity of a metric space X is the smallest k for which there is a partition of X x X into ENRs E_0,...,E_k on each of which there is a continuous choice of minimal geodesic sigma(x_0,x_1) from x_0 to x_1. We prove that the geodesic complexity of an n-dimensional Klein bottle equals 2n. Its topological complexity remains unknown for n>2. △ Less

Submitted 16 December, 2019; originally announced December 2019.

MSC Class: 53N22; 55M30; 68T40

Proximal methods avoid active strict saddles of weakly convex functions

Authors: Damek Davis, Dmitriy Drusvyatskiy

Abstract: We introduce a geometrically transparent strict saddle property for nonsmooth functions. This property guarantees that simple proximal algorithms on weakly convex problems converge only to local minimizers, when randomly initialized. We argue that the strict saddle property may be a realistic assumption in applications, since it provably holds for generic semi-algebraic optimization problems. We introduce a geometrically transparent strict saddle property for nonsmooth functions. This property guarantees that simple proximal algorithms on weakly convex problems converge only to local minimizers, when randomly initialized. We argue that the strict saddle property may be a realistic assumption in applications, since it provably holds for generic semi-algebraic optimization problems. △ Less

Submitted 16 February, 2021; v1 submitted 15 December, 2019; originally announced December 2019.

Comments: 43 pages, 2 figures

MSC Class: 65K05; 65K10; 90C15; 90C30; 90C06

The elliptic Grothendieck-Springer resolution as a simultaneous log resolution of algebraic stacks

Authors: Dougal Davis

Abstract: Let $G$ be a simply connected simple algebraic group. The classical multiplicative and additive Grothendieck-Springer resolutions are simultaneous resolutions of singularities for the maps from $G$ and its Lie algebra to their invariant theory quotients by the conjugation action of $G$. In this paper, we construct an elliptic version of the Grothendieck-Springer resolution, which is a simultaneous… ▽ More Let $G$ be a simply connected simple algebraic group. The classical multiplicative and additive Grothendieck-Springer resolutions are simultaneous resolutions of singularities for the maps from $G$ and its Lie algebra to their invariant theory quotients by the conjugation action of $G$. In this paper, we construct an elliptic version of the Grothendieck-Springer resolution, which is a simultaneous log resolution of a family whose total space is the stack of principal $G$-bundles on an elliptic curve. Our construction extends a well-known simultaneous resolution of the coarse moduli space map for semistable principal bundles to the stack of all principal bundles. We also prove elliptic versions of the Chevalley isomorphism and the Kostant and Steinberg section theorems, on which our construction relies. △ Less

Submitted 28 April, 2020; v1 submitted 12 August, 2019; originally announced August 2019.

Comments: 48 pages. Added a few extra remarks and clarified the construction of the fundamental diagram, in response to reviewer's comments

MSC Class: 14H60 (primary); 14E15; 14L35; 14L40 (secondary)

A lift from group cohomology to spectra for trivial profinite actions

Authors: Daniel G. Davis

Abstract: Let $G$ be a profinite group, $X$ a discrete $G$-spectrum with trivial action, and $X^{hG}$ the continuous homotopy fixed points. For any $N \trianglelefteq_o G$ ("$o$" for open), $X = X^N$ is a $G/N$-spectrum with trivial action. We construct a zigzag $\text{colim}\,_N \,X^{hG/N} \buildrelΦ\over\longrightarrow \text{colim}\,_N \,(X^{hN})^{hG/N} \buildrelΨ\over\longleftarrow X^{hG}$, where $Ψ$ is… ▽ More Let $G$ be a profinite group, $X$ a discrete $G$-spectrum with trivial action, and $X^{hG}$ the continuous homotopy fixed points. For any $N \trianglelefteq_o G$ ("$o$" for open), $X = X^N$ is a $G/N$-spectrum with trivial action. We construct a zigzag $\text{colim}\,_N \,X^{hG/N} \buildrelΦ\over\longrightarrow \text{colim}\,_N \,(X^{hN})^{hG/N} \buildrelΨ\over\longleftarrow X^{hG}$, where $Ψ$ is a weak equivalence. When $Φ$ is a weak equivalence, this zigzag gives an interesting model for $X^{hG}$ (for example, its Spanier-Whitehead dual is $\text{holim}\,_N \,F(X^{hG/N}, S^0)$). We prove that this happens in the following cases: (1) $|G| < \infty$; (2) $X$ is bounded above; (3) there exists $\{U\}$ cofinal in $\{N\}$, such that for each $U$, $H^s_c(U, π_\ast(X)) = 0$, for $s > 0$. Given (3), for each $U$, there is a weak equivalence $X \buildrel\simeq\over\longrightarrow X^{hU}$ and $X^{hG} \simeq X^{hG/U}$. For case (3), we give a series of corollaries and examples. As one instance of a family of examples, if $p$ is a prime, $K(n_p,p)$ the $n_p$th Morava $K$-theory $K(n_p)$ at $p$ for some $n_p \geq 1$, and $\mathbb{Z}_p$ the $p$-adic integers, then for each $m \geq 2$, (3) is satisfied when $G \leqslant \prod_{p \leq m} \mathbb{Z}_p$ is closed, $X = \bigvee_{p > m} (H\mathbb{Q} \vee K(n_p,p))$, and $\{U\} := \{N_G \mid N_G \trianglelefteq_o G\}$. △ Less

Submitted 5 August, 2019; originally announced August 2019.

Comments: 18 pages; submitted for publication

MSC Class: 55P42; 55P91

From low probability to high confidence in stochastic convex optimization

Authors: Damek Davis, Dmitriy Drusvyatskiy, Lin Xiao, Junyu Zhang

Abstract: Standard results in stochastic convex optimization bound the number of samples that an algorithm needs to generate a point with small function value in expectation. More nuanced high probability guarantees are rare, and typically either rely on "light-tail" noise assumptions or exhibit worse sample complexity. In this work, we show that a wide class of stochastic optimization algorithms for strong… ▽ More Standard results in stochastic convex optimization bound the number of samples that an algorithm needs to generate a point with small function value in expectation. More nuanced high probability guarantees are rare, and typically either rely on "light-tail" noise assumptions or exhibit worse sample complexity. In this work, we show that a wide class of stochastic optimization algorithms for strongly convex problems can be augmented with high confidence bounds at an overhead cost that is only logarithmic in the confidence level and polylogarithmic in the condition number. The procedure we propose, called proxBoost, is elementary and builds on two well-known ingredients: robust distance estimation and the proximal point method. We discuss consequences for both streaming (online) algorithms and offline algorithms based on empirical risk minimization. △ Less

Submitted 16 October, 2019; v1 submitted 31 July, 2019; originally announced July 2019.

Comments: 37 pages

MSC Class: 65K05; 65K10; 90C15; 90C25

Stochastic algorithms with geometric step decay converge linearly on sharp functions

Authors: Damek Davis, Dmitriy Drusvyatskiy, Vasileios Charisopoulos

Abstract: Stochastic (sub)gradient methods require step size schedule tuning to perform well in practice. Classical tuning strategies decay the step size polynomially and lead to optimal sublinear rates on (strongly) convex problems. An alternative schedule, popular in nonconvex optimization, is called \emph{geometric step decay} and proceeds by halving the step size after every few epochs. In recent work,… ▽ More Stochastic (sub)gradient methods require step size schedule tuning to perform well in practice. Classical tuning strategies decay the step size polynomially and lead to optimal sublinear rates on (strongly) convex problems. An alternative schedule, popular in nonconvex optimization, is called \emph{geometric step decay} and proceeds by halving the step size after every few epochs. In recent work, geometric step decay was shown to improve exponentially upon classical sublinear rates for the class of \emph{sharp} convex functions. In this work, we ask whether geometric step decay similarly improves stochastic algorithms for the class of sharp nonconvex problems. Such losses feature in modern statistical recovery problems and lead to a new challenge not present in the convex setting: the region of convergence is local, so one must bound the probability of escape. Our main result shows that for a large class of stochastic, sharp, nonsmooth, and nonconvex problems a geometric step decay schedule endows well-known algorithms with a local linear rate of convergence to global minimizers. This guarantee applies to the stochastic projected subgradient, proximal point, and prox-linear algorithms. As an application of our main result, we analyze two statistical recovery tasks---phase retrieval and blind deconvolution---and match the best known guarantees under Gaussian measurement models and establish new guarantees under heavy-tailed distributions. △ Less

Submitted 22 July, 2019; originally announced July 2019.

MSC Class: 65K05; 65K10; 90C15; 90C30; 90C06

On the unordered configuration space C(RP^n,2)

Authors: Donald M. Davis

Abstract: We prove that, if n is a 2-power, the unordered configuration space C(RP^n,2) cannot be immersed in R^{4n-2} nor embedded as a closed subspace of R^{4n-1}, optimal results, while if n is not a 2-power, C(RP^n,2) can be immersed in R^{4n-3}. We also obtain cohomological lower bounds for the topological complexity of C(RP^n,2), which are nearly optimal when n is a 2-power. We also give a new descrip… ▽ More We prove that, if n is a 2-power, the unordered configuration space C(RP^n,2) cannot be immersed in R^{4n-2} nor embedded as a closed subspace of R^{4n-1}, optimal results, while if n is not a 2-power, C(RP^n,2) can be immersed in R^{4n-3}. We also obtain cohomological lower bounds for the topological complexity of C(RP^n,2), which are nearly optimal when n is a 2-power. We also give a new description of the mod-2 cohomology algebra of the Grassmann manifold G_{n+1,2}. △ Less

Submitted 24 May, 2019; originally announced May 2019.

MSC Class: 55R80; 57R42; 55M30; 55S15

Low-rank matrix recovery with composite optimization: good conditioning and rapid convergence

Authors: Vasileios Charisopoulos, Yudong Chen, Damek Davis, Mateo Díaz, Lijun Ding, Dmitriy Drusvyatskiy

Abstract: The task of recovering a low-rank matrix from its noisy linear measurements plays a central role in computational science. Smooth formulations of the problem often exhibit an undesirable phenomenon: the condition number, classically defined, scales poorly with the dimension of the ambient space. In contrast, we here show that in a variety of concrete circumstances, nonsmooth penalty formulations d… ▽ More The task of recovering a low-rank matrix from its noisy linear measurements plays a central role in computational science. Smooth formulations of the problem often exhibit an undesirable phenomenon: the condition number, classically defined, scales poorly with the dimension of the ambient space. In contrast, we here show that in a variety of concrete circumstances, nonsmooth penalty formulations do not suffer from the same type of ill-conditioning. Consequently, standard algorithms for nonsmooth optimization, such as subgradient and prox-linear methods, converge at a rapid dimension-independent rate when initialized within constant relative error of the solution. Moreover, nonsmooth formulations are naturally robust against outliers. Our framework subsumes such important computational tasks as phase retrieval, blind deconvolution, quadratic sensing, matrix completion, and robust PCA. Numerical experiments on these problems illustrate the benefits of the proposed approach. △ Less

Submitted 22 April, 2019; originally announced April 2019.

Comments: 80 pages

MSC Class: 65K10; 90C06

The Shape of Thurston's Master Teapot

Authors: Harrison Bray, Diana Davis, Kathryn Lindsey, Chenxi Wu

Abstract: We establish basic geometric and topological properties of Thurston's Master Teapot and the Thurston set for superattracting unimodal self-maps of intervals. In particular, the Master Teapot is connected, contains the unit cylinder, and its intersection with a set $\mathbb{D} \times \{c\}$ grows monotonically with $c$. We show that the Thurston set described above is not equal to the Thurston set… ▽ More We establish basic geometric and topological properties of Thurston's Master Teapot and the Thurston set for superattracting unimodal self-maps of intervals. In particular, the Master Teapot is connected, contains the unit cylinder, and its intersection with a set $\mathbb{D} \times \{c\}$ grows monotonically with $c$. We show that the Thurston set described above is not equal to the Thurston set for postcritically finite tent maps, and we provide an arithmetic explanation for why certain gaps appear in plots of finite approximations of the Thurston set. △ Less

Submitted 11 April, 2020; v1 submitted 27 February, 2019; originally announced February 2019.

