A combinatorial characterization of smooth LTCs and applications

The study of locally testable codes (LTCs) has benefited from a number of nontrivial constructions discovered in recent years. Yet, we still lack a good understanding of what makes a linear error correcting code locally testable and as a result we do not know what is the rate‐limit of LTCs and whether asymptotically good linear LTCs with constant query complexity exist. In this paper, we provide a combinatorial characterization of smooth locally testable codes, which are locally testable codes whose associated tester queries every bit of the tested word with equal probability. Our main contribution is a combinatorial property defined on the Tanner graph associated with the code tester (“well‐structured tester”). We show that a family of codes is smoothly locally testable if and only if it has a well‐structured tester. As a case study we show that the standard tester for the Hadamard code is “well‐structured,” giving an alternative proof of the local testability of the Hadamard code, originally proved by (Blum, Luby, Rubinfeld, J. Comput. Syst. Sci. 47 (1993) 549–595) (STOC 1990). Additional connections to the works of (Ben‐Sasson, Harsha, Raskhodnikova, SIAM J. Comput 35 (2005) 1–21) (SICOMP 2005) and of (Lachish, Newman and Shapira, Comput Complex 17 (2008) 70–93) are also discussed. © 2016 Wiley Periodicals, Inc. Random Struct. Alg., 49, 280–307, 2016     

Testing algebraic geometric codes

Property testing was initially studied from various motivations in 1990’s. A code C  GF (r)n is locally testable if there is a randomized algorithm which can distinguish with high possibility the codewords from a vector essentially far from the code by only accessing a very small (typically constant) number of the vector’s coordinates. The problem of testing codes was firstly studied by Blum, Luby and Rubinfeld and closely related to probabilistically checkable proofs (PCPs). How to characterize locally te...     

A combination of testability and decodability by tensor products

Ben‐Sasson and Sudan (RSA 2006) showed that taking the repeated tensor product of linear codes with very large distance results in codes that are locally testable. Due to the large distance requirement the associated tensor products could be applied only over sufficiently large fields. Then Meir (SICOMP 2009) used this result to present a combinatorial construction of locally testable codes with largest known rate. As a consequence, this construction was obtained over sufficiently large fields. In this paper we improve the result of Ben‐Sasson and Sudan and show that for any linear codes the associated tensor products are locally testable. Consequently, the construction of Meir can be taken over any field, including the binary field. Moreover, a combination of our result with the result of Spielman (IEEE IT, 1996) implies a construction of linear codes (over any field) that combine the following properties:

• have constant rate and constant relative distance;
• have blocklength n and are testable with n^? queries, for any constant ? > 0;
• linear time encodable and linear‐time decodable from a constant fraction of errors.

Furthermore, a combination of our result with the result of Guruswami et al. (STOC 2009) implies a similar corollary for list‐decodable codes. © 2013 Wiley Periodicals, Inc. Random Struct. Alg., 46, 572–598, 2015     

Hermitian codes as generalized Reed-Solomon codes

Hermitian codes obtained from Hermitian curves are shown to be concatenated generalized Reed-Solomon codes. This interpretation of Hermitian codes is used to investigate their structure. An efficient encoding algorithm is given for Hermitian codes. A new general decoding algorithm is given and applied to Hermitian codes to give a decoding algorithm capable of decoding up to the full error correcting capability of the code.This work is supported by a Natural Science and Engineering Research Council Grant A7382.     

Perfect codes from the dual point of view I

The two concepts dual code and parity check matrix for a linear perfect 1-error correcting binary code are generalized to the case of non-linear perfect codes. We show how this generalization can be used to enumerate some particular classes of perfect 1-error correcting binary codes. We also use it to give an answer to a problem of Avgustinovich: whether or not the kernel of every perfect 1-error correcting binary code is always contained in some Hamming code.     

