Searcharxiv⌕ Search

arXiv subjects

Richard Montgomery

Publications and source records attributed to Richard Montgomery.

At least 91 records · Page 5Linked to original sources
arXiv2015

No hyperbolic pants for the 4-body problem

The $N$-body problem with a $1/r^2$ potential has, in addition to translation and rotational symmetry, an effective scale symmetry which allows its zero energy flow to be reduced to a geodesic flow on complex projective $N-2$-space, minus a hyperplane arrangement. When $N=3$ we get a geodesic flow on the two-sphere minus three points. If, in addition we assume that the three masses are equal, then it was proved in [1] that the corresponding metric is hyperbolic: its Gaussian curvature is negative except at two points. Does the negative curvature property persist for $N=4$, that is, in the equal mass $1/r^2$ 4-body problem? Here we prove `no' by computing that the corresponding Riemannian metric in this $N=4$ case has positive sectional curvature at some two-planes. This `no' answer dashes hopes of naively extending hyperbolicity from $N=3$ to $N>3$.

math.DS↗
arXiv2014

Almost all friendly matrices have many obstructions

A symmetric $m\times m$ matrix $M$ with entries taken from $\{0,1,\ast\}$ gives rise to a graph partition problem, asking whether a graph can be partitioned into $m$ vertex sets matched to the rows (and corresponding columns) of $M$ such that, if $M_{ij}=1$, then any two vertices between the corresponding vertex sets are joined by an edge, and if $M_{ij}=0$ then any two vertices between the corresponding vertex sets are not joined by an edge. The entry $\ast$ places no restriction on the edges between the corresponding sets. This problem generalises graph colouring and graph homomorphism problems. A graph with no $M$-partition but such that every proper subgraph does have an $M$-partition is called a minimal obstruction. Feder, Hell and Xie have defined friendly matrices and shown that non-friendly matrices have infinitely many minimal obstructions. They showed through examples that friendly matrices can have finitely or infinitely many minimal obstructions and gave an example of a friendly matrix with an NP-hard partition problem. Here we show that almost all friendly matrices have infinitely many minimal obstructions and an NP-hard partition problem.

math.CO↗
arXiv2014

Realizing All Free Homotopy Classes for the Newtonian Three-Body Problem

The configuration space of the planar three-body problem when collisions are excluded has a rich topology which supports a large set of free homotopy classes. Most classes survive modding out by rotations. Those that survive are called the reduced free homotopy classes and have a simple description when projected onto the shape sphere. They are coded by syzygy sequences. We prove that every reduced free homotopy class, and thus every reduced syzygy sequence, is realized by a reduced periodic solution to the Newtonian planar three-body problem. The realizing solutions have nonzero angular momentum, repeatedly come very close to triple collision, and have lots of "stutters"--repeated syzygies of the same type. The heart of the proof is contained in the work by one of us on symbolic dynamics arising out of the central configurations after the triple collision is blown up using McGehee's method.

math.DS↗
arXiv2014

Who's Afraid of the Hill Boundary?

The Jacobi-Maupertuis metric allows one to reformulate Newton's equations as geodesic equations for a Riemannian metric which degenerates at the Hill boundary. We prove that a JM geodesic which comes sufficiently close to a regular point of the boundary contains pairs of conjugate points close to the boundary. We prove the conjugate locus of any point near enough to the boundary is a hypersurface tangent to the boundary. Our method of proof is to reduce analysis of geodesics near the boundary to that of solutions to Newton's equations in the simplest model case: a constant force. This model case is equivalent to the beginning physics problem of throwing balls upward from a fixed point at fixed speeds and describing the resulting arcs, see Fig. 2.

math.DS↗
arXiv2014

Logarithmically-small Minors and Topological Minors

Mader proved that for every integer $t$ there is a smallest real number $c(t)$ such that any graph with average degree at least $c(t)$ must contain a $K_t$-minor. Fiorini, Joret, Theis and Wood conjectured that any graph with $n$ vertices and average degree at least $c(t)+ε$ must contain a $K_t$-minor consisting of at most $C(ε,t)\log n$ vertices. Shapira and Sudakov subsequently proved that such a graph contains a $K_t$-minor consisting of at most $C(ε,t)\log n \log\log n$ vertices. Here we build on their method using graph expansion to remove the $\log\log n$ factor and prove the conjecture. Mader also proved that for every integer $t$ there is a smallest real number $s(t)$ such that any graph with average degree larger than $s(t)$ must contain a $K_t$-topological minor. We prove that, for sufficiently large $t$, graphs with average degree at least $(1+ε)s(t)$ contain a $K_t$-topological minor consisting of at most $C(ε,t)\log n$ vertices. Finally, we show that, for sufficiently large $t$, graphs with average degree at least $(1+ε)c(t)$ contain either a $K_t$-minor consisting of at most $C(ε,t)$ vertices or a $K_t$-topological minor consisting of at most $C(ε,t)\log n$ vertices.