Comments: 26 pages, 5 figures

K-theory and immersions of spatial polygon spaces

Authors: Donald M Davis

Abstract: For ell a generic n-tuple of positive numbers, N(ell) denotes the space of isometry classes of oriented n-gons in R^3 with side lengths specified by ell. We determine the algebra K(N(ell)) and use this to obtain nonimmersions of the 2(n-3)-manifold N(ell) in Euclidean space for several families of ell. We also use obstruction theory to tell exactly when N(ell) immerses in R^{4n-14} for two familie… ▽ More For ell a generic n-tuple of positive numbers, N(ell) denotes the space of isometry classes of oriented n-gons in R^3 with side lengths specified by ell. We determine the algebra K(N(ell)) and use this to obtain nonimmersions of the 2(n-3)-manifold N(ell) in Euclidean space for several families of ell. We also use obstruction theory to tell exactly when N(ell) immerses in R^{4n-14} for two families of ell. △ Less

Submitted 13 January, 2019; originally announced January 2019.

MSC Class: 57R42; 55N15; 55R80; 57R20

Composite optimization for robust blind deconvolution

Authors: Vasileios Charisopoulos, Damek Davis, Mateo Díaz, Dmitriy Drusvyatskiy

Abstract: The blind deconvolution problem seeks to recover a pair of vectors from a set of rank one bilinear measurements. We consider a natural nonsmooth formulation of the problem and show that under standard statistical assumptions, its moduli of weak convexity, sharpness, and Lipschitz continuity are all dimension independent. This phenomenon persists even when up to half of the measurements are corrupt… ▽ More The blind deconvolution problem seeks to recover a pair of vectors from a set of rank one bilinear measurements. We consider a natural nonsmooth formulation of the problem and show that under standard statistical assumptions, its moduli of weak convexity, sharpness, and Lipschitz continuity are all dimension independent. This phenomenon persists even when up to half of the measurements are corrupted by noise. Consequently, standard algorithms, such as the subgradient and prox-linear methods, converge at a rapid dimension-independent rate when initialized within constant relative error of the solution. We then complete the paper with a new initialization strategy, complementing the local search algorithms. The initialization procedure is both provably efficient and robust to outlying measurements. Numerical experiments, on both simulated and real data, illustrate the developed theory and methods. △ Less

Submitted 18 January, 2019; v1 submitted 6 January, 2019; originally announced January 2019.

Comments: 60 pages, 14 figures

MSC Class: 65K10; 90C06

The typical measure preserving transformation is not an interval exchange transformation

Authors: Jon Chaika, Diana Davis

Abstract: We show that the typical measure preserving transformation is not isomorphic to any interval exchange transformation. We show that the typical measure preserving transformation is not isomorphic to any interval exchange transformation. △ Less

Submitted 26 December, 2018; originally announced December 2018.

Periodic paths on the pentagon, double pentagon and golden L

Authors: Diana Davis, Samuel Lelievre

Abstract: We give a tree structure on the set of all periodic directions on the golden L, which gives an associated tree structure on the set of periodic directions for the pentagon billiard table and double pentagon surface. We use this to give the periods of periodic directions on the pentagon and double pentagon. We also show examples of many periodic billiard trajectories on the pentagon, which are stri… ▽ More We give a tree structure on the set of all periodic directions on the golden L, which gives an associated tree structure on the set of periodic directions for the pentagon billiard table and double pentagon surface. We use this to give the periods of periodic directions on the pentagon and double pentagon. We also show examples of many periodic billiard trajectories on the pentagon, which are strikingly beautiful, and we describe some of their properties. Finally, we give conjectures and future directions based on experimental computer evidence. △ Less

Submitted 26 October, 2018; originally announced October 2018.

Comments: 29 pages, 20 figures, 1 appendix; Sage code in .tex file