Abstract
This paper investigates the isoperimetric problem—finding the shape that encloses a fixed area with minimum perimeter—on planes where space is weighted by a density function. We specifically examine planes with generalized curvature zero. We solve the isoperimetric problem for the plane with exponential density $e^x$ and radial density $r^p$ (for $p < 0$). Additionally, we present a proof by Robert Bryant demonstrating that the Gauss plane possesses a unique closed geodesic.
Introduction
In classical Euclidean geometry, the solution to the isoperimetric problem is always a circle. However, in a manifold with density, both area and perimeter are weighted by a positive function $f = e^\phi$. This weighting changes the geometric landscape: because area is a two-dimensional measure and perimeter is one-dimensional, they scale differently. Density is not merely a coordinate transformation but a fundamental change in the space's properties.
This study focuses on planes with generalized curvature zero. In these spaces, the log of the density function, $\phi$, is harmonic (satisfying Laplace’s equation $\Delta \phi = 0$). Such planes share structural similarities with flat Euclidean space while exhibiting unique behaviors regarding optimal shapes.
The Exponential Density: $e^x$
For a plane with density $f(x,y) = e^x$, the authors prove that an optimal isoperimetric region does not exist.
In this environment, one can move a shape further into the "thinner" regions of the plane (as $x \to -\infty$) to reduce its perimeter while maintaining a constant area. While the ratio of perimeter to area approaches an infimum of 1, this limit is never actually attained by any bounded region. Consequently, there is no "best" shape, only a sequence of shapes that get progressively closer to the theoretical minimum.
Radial Densities: $r^p$
The behavior of the isoperimetric problem under radial density $f(r) = r^p$ varies significantly based on the exponent $p$:
- $p < -2$: A circle centered at the origin is the unique isoperimetric region. The high density at the center effectively "anchors" the optimal shape.
- $-2 \le p < 0$: The problem effectively "breaks." In this range, it is possible to enclose any given area with an infinitely small perimeter, meaning no stable minimum exists.
- $p > 0$: An isoperimetric region is guaranteed to exist and must contain the origin. However, a circle centered at the origin is unstable. The authors conjecture that the optimal shape is an off-center, convex disk that hugs the origin, though a formal proof of the exact boundary remains an open challenge.
The Gauss Plane and the $\mathcal{J}$-plane
The paper explores two specific harmonic densities: the Gauss plane ($e^{-r^2/2}$) and the $\mathcal{J}$-plane ($\exp(x^2 - y^2)$).
The Gauss Plane
A major contribution of this work is a proof by Robert Bryant regarding the Gauss plane. He demonstrates that the only simple closed geodesic—a path that returns to its start without self-intersection—is a circle centered at the origin. This confirms that the radial symmetry of the Gaussian density dictates the only possible periodic "straight" path in that space.
The $\mathcal{J}$-plane
In the $\mathcal{J}$-plane, the authors find that the geometry is far more elusive. Most geodesics in this space have infinite length and do not close. While they characterize the behavior of these paths at the "ends" of the plane, the complete mapping of how these paths curve in all directions remains partially unresolved.
Conclusion
By extending the isoperimetric problem to planes with density, this research reveals how harmonic density functions can either preserve, distort, or entirely eliminate the existence of optimal geometric shapes. While radial densities often favor the origin, exponential densities demonstrate that some spaces are simply too "thin" to support a shape of least resistance.
Summary updated