Abstract:We prove that a binary linear code of block length $n$ that is locally correctable with $3$ queries against a fraction $\delta > 0$ of adversarial errors must have dimension at most $O_{\delta}(\log^2 n \cdot \log \log n)$. This is almost tight in view of quadratic Reed-Muller codes being a $3$-query locally correctable code (LCC) with dimension $\Theta(\log^2 n)$. Our result improves, for the binary field case, the $O_{\delta}(\log^8 n)$ bound obtained in the recent breakthrough of (Kothari and Manohar, 2023) (
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: in the presence of adversarial errors, what is the maximum dimension of 3 - query Locally Correctable Binary Linear Codes (3 - LCCs). Specifically, the author proves that for a binary linear code of length \(n\), if it can be locally corrected using 3 queries in the case of an adversarial error proportion of \(\delta>0\), then its dimension is at most \(O_\delta(\log^2 n\cdot\log\log n)\). This result is almost optimal because the quadratic Reed - Muller code is a 3 - query locally correctable code with a dimension of \(\Theta(\log^2 n)\).
### Main Contributions
1. **Improved Upper Bound**: The paper improves the previous upper bound on 3 - query locally correctable codes over the binary field, from \(O_\delta(\log^8 n)\) to \(O_\delta(\log^2 n\cdot\log\log n)\).
2. **Direct Method**: Different from the previous method of indirectly proving by constructing 2 - query Locally Decodable Codes (LDCs), this paper adopts a more direct method, proving the result by restricting the covering radius of the dual code.
3. **Application of Rainbow Cycles**: The paper utilizes the concept of "rainbow cycles" in graph theory, finding rainbow cycles in appropriately colored graphs to analyze the local correction property.
### Technical Details
- **Definition of Locally Correctable Codes**: A binary linear code \(C\subseteq\{0, 1\}^n\) is called an \((r, \delta)\)-locally correctable code if any given bit can be recovered with high probability by randomly querying \(r\) positions, even if there are no more than \(\delta n\) errors in these positions.
- **Core Theorem**: The core theorem of the paper is proved by restricting the covering radius of the dual code. Specifically, if \(C\) is a \((3, \delta)\)-locally correctable code, then each vector \(x\in\mathbb{F}_2^k\) can be represented as a linear combination of at most \(O_\delta(\log n\cdot\log\log n)\) row vectors of the generator matrix.
- **Use of Rainbow Cycles**: By finding rainbow cycles in appropriately colored graphs, the paper proves that the length of the linear combination can be effectively compressed, thus achieving the required upper bound.
### Conclusion
By introducing new techniques and methods, the paper significantly improves the upper bound of 3 - query locally correctable binary linear codes, providing a new perspective for understanding the structure and limitations of such codes. This result is not only of great theoretical significance but also provides guidance for the design of error - correction coding in practical applications.