Mathematical analysis uncovers exact subdivision matrix spectra for generalized cubic B-splines across all valences, confirming stable local convergence and vanishing spectral gaps.
The spectrum of a subdivision matrix controls local convergence near an extraordinary mesh element. For every surface valence n ≥ 3, Dietz's generalized cubic B-spline construction produces an initial-element matrix and a released double-ring matrix assembled from it and regular cubic stencils. The source established two independent 1/2-eigenvectors but left open whether 1/2 is subdominant, has exactly the expected multiplicity, and is semisimple. We resolve all three questions and give the complete spectra of both released surface matrices. A discrete Fourier transform reduces the initial matrix to one 3 × 3 block and n − 1 blocks of size 2 × 2; a phase-free block-triangular form gives the outer spectrum. Both matrices have a simple dominant eigenvalue 1, a semisimple subdominant eigenvalue 1/2 of algebraic and geometric multiplicity two, and no other eigenvalue of modulus at least 1/2. For n ≥ 5, the subsubdominant eigenvalue is 1/2 − 3π²/(4n²) + π⁴/(2n⁴) + O(n⁻⁶), explaining the vanishing spectral gap. A proved tensor identity gives the expected multiplicity three for the semi-regular initial-volume family. The analogous double-shell conclusion remains conditional on the tensor identity stated with a proof sketch in Dietz's source. Exact-identity checks, independent reconstructions, and Lean certificates accompany the paper. Fully irregular three-dimensional matrices and a complete C¹ theorem are not covered.
No takes yet. Share an insight, caveat, or question.
Lennart Rudolph (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: