English
The Bonnet Problem: Two Tori with Identical Metric and Mean Curvature but Different Global Shape
An Unlikely Introduction to the Bonnet Problem in Differential Geometry
While a certain tennis player is already preparing for his next tournament, let us talk about tori in topology.
Imagine you have never seen the Earth from space. You could still figure out it is round: all it takes is measuring distances and angles on the surface. Geometry offers a similar idea: sometimes a small set of local measurements is enough to reconstruct an entire shape.
What the Bonnet Problem Tells Us About Surfaces
In 1867, Pierre Ossian Bonnet showed that if you know the metric of a surface — how distances are measured on it — and the mean curvature at every point — how the surface bends on average in space — then this information is usually sufficient to determine the surface uniquely.
Usually, but not always. And it is precisely that “not always” that mathematicians were hunting for nearly a century and a half.
The Long Search for Counterexamples
The trouble was that all known exceptions turned out to be non-compact: they stretched to infinity or had boundaries. For “nice” closed surfaces like the sphere and the torus, no counterexample could be found.
By the early 1980s it became clear that for genus 0 surfaces — those with no holes — the Bonnet result holds completely. For tori, however, a subtle loophole remained: in theory, there could exist at most two distinct tori sharing the same local geometric data.
Two Twisty Tori: A Resolution of the Bonnet Problem for Tori
In October 2025, Alexander Bobenko, Tim Hoffmann, and Andrew Sageman-Furnas finally produced such a pair: two highly twisted tori that agree on metric and mean curvature everywhere yet are globally different in shape.
The most elegant part of the story is how they got there. Their path began with discrete surfaces — a pixelated kind of geometry where smoothness is replaced by a mesh. In 2018, a computer search turned up a strange rhino-like torus that looked suspicious but turned out to be a crucial clue.
From Discrete Geometry to Smooth Bonnet Problem Tori
Then came an unexpected bridge to Darboux: classical formulas about curvature lines that had long “run off to infinity.” By adjusting those formulas so the curvature lines closed up, the researchers grew smooth examples out of the discrete scaffolding.
First came a mirror-image pair — technically valid, but visually unsatisfying — and then a more explicitly distinct pair of tori.
Open Questions Around Bonnet Problem Tori
There is a catch: the tori found this way self-intersect, as if the surface pierces itself in three-dimensional space. And so the next question arises almost inevitably: can such Bonnet problem tori exist without self-intersections?
Sometimes mathematics reminds us: locally everything may look the same, yet globally you can inhabit an entirely different world.