Conformal geodesics are not variational in higher dimensions
Boris Kruglikov
TL;DR
The paper addresses whether conformal geodesic equations admit a conformally invariant variational formulation in dimensions $n>3$. Using the inverse problem of the calculus of variations, jet-space analysis, and dimension-reduction techniques, it establishes that both parametrized and unparametrized conformal geodesics are not variational in higher dimensions, with the 3D case remaining special (unparametrized conformal circles are variational via a torsion-based Lagrangian). The argument combines explicit flat-model obstructions (via Vainberg–Tonti reduction) with a generic-rank analysis to extend non-variationality to general metrics. This leads to a proposed selection principle: variationality of conformal geodesics singles out dimension three, with important implications for the role of action principles in conformal geometry and general relativity.
Abstract
Variationality of the equation of conformal geodesics is an important problem in geometry with applications to general relativity. Recently it was proven that, in three dimensions, this system of equations for un-parametrized curves is the Euler-Lagrange equations of a certain conformally invariant functional, while the parametrized system in three dimensions is not variational. We demonstrate that variationality fails in higher dimensions for both parametrized and un-parametrized conformal geodesics, indicating that variational principle may be the selection principle for the physical dimension.