math.CO↗
arXiv2014

Sharp threshold for embedding combs and other spanning trees in random graphs

When $k|n$, the tree $\mathrm{Comb}_{n,k}$ consists of a path containing $n/k$ vertices, each of whose vertices has a disjoint path length $k-1$ beginning at it. We show that, for any $k=k(n)$ and $ε>0$, the binomial random graph $\mathcal{G}(n,(1+ε)\log n/ n)$ almost surely contains $\mathrm{Comb}_{n,k}$ as a subgraph. This improves a recent result of Kahn, Lubetzky and Wormald. We prove a similar statement for a more general class of trees containing both these combs and all bounded degree spanning trees which have at least $εn/ \log^9n$ disjoint bare paths length $\lceil\log^9 n\rceil$. We also give an efficient method for finding large expander subgraphs in a binomial random graph. This allows us to improve a result on almost spanning trees by Balogh, Csaba, Pei and Samotij.

math.CO↗
arXiv2014

The Three-body problem and the shape sphere

[This is an expository article. I have submitted it to the American Mathematical Monthly.] The three-body problem defines a dynamics on the space of triangles in the plane. The shape sphere is the moduli space of oriented similarity classes of planar triangles and lies inside shape space, a Euclidean 3-space parametrizing oriented congruence classes of triangles. We derive and investigate the geometry and dynamics induced on these spaces by the three-body problem. We present two theorems concerning the three-body problem whose discovery was made through the shape space perspective

math.DS↗
arXiv2013

MICZ-Kepler = dynamics on the cone over the rotation group

We show that the n-dimensional MICZ-Kepler system arises from symplectic reduction of a simple mechanical system on the cone over the rotation group SO(n). As a corollary we derive an elementary formula for its general solution. The punch-line of our computation is that the additional MICZ-Kepler $|ϕ|^2/r^2$ type potential term is the rotational part of the cone's kinetic energy.

math-ph↗
arXiv2012

Keplerian Dynamics on the Heisenberg Group and Elsewhere

Posing Kepler's problem of motion around a fixed "sun" requires the geometric mechanician to choose a metric and a Laplacian. The metric provides the kinetic energy. The fundamental solution to the Laplacian (with delta source at the "sun") provides the potential energy. Posing Kepler's three laws (with input from Galileo) requires symmetry conditions. The metric space must be homogeneous, isotropic, and admit dilations. Any Riemannian manifold enjoying these three symmetry properties is Euclidean. So if we want a semblance of Kepler's three laws to hold but also want to leave the Euclidean realm, we are forced out of the realm of Riemannian geometries. The Heisenberg group (a subRiemannian geometry) and lattices provide the simplest examples of metric spaces enjoying a semblance of all three of the Keplerian symmetries. We report success in posing, and solving, the Kepler problem on the Heisenberg group. We report failures in posing the Kepler problem on the rank two lattice and partial success in solving the problem on the integers. We pose a number of questions.

math.DS↗
arXiv2012

Symmetric Regularization, Reduction and Blow-Up of the Planar Three-Body Problem

We carry out a sequence of coordinate changes for the planar three-body problem which successively eliminate the translation and rotation symmetries, regularize all three double collision singularities and blow-up the triple collision. Parametrizing the configurations by the three relative position vectors maintains the symmetry among the masses and simplifies the regularization of binary collisions. Using size and shape coordinates facilitates the reduction by rotations and the blow-up of triple collision while emphasizing the role of the shape sphere. By using homogeneous coordinates to describe Hamiltonian systems whose configurations spaces are spheres or projective spaces, we are able to take a modern, global approach to these familiar problems. We also show how to obtain the reduced and regularized differential equations in several convenient local coordinates systems.

math.CA↗
arXiv2011

From Brake to Syzygy

In the planar three-body problem, we study solutions with zero initial velocity (brake orbits). Following such a solution until the three masses become collinear (syzygy), we obtain a continuous, flow-induced Poincaré map. We study the image of the map in the set of collinear configurations and define a continuous extension to the Lagrange triple collision orbit. In addition we provide a variational characterization of some of the resulting brake-to-syzygy orbits and find simple examples of periodic brake orbits.

math.DS↗
arXiv2009

Curve Singularities and Monster / Semple Towers

In earlier work, we introduced the `Monster tower', a tower of fibrations associated to planar curves. We constructed an algorithm for classifying its points with respect to the equivalence relation generated by the action of the contact pseudogroup on the tower. Here, we construct the analogous tower for curves in $n$-space. (This tower is known as the Semple Bundle in Algebraic Geometry.) The pseudo-group of diffeomorphisms of $n$-space acts on each level of the extended tower. We take initial steps toward classifying points of this extended Monster tower under this pseudogroup action. Arnol'd's list of stable simple curve singularities plays a central role in these initial steps. We end with a list of open problems.

