AI & ML
A Nonadaptive Sparse Fourier Transform in Every Dimension
arXiv cs.NA0 views
arXiv:2611.25334v1 Announce Type: cross Abstract: Let $G=(\mathbb{Z}/2^a\mathbb{Z})^d$, of order $N=2^{ad}$. Given point-query access to any complex signal $X$ with Fourier transform $\hat{X}$ over $G$, we provide a randomized algorithm that constructs a $k$-sparse estimate $\widetilde{X}$ via non-adaptive queries such that, with large probability, $\|\widetilde{X}-\hat{X}\|_2^2 \leq C \cdot \|H_k(\hat{X})-\hat{X}\|_2^2$, for an absolute constant $C>0$, where $H_k(\cdot)$ denotes the best $k$-sparse approximation. The constant $C$ can be improved to any desired $1+\varepsilon$ by either supplying a constant-factor estimate on the $k$-sparse tail norm or allowing an arbitrarily small additive error term $\delta \|\hat{X}\|_2^2$ on the right-hand side. With a supplied tail estimate, we can also satisfy an $\ell_\infty/\ell_2$ heavy hitters guarantee. For fixed $\delta$, the number of samples is $O((k/\varepsilon^2)(\log N)\log(k/\varepsilon))$, and the recovery algorithm requires a polynomial in $(k/\varepsilon)\log N$ exact operations. The algorithm naturally extends to $G=(\mathbb{Z}/b^a\mathbb{Z})^d$ for any fixed integer $b>1$.
Source:https://arxiv.org/abs/2611.25334PreviousAccelerating Diffusion Maps on a Quantum Computer: Spectral Filtering and Galerkin PDE SolversNextFinite-difference least square methods for solving Hamilton-Jacobi equations using neural networks
Related posts

Serving Qwen3.5-0.8B with a 0.016% weight edit: measure the loops, pin the sampling on the server
0dev.to
Design Tokens for AI: Extract a Website's Real Colours, Fonts, Shadows and Icons So an LLM Builds On Brand
0dev.to
Trying to get our Arms around AI efficiency
0Stack Overflow Blog
Why an Embedded Game Fails on Some Phones, and How to Debug It
0dev.to
Microsoft 365 vs Office 2024: The Break-Even Math
0dev.to
Master AI Evals: Build Reliable, Scalable Model Evaluation Pipelines
0dev.to