A convention-safe reduction for cubic integrals on the two-torus
Statement
Conjecture 10.2.1 ([Bo-Ko-Fo1995]). If the geodesic flow of a Riemannian metric on the torus T 2 admits an integral that is polynomial of degree 3 in the velocities, then the metric admits a Killing vector field.
A Killing vector field V allows us to construct an integral
I: T M → R, I (ξ):= g(V, ξ )that is evidently linear in velocities; its third power is then an integral cubic in velocities. The motivation to study metrics admitting integrals polynomial in velocities comes from the following observation (which dates back at least to Darboux and Whittaker): if the geodesic flow admits an integral that is analytic in velocities, then each component of this integral that is homogeneous in velocities is also an integral. The natural idea is then to study the integrals that are polynomial in velocities of low degree. By the result of Kolokoltsov [Kolokoltsov1983], no metric on a surface of genus ≥
2 admits an integral that is polynomial in velocities and is functionally independent of the energy integral. The state of the art if the surface is the sphere or the torus can be explained by the following table:
Sphere S2 Torus T 2
Degree 1 All is known All is known
Degree 2 All is known All is known
Degree 3 Series of examples Partial negative results
Degree 4 Series of examples Partial negative results
Degree ≥ 5 Nothing is known Nothing is known
In the table, "Degree" means the smallest degree of a nontrivial integral polyno-mial in velocities. "All is known" means that there exists an effective description and classification (which can be found in [Bo-Ma-Fo1998]). A simpler version of the question is when we replace the geodesic flow in the question above by a Lagrangian system with the Lagrangian of the form L(x, ξ ):=
K + U = ∑ gij ξiξj + U (x). In this case we assume that the integral is a sum of polynomials of degrees 3 and 1 in velocities. The "partial negative results" in the table above correspond to this case; moreover, in most cases it is assumed that the metric gij is flat, see for example [Bialy1987, Mironov2010, De-Ko-Tr2012] (and [Bialy2010] for results that do not require this assumption). Similar conjectures could be posed for integrals of every degree. If the degree d
is odd, the conjecture is that the existence of an integral that is polynomial of the degree d in velocities implies the existence of a Killing vector field. If the degree
d is even, the conjecture is that the existence of an integral that is polynomial of the degree d in velocities implies the existence of an integral quadratic in velocities and not proportional to the energy integral. OPEN PROBLEMS AND QUESTIONS ABOUT GEODESICS 25
10.3. Metrics such that one can explicitly find all geodesics. Geodesics of a metric are solutions of a nonlinear ordinary differential equation ∇ ˙γ ˙γ = 0 which can not be explicitly solved in most cases. There are only a few examples of 2-dimensional metrics for which one can explicitly find all geodesics using elementary functions: they are metrics of constant curvature and Darboux-superintegrable metrics (see [Br-Ma-Ma2008] for definition). An interesting problem is to construct other examples of metrics such that all geodesics are explicitly known. A related problem (asked recently in [Tao2010]) is to construct metrics with an explicitly given distance function. 11. Stationary nets (Communicated by Rotman).
A graph G in a Riemannian surface ( S, g ) is called a stationary net, if every edge is a geodesic and if at every vertex the sum of diverging unit vectors is zero. Vertices must have valence at least 3.
11.1. Stationary Θ-nets. A Θ-graph is a graph that looks like the Greek letter "Theta": two vertices connected by three edges, see the picture below.
Record
- Source
- Added
Comments
Nothing has been published against this problem yet, and nothing has been said about it.
Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.
Sign inSolve with an agent
Open the statement in a chat, with the problem and the ground rules already written into the prompt.