Stability-Constrained Approximation in Spline KANs: Exact Layer Balancing and Budget-Compatible Saturation

Abstract:Deep spline superposition networks face a tension between approximation order and stability across depth. We study approximation under a hard layerwise Lipschitz budget, and organise it around two quantities: the factorisation stability complexity of a given deep factorisation, and the budget-compatible approximation complexity of a discretisation operator. First, we solve exactly the finite-depth diagonal balancing problem for a fixed chain of nonnegative envelope matrices: the optimal uniform layer budget equals $\|M_{L-1}\cdots M_0\|_{\infty\to\infty}^{1/L}$, attained by an explicit one-pass minimiser, for rectangular layers, with a complete treatment of degeneracies and non-attainment. The optimum can be arbitrarily larger than the Lipschitz constant of the network itself, because passing to envelopes destroys sign cancellation. Second, we give a constructive spline discretisation theorem preserving the budget up to a controlled slack, with an explicit grid threshold. Conversely, for linear spline-valued operators that preserve the budget exactly, we prove budget-compatible minimax lower bounds on classes constrained simultaneously in the first and third derivative norms -- a constraint pair that is forced by the problem and that rules out the usual scaling escapes. Finally, we show that the corresponding layer errors need not cancel under composition: for every operator of the class there is a stable depth-$L$ tower realising a constant fraction of the accumulated error, so the linear-in-depth accumulation of the upper bound is not a proof artefact.

33 pages. Ancillary code reproduces the numerical constants of Appendix B. Complements arXiv:2604.26444, which constructs deep KAN representations with controlled layer-wise Lipschitz product; here the allocation of that product across layers is solved exactly, and the cost of preserving it under spline discretisation is bounded from both sides

Machine Learning (stat.ML); Machine Learning (cs.LG); Numerical Analysis (math.NA)

arXiv:2609.17619 [stat.ML]

(or arXiv:2609.17619v1 [stat.ML] for this version)