Force balance controls the relaxation time of the gradient descent algorithm in the satisfiable phase

We numerically study the relaxation dynamics of the single-layer perceptron with the spherical constraint. This is the simplest model of neural networks and serves a prototypical mean-field model of both convex and nonconvex optimization problems. The relaxation time of the gradient descent algorithm rapidly increases near the SAT-UNSAT (satisfiable-unsatisfiable) transition point. We numerically confirm that the first nonzero eigenvalue of the Hessian controls the relaxation time. This first eigenvalue vanishes much faster upon approaching the SAT-UNSAT transition point than the prediction of the Marchenko-Pastur law in random matrix theory derived under the assumption that the set of unsatisfied constraints are uncorrelated. This leads to a nontrivial critical exponent of the relaxation time in the SAT phase. Using a simple scaling analysis, we show that the isolation of this first eigenvalue from the bulk of spectrum is attributed to the force balance at the SAT-UNSAT transition point. Finally, we show that the estimated critical exponent of the relaxation time in the nonconvex region agrees very well with that of frictionless spherical particles, which have been studied in the context of the jamming transition of granular materials.