Traceability codes

Traceability codes are combinatorial objects introduced by Chor, Fiat and Naor in 1994 to be used in traitor tracing schemes to protect digital content. A k-traceability code is used in a scheme to trace the origin of digital content under the assumption that no more than k users collude. It is well known that an error correcting code of high minimum distance is a traceability code. When does this ‘error correcting construction’ produce good traceability codes? The paper explores this question.Let ? be a fixed positive integer. When q is a sufficiently large prime power, a suitable Reed-Solomon code may be used to construct a 2-traceability code containing q⌈?/4⌉ codewords. The paper shows that this construction is close to best possible: there exists a constant c, depending only on ?, such that a q-ary 2-traceability code of length ? contains at most cq⌈?/4⌉ codewords. This answers a question of Kabatiansky from 2005.Barg and Kabatiansky (2004) asked whether there exist families of k-traceability codes of rate bounded away from zero when q and k are constants such that q?k². These parameters are of interest since the error correcting construction cannot be used to construct k-traceability codes of constant rate for these parameters: suitable error correcting codes do not exist when q?k² because of the Plotkin bound. Kabatiansky (2004) answered Barg and Kabatiansky's question (positively) in the case when k=2. This result is generalised to the following: whenever k and q are fixed integers such that k?2 and q?k²−⌈k/2⌉+1, or such that k=2 and q=3, there exist infinite families of q-ary k-traceability codes of constant rate.     

Binary permutation sequences as subsets of Levenshtein codes,spectral null codes,run-length limited codes and constant weight codes

We investigate binary sequences which can be obtained by concatenating the columns of (0,1)-matrices derived from permutation sequences. We then prove that these binary sequences are subsets of a surprisingly diverse ensemble of codes, namely the Levenshtein codes, capable of correcting insertion/deletion errors; spectral null codes, with spectral nulls at certain frequencies; as well as being subsets of run-length limited codes, Nyquist null codes and constant weight codes. This paper was presented in part at the IEEE Information Theory Workshop, Chengdu, China, October, 2006.     

A number theoretic formula and asymptotic optimality of cardinalities of BAD correcting codes

This paper presents a formula for the cardinality of a class of non-linear error correcting codes for Balanced Adjacent Deletions that are provided as an extension of standard deletion from the point of the view of Weyl groups. Furthermore, we show that the cardinality is approximately optimal over any single BAD correcting codes. In other words, the ratio of the cardinality of the code and that of maximum cardinality BAD correcting code converges to 1 for sufficiently large length.     

Stepsize selection for tolerance proportionality in explicit Runge–Kutta codes

The potential for adaptive explicit Runge–Kutta (ERK) codes to produce global errors that decrease linearly as a function of the error tolerance is studied. It is shown that this desirable property may not hold, in general, if the leading term of the locally computed error estimate passes through zero. However, it is also shown that certain methods are insensitive to a vanishing leading term. Moreover, a new stepchanging policy is introduced that, at negligible extra cost, ensures a robust global error behaviour. The results are supported by theoretical and numerical analysis on widely used formulas and test problems. Overall, the modified stepchanging strategy allows a strong guarantee to be attached to the complete numerical process. This revised version was published online in June 2006 with corrections to the Cover Date.     

Families of twisted tensor product codes

Using geometric properties of the variety ${\mathcal V_{r,t}}$ , the image under the Grassmannian map of a Desarguesian (t ? 1)-spread of PG(rt ? 1, q), we introduce error correcting codes related to the twisted tensor product construction, producing several families of constacyclic codes. We determine the precise parameters of these codes and characterise the words of minimum weight.     

On the construction of perfect deletion-correcting codes using design theory

We consider a way to construct perfect codes capable of correcting 2 or more deletions using design-theory. As a starting point we use an (ordered) block design to construct a perfect deletion correcting code. Using this code we are able to construct more perfect deletion correcting codes over smaller or larger alphabets by removing or adding symbols in a smart way.In this way we are able to find all perfect 2-deletion correcting codes of length 4, and all perfect 3-deletion correcting codes of length 5 with different coordinates. The perfect 3-deletion correcting codes of length 5 with repeated symbols can be constructed for almost all possible alphabet sizesv, except forv=13, 14, 15, and 16, and forv7, 8 (mod 10),v17. For these values ofv we are neither able to prove the existence, nor the non-existence of perfect 3-deletion correcting codes of length 5 over an alphabet of sizev.     

