Tight Bounds for Sketching the Operator Norm, Schatten Norms, and Subspace Embeddings
February 20, 2022 Β· Declared Dead Β· π International Workshop and International Workshop on Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques
"No code URL or promise found in abstract"
Evidence collected by the PWNC Scanner
Authors
Yi Li, David P. Woodruff
arXiv ID
2202.09797
Category
cs.DS: Data Structures & Algorithms
Citations
32
Venue
International Workshop and International Workshop on Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques
Last Checked
3 months ago
Abstract
We consider the following oblivious sketching problem: given $Ξ΅\in (0,1/3)$ and $n \geq d/Ξ΅^2$, design a distribution $\mathcal{D}$ over $\mathbb{R}^{k \times nd}$ and a function $f: \mathbb{R}^k \times \mathbb{R}^{nd} \rightarrow \mathbb{R}$, so that for any $n \times d$ matrix $A$, $$\Pr_{S \sim \mathcal{D}} [(1-Ξ΅) \|A\|_{op} \leq f(S(A),S) \leq (1+Ξ΅)\|A\|_{op}] \geq 2/3,$$ where $\|A\|_{op}$ is the operator norm of $A$ and $S(A)$ denotes $S \cdot A$, interpreting $A$ as a vector in $\mathbb{R}^{nd}$. We show a tight lower bound of $k = Ξ©(d^2/Ξ΅^2)$ for this problem. Our result considerably strengthens the result of Nelson and Nguyen (ICALP, 2014), as it (1) applies only to estimating the operator norm, which can be estimated given any OSE, and (2) applies to distributions over general linear operators $S$ which treat $A$ as a vector and compute $S(A)$, rather than the restricted class of linear operators corresponding to matrix multiplication. Our technique also implies the first tight bounds for approximating the Schatten $p$-norm for even integers $p$ via general linear sketches, improving the previous lower bound from $k = Ξ©(n^{2-6/p})$ [Regev, 2014] to $k = Ξ©(n^{2-4/p})$. Importantly, for sketching the operator norm up to a factor of $Ξ±$, where $Ξ±- 1 = Ξ©(1)$, we obtain a tight $k = Ξ©(n^2/Ξ±^4)$ bound, matching the upper bound of Andoni and Nguyen (SODA, 2013), and improving the previous $k = Ξ©(n^2/Ξ±^6)$ lower bound. Finally, we also obtain the first lower bounds for approximating Ky Fan norms.
Community Contributions
Found the code? Know the venue? Think something is wrong? Let us know!
π Similar Papers
In the same crypt β Data Structures & Algorithms
π
π
The Cartographer
R.I.P.
π»
Ghosted
Route Planning in Transportation Networks
R.I.P.
π»
Ghosted
Near-linear time approximation algorithms for optimal transport via Sinkhorn iteration
R.I.P.
π»
Ghosted
Hierarchical Clustering: Objective Functions and Algorithms
R.I.P.
π»
Ghosted
Graph Isomorphism in Quasipolynomial Time
π
π
The Cartographer
Simulation optimization: A review of algorithms and applications
Died the same way β π» Ghosted
R.I.P.
π»
Ghosted
Federated Learning: Strategies for Improving Communication Efficiency
R.I.P.
π»
Ghosted
In-Datacenter Performance Analysis of a Tensor Processing Unit
R.I.P.
π»
Ghosted
Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning
R.I.P.
π»
Ghosted