Table of Contents
Fetching ...

Improved Lower Bounds for all Odd-Query Locally Decodable Codes

Arpon Basu, Jun-Ting Hsieh, Pravesh K. Kothari, Andrew D. Lin

TL;DR

An argument based on spectral bounds on Kikuchi Matrices that lower bounds the blocklength of any LDC whose local decoding sets satisfy t-approximate strong regularity for any t ⩽ q works despite having no non-trivial absolute upper bound on the co-degrees of any set of vertices.

Abstract

We prove that for every odd $q\geq 3$, any $q$-query binary, possibly non-linear locally decodable code ($q$-LDC) $E:\{\pm1\}^k \rightarrow \{\pm1\}^n$ must satisfy $k \leq \tilde{O}(n^{1-2/q})$. For even $q$, this bound was established in a sequence of prior works. For $q=3$, the above bound was achieved in a recent work of Alrabiah, Guruswami, Kothari and Manohar using an argument that crucially exploits known exponential lower bounds for $2$-LDCs. Their strategy hits an inherent bottleneck for $q \geq 5$. Our key insight is identifying a general sufficient condition on the hypergraph of local decoding sets called $t$-approximate strong regularity. This condition demands that 1) the number of hyperedges containing any given subset of vertices of size $t$ (i.e., its co-degree) be equal to the same but arbitrary value $d_t$ up to a multiplicative constant slack, and 2) all other co-degrees be upper-bounded relative to $d_t$. This condition significantly generalizes related proposals in prior works that demand absolute upper bounds on all co-degrees. We give an argument based on spectral bounds on Kikuchi Matrices that lower bounds the blocklength of any LDC whose local decoding sets satisfy $t$-approximate strong regularity for any $t \leq q$. Crucially, unlike prior works, our argument works despite having no non-trivial absolute upper bound on the co-degrees of any set of vertices. To apply our argument to arbitrary $q$-LDCs, we give a new, greedy, approximate strong regularity decomposition that shows that arbitrary, dense enough hypergraphs can be partitioned (up to a small error) into approximately strongly regular pieces satisfying the required relative bounds on the co-degrees.

Improved Lower Bounds for all Odd-Query Locally Decodable Codes

TL;DR

An argument based on spectral bounds on Kikuchi Matrices that lower bounds the blocklength of any LDC whose local decoding sets satisfy t-approximate strong regularity for any t ⩽ q works despite having no non-trivial absolute upper bound on the co-degrees of any set of vertices.

Abstract

We prove that for every odd , any -query binary, possibly non-linear locally decodable code (-LDC) must satisfy . For even , this bound was established in a sequence of prior works. For , the above bound was achieved in a recent work of Alrabiah, Guruswami, Kothari and Manohar using an argument that crucially exploits known exponential lower bounds for -LDCs. Their strategy hits an inherent bottleneck for . Our key insight is identifying a general sufficient condition on the hypergraph of local decoding sets called -approximate strong regularity. This condition demands that 1) the number of hyperedges containing any given subset of vertices of size (i.e., its co-degree) be equal to the same but arbitrary value up to a multiplicative constant slack, and 2) all other co-degrees be upper-bounded relative to . This condition significantly generalizes related proposals in prior works that demand absolute upper bounds on all co-degrees. We give an argument based on spectral bounds on Kikuchi Matrices that lower bounds the blocklength of any LDC whose local decoding sets satisfy -approximate strong regularity for any . Crucially, unlike prior works, our argument works despite having no non-trivial absolute upper bound on the co-degrees of any set of vertices. To apply our argument to arbitrary -LDCs, we give a new, greedy, approximate strong regularity decomposition that shows that arbitrary, dense enough hypergraphs can be partitioned (up to a small error) into approximately strongly regular pieces satisfying the required relative bounds on the co-degrees.

Paper Structure

This paper contains 22 sections, 17 theorems, 41 equations, 3 figures, 2 algorithms.

Key Result

Theorem 1.1

For any $q$-LDC $E:\{\pm 1\}^k \rightarrow \{\pm 1\}^n$ with a constant distance, $k\leqslant O(n^{1-2/q} \log^4 n)$. If $E$ is linear, then, $k \leqslant O(n^{1-2/q} \log^2 n)$.

Figures (3)

  • Figure 1: This example shows, for $q = 5$, a scenario where sets $S$ as shown have $\geqslant d_3$ edges where one of the colors is $i$.
  • Figure 2: In this example, the hyperedges $C, C'$ induce an edge between $S, T$. Here, $q = 5, t = 2$. Vertices of the form $(x, 1)$ are colored green, while vertices of the form $(x, 2)$ are colored blue. Note that $S\oplus T = (C\setminus Q_\theta)\times\{1\}\bigcup\ (C'\setminus Q_\theta)\times\{2\}$. Also note how $S$ contains $2 = \left\lceil\frac{q - t}{2}\right\rceil$ green elements and $1 = \left\lfloor\frac{q - t}{2}\right\rfloor$ blue element from $S\oplus T$, while $T$ contains $1 = \left\lfloor\frac{q - t}{2}\right\rfloor$ green element and $2 = \left\lceil\frac{q - t}{2}\right\rceil$ blue elements from $S\oplus T$.
  • Figure 3: Example of a contributing monomial to $\mathsf{Deg}(s,s')$. Here, $q = 5, t = 2$. As in \ref{['fig:KikuchiGraph']}, vertices of the form $(x, 1)$ are colored green, while vertices of the form $(x, 2)$ are colored blue. The elements of $Q_\theta$ are colored yellow. Note that $q - t = 3$ is odd, and $S$ indicated by $(s,s')$ contains $2 = \left\lceil\frac{q - t}{2}\right\rceil$ green elements from $\left(C\setminus Q_\theta\right)\times\{1\}$, and $1 = \left\lfloor\frac{q - t}{2}\right\rfloor$ blue element from $\left(C'\setminus Q_\theta\right)\times\{2\}$.

Theorems & Definitions (53)

  • Theorem 1.1
  • Theorem 2.1
  • Definition 2.3: Weak Rainbow Even Covers
  • Remark 2.5: Prior works
  • Definition 2.6: Good index $t$
  • Definition 2.7: Approximate Strong Regularity
  • Lemma 2.8: Weak Rainbow Bound for Approximately Strongly Regular Hypergraphs
  • proof : Proof of \ref{['lem:even-cover-regular-hypergraph']}
  • Claim 2.9
  • proof : Proof of \ref{['thm:linear-LDCs']}
  • ...and 43 more