Answers are grounded in this paper and cite their sources.
The ₚᵖ/qᵖ ratio model is scale invariant and well suited to sparse signal recovery, but its nonconvex, nonsmooth, and fractional structure complicates unified theoretical and algorithmic analysis. For 0<p≤1 and 1<q<∞, this paper develops a unified framework for exact recovery, stable recovery of sparse and compressible signals, and convergence of a fractional algorithm. First, we establish a local optimality criterion and a null-space condition for uniform exact recovery, and derive high-probability sample-complexity bounds under independent isotropic sub-Gaussian measurements. For noisy recovery of s-sparse signals, we relax the sufficient condition based on the restricted isometry property (RIP) when $0<p<1$, while recovering the existing condition when $p=1$. Over the common admissible range of the restricted isometry constant (RIC), our error bound is tighter or identical. For compressible signals, we derive error estimates based on the RIP and the restricted orthogonality property (ROP) with dimension-independent constants, removing the explicit ambient-dimension dependence in existing results for $1<q<2$, and obtain an RIP-only sufficient condition that requires no separate restricted orthogonality constant (ROC) estimate. Finally, we propose a prox-linear Dinkelbach algorithm and prove that every cluster point is critical. Numerical experiments demonstrate that suitable choices of $(p,q)$ perform well for high-dynamic-range sparse signals and coherent sensing matrices.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: