
Constant-dimension codes (CDCs) play a key role in random linear network coding, and a central problem is to construct CDCs with as many codewords as possible. The parallel construction has produced many strong lower bounds, but it applies primarily in the case k≥ d . In this paper, we introduce column-restricted rank-metric codes (CRMCs) and use them to extend the parallel construction to the case d>k , thereby improving several known lower bounds for CDCs.
For distinct primes p and q, the non-unitary rings H_pq and H_qp are introduced. The case (p,q)=(2,3) appeared in [1]. We study self-orthogonal codes over these rings with respect to the symplectic inner product. In addition, we explore symplectic linear complementary dual (symplectic-LCD), symplectic quasi self-dual (symplectic-QSD) and symplectic self-dual (symplectic-SD) codes over the rings H_pq and H_qp . An application of this study is given by classifying all symplectic self-orthogonal codes over these rings up to symplectic equivalence in short lengths for some primes p and q.
A (t, n)-threshold visual cryptography scheme provides a nonconventional secret sharing scheme for digital black-white images in which n shares are generated from a secret image and the secret image is reproduced by superimposition of more than or equal to t shares. In this paper, we focus on the relative difference α _n that represents the contrast between black and white pixels in a reproduced secret image when all the n shares are superimposed. We succeed in obtaining the tight upper bounds on α _n for the cases of t=2 and t=3 . In particular, the upper bound for the case of t=3 is obtained using a combinatorial argument. This upper bound for t=3 yields the optimal construction of the (3, n)-threshold visual cryptography scheme using a BIBD associated with a Hadamard matrix.
Evolving secret sharing (Komargodski, Naor, and Yogev, TCC 2016) allows a dealer to share a secret value in an online manner, without knowing the access structure or the maximum number of parties in advance, and without ever updating the shares of older players. In this paper, we continue the study of evolving secret sharing schemes in the computational setting, as first considered by Francati and Venturi (ASIACRYPT 2024), where the number of parties is upper bounded by an unknown polynomial and the privacy property only holds against computationally-bounded adversaries. We provide three main contributions. First, we formalize adaptive security for evolving secret sharing, introducing two natural notions—semi-adaptive and fully-adaptive—and establish a precise relationship between them. We show that semi-adaptive security implies fully-adaptive security with a loss that depends on the description size of the evolving access structure. For succinctly representable structures, this loss is polynomial. Second, we present generic compilers that transform non-evolving secret sharing schemes into evolving ones. Assuming one-way functions, we obtain evolving schemes in which the n-th share essentially aggregates the shares corresponding to the other n-1 parties; assuming trapdoor permutations in the random oracle model and starting from an adaptively secure non-evolving scheme, we obtain fully-adaptively secure evolving schemes, that incur the same aggregation cost, plus an additional overhead linear in the number of parties. Together, these results establish a natural upper bound on the share size of evolving secret sharing for natural families of access structures. Third, we prove that an existing construction for monotone circuits evolving access structures achieves adaptive security for constant-depth circuits, under standard cryptographic assumptions. Overall, our results provide a systematic treatment of adaptive security in evolving secret sharing and demonstrate that it can be generically achieved with controlled overhead under standard assumptions.
Locally recoverable codes with locality (r,δ ) (or (r,δ ) -LRCs) were introduced by Prakash et al. [33] to address the challenge of tolerating multiple node failures in distributed storage systems. In the special case δ =2 , (r, 2)-LRCs reduce to classical locally recoverable codes with locality r (or r-LRCs). More recently, quantum (r,δ ) -LRCs (abbreviated as (r,δ ) -qLRCs) have been developed by incorporating locality constraints into the framework of quantum information theory, motivated by their potential applications in large-scale quantum data storage [9, 11]. Let q be a prime power and m be a positive integer. In this paper, we construct two classes of (r,δ ) -LRCs of length λ (q+1) derived from cyclic codes. These codes are shown to be distance-optimal (or d-optimal) with respect to the Singleton-like bound for (r,δ ) -LRCs. Furthermore, we obtain two classes of (r,δ ) -qLRCs of length n | (q-1) from cyclic codes. These quantum codes are either d-optimal or almost d-optimal with respect to the corresponding Singleton-like bound for (r,δ ) -qLRCs. In addition, under the condition (q-1,m)=1 , we prove that the resulting Hamming r-qLRCs are dimension-optimal (or k-optimal) with respect to the Sphere-packing-like bound for r-LRCs obtained via the Euclidean construction, and are almost k-optimal with respect to the Cadambe–Mazumdar-like bound.