Brascamp-Lieb constants arise at the intersection of functional analysis, information theory, and geometric measure theory. They bound the integral of a product of functions evaluated along different linear projections. Computing them requires solving an optimization problem — finding the Gaussian that maximizes the ratio. The constants were treated as analytic objects: smooth, continuous, computed by optimization.
Chindris and Derksen (arXiv:2603.09057) prove they are semi-algebraic. The Brascamp-Lieb constants, viewed as functions of the projection data, satisfy polynomial relations. They are constrained by the same kind of equations that define algebraic varieties — not smooth functions of continuous parameters but solutions to polynomial systems.
Semi-algebraicity means the constants live in a far more rigid structure than their analytic definition suggests. An analytic function can take any value consistent with continuity. A semi-algebraic function can only take values that satisfy specific polynomial constraints. The space of possible Brascamp-Lieb constants is not a smooth manifold but an algebraic set — cut out by equations, not merely parameterized by coordinates.
The practical consequence: algorithms for computing these constants can exploit algebraic structure rather than relying on generic optimization. The theoretical consequence: the constants that seemed to emerge from a continuous optimization landscape actually sit on an algebraic skeleton. The optimization was finding solutions to equations it didn't know existed.
The deeper point is about hidden rigidity. Objects defined by optimization problems appear to have the smoothness of their defining landscapes. But the solutions can be constrained by algebraic relations invisible from the optimization perspective. The landscape is smooth; the solution set is algebraic. The rigidity was always there, below the differentiable surface.