AI & ML Paradigm Challenge

We’re trying to solve the world’s hardest AI problems by pretending they’re just basic high school math, and it’s holding us back.

April 10, 2026

Original Paper

Smooth, globally Polyak-Łojasiewicz functions are nonlinear least-squares

Nicolas Boumal, Christopher Criscitiello, Quentin Rebjock

arXiv · 2604.07972

The Takeaway

Researchers proved that the 'PŁ condition'—a staple for ensuring AI models can be trained efficiently—imposes a rigid geometric structure that essentially turns any problem into a sum of squares. This changes how practitioners view the optimization landscape, suggesting that many 'complex' functions are mathematically more restricted than we thought.

From the abstract

The Polyak-Łojasiewicz (PŁ) condition is often invoked in nonconvex optimization because it allows fast convergence of algorithms beyond strong convexity. A function $f \colon \mathcal{M} \to \mathbb{R}$ on a Riemannian manifold $\mathcal{M}$ is globally PŁ if $\|\nabla f(x)\|^2 \geq 2\mu(f(x) - f^*)$ for all $x$, where $f^* = \inf f$ and $\mu > 0$. How much does this pointwise, first-order inequality constrain $f$ and its set of minimizers $S$?We show that if $f$ is also smooth ($C^\infty$) and