math.DG↗
arXiv2008

The Chains of Left-invariant CR-structures on SU(2)

We compute the chains associated to the left-invariant CR structures on the three-sphere. These structures are characterized by a single real modulus $a$. For the standard structure $a=1$, the chains are well-known and are closed curves. We show that for almost all other values of the modulus $a$ either two or three types of chains are simultaneously present : (I) closed curves, (II) quasi-periodic curves dense on two-torii, or (III) chains homoclinic between closed curves. For $1 < a < \sqrt{3}$ no curves of the last type occur. A bifurcation occurs at $a = \sqrt{3}$ and from that point on all three types of chains are guaranteed to exist, and exhaust all chains. The method of proof is to use the Fefferman metric characterization of chains, combined with tools from geometric mechanics. The key to the computation is a reduced Hamiltonian system, similar to Euler's rigid body system, and depending on $a$, which is integrable.

math.CV↗
arXiv2006

G_2 and the "Rolling Distribution"

Associated to the problem of rolling one surface along another there is a five-manifold M with a rank two distribution. If the two surfaces are spheres then M is the product of the rotation group SO_3 with the two-sphere and its distribution enjoys an obvious symmetry group; the product of two SO_3's, one for each sphere. But if the ratio of radii of the spheres is 1:3 and if the distribution is lifted to the universal cover S^3 \times S^2 of M, then the symmetry group becomes much larger: the split real form of the Lie group G_2. This fact goes back to Cartan in a sense, and can be found in a paper by Bryant and Hsu. We prove this fact through two explicit constructions, relying on the theory of roots and weights for the Lie algebra of G_2, and on its 7-dimensional representation.

math.DG↗
arXiv2006

The only syzygy-free solution is Lagrange's

A syzygy in the three-body problem is a collinear instant. We prove that with the exception of Lagrange's solution every solution to the zero angular momentum Newtonian three-body problem suffers syzygies. The proof works for all mass ratios.

math.DS↗
arXiv2004

Momentum Maps and Classical Relativistic Fields. Part I: Covariant Field Theory

This is the first paper of a five part work in which we study the Lagrangian and Hamiltonian structure of classical field theories with constraints. Our goal is to explore some of the connections between initial value constraints and gauge transformations in such theories (either relativistic or not). To do this, in the course of these four papers, we develop and use a number of tools from symplectic and multisymplectic geometry. Of central importance in our analysis is the notion of the ``energy-momentum map'' associated to the gauge group of a given classical field theory. We hope to demonstrate that many different and apparently unrelated facets of field theories can be thereby tied together and understood in an essentially new way. In Part I we develop some of the basic theory of classical fields from a spacetime covariant viewpoint. We begin with a study of the covariant Lagrangian and Hamiltonian formalisms, on jet bundles and multisymplectic manifolds, respectively. Then we discuss symmetries, conservation laws, and Noether's theorem in terms of ``covariant momentum maps.''

math-ph↗
arXiv2004

Nonholonomic systems via moving frames: Cartan equivalence and Chaplygin Hamiltonization

A nonholonomic system consists of a configuration space Q, a Lagrangian L, and an nonintegrable constraint distribution H, with dynamics governed by Lagrange-d'Alembert's principle. We present two studies both using adapted moving frames. In the first study we apply Cartan's method of equivalence to investigate the geometry underlying a nonholonomic system. As an example we compute the differential invariants for a nonholonomic system on a four-dimensional configuration manifold endowed with a rank two (Engel) distribution. In the second part we study G-Chaplygin systems. These are systems where the constraint distribution is given by a connection on a principal fiber bundle with total space Q and base space S=Q/G, and with a G-equivariant Lagrangian. These systems compress to an almost Hamiltonian system on $T^{*}S$. Under an $s \in S$ dependent time reparameterization a number of compressed systems become Hamiltonian. A necessary condition for Hamiltonization is the existence of an invariant measure on the original system. Assuming an invariant measure we describe the obstruction to Hamiltonization. Chaplygin's "rubber" sphere, a ball with unequal inertia coefficients rolling without slipping or spinning (about the vertical axis) on a plane is Hamiltonizable when compressed to $T^{*}SO(3)$. Finally we discuss reduction of internal symmetries. Chaplygin's "marble" (where spinning is allowed) is not Hamiltonizable when compressed to $T^{*}SO(3)$; we conjecture that it is also not Hamiltonizable when reduced to $T^{*}S^{2}$.

math-ph↗