Lower bounds for adaptive locally decodable codes

An error‐correcting code is said to be locally decodable if a randomized algorithm can recover any single bit of a message by reading only a small number of symbols of a possibly corrupted encoding of the message. Katz and Trevisan 12 showed that any such code C : {0, 1}ⁿ → Σ^m with a decoding algorithm that makes at most q probes must satisfy m = Ω((n/log |Σ|)^q/(q?1)). They assumed that the decoding algorithm is non‐adaptive, and left open the question of proving similar bounds for adaptive decoders. We show m = Ω((n/log |Σ|)^q/(q?1)) without assuming that the decoder is nonadaptive. © 2005 Wiley Periodicals, Inc. Random Struct. Alg., 2005     

Improved decoding of affine-variety codes

General error locator polynomials are polynomials able to decode any correctable syndrome for a given linear code. Such polynomials are known to exist for all cyclic codes and for a large class of linear codes. We provide some decoding techniques for affine-variety codes using some multidimensional extensions of general error locator polynomials. We prove the existence of such polynomials for any correctable affine-variety code and hence for any linear code. We propose two main different approaches, that depend on the underlying geometry. We compute some interesting cases, including Hermitian codes. To prove our coding theory results, we develop a theory for special classes of zero-dimensional ideals, that can be considered generalizations of stratified ideals. Our improvement with respect to stratified ideals is twofold: we generalize from one variable to many variables and we introduce points with multiplicities.     

On perfect deletion‐correcting codes

In this article, we present constructions for perfect deletion‐correcting codes. The first construction uses perfect deletion‐correcting codes without repetition of letters to construct other perfect deletion‐correcting codes. This is a generalization of the construction shown in 1 . In the third section, we investigate several constructions of perfect deletion‐correcting codes using designs. In the last section, we investigate perfect deletion‐correcting codes containing few codewords. © 2003 Wiley Periodicals, Inc.     

A unified framework for testing linear‐invariant properties

In the study of property testing, a particularly important role has been played by linear invariant properties, i.e., properties of Boolean functions on the hypercube which are closed under linear transformations of the domain. Examples of such properties include linearity, Reed‐Muller codes, and Fourier sparsity. In this work, we describe a framework that can lead to a unified analysis of the testability of all linear‐invariant properties, drawing on techniques from additive combinatorics and from graph theory. Our main contributions here are the following:

相似文献   

Symmetric LDPC codes and local testing

Coding theoretic and complexity theoretic considerations naturally lead to the question of generating symmetric, sparse, redundant linear systems. This paper provides a new way of construction with better parameters and new lower bounds.Low Density Parity Check (LDPC) codes are linear codes defined by short constraints (a property essential for local testing of a code). Some of the best (theoretically and practically) used codes are LDPC. Symmetric codes are those in which all coordinates “look the same,” namely there is some transitive group acting on the coordinates which preserves the code. Some of the most commonly used locally testable codes (especially in PCPs and other proof systems), including all “low-degree” codes, are symmetric. Requiring that a symmetric binary code of length n has large (linear or near-linear) distance seems to suggest a “con ict” between 1/rate and density (constraint length). In known constructions, if one is constant, then the other is almost the worst possible - n/poly(logn).Our main positive result simultaneously achieves symmetric low density, constant rate codes generated by a single constraint. We present an explicit construction of a symmetric and transitive binary code of length n, near-linear distance n/(log logn)², of constant rate and with constraints of length (logn)⁴. The construction is in the spirit of Tanner codes, namely the codewords are indexed by the edges of a sparse regular expander graph. The main novelty is in our construction of a transitive (non Abelian!) group acting on these edges which preserves the code. Our construction is one instantiation of a framework we call Cayley Codes developed here, that may be viewed as extending zig-zag product to symmetric codes.Our main negative result is that the parameters obtained above cannot be significantly improved, as long as the acting group is solvable (like the one we use). More specifically, we show that in constant rate and linear distance codes (aka “good” codes) invariant under solvable groups, the density (length of generating constraints) cannot go down to a constant, and is bounded below by (log^(Ω(?)) n)(an Ω(?) iterated logarithm) if the group has a derived series of length ?. This negative result precludes natural local tests with constantly many queries for such solvable “good” codes.     

On transform-domain error and erasure correction by Gabidulin codes

Gabidulin codes are the rank metric analogues of Reed–Solomon codes and have found many applications including network coding. In this paper, we propose a transform-domain algorithm correcting both errors and erasures with Gabidulin codes. Interleaving or the direct sum of Gabidulin codes allows both decreasing the redundancy and increasing the error correcting capability for network coding. We generalize the proposed decoding algorithm for interleaved Gabidulin codes. The transform-domain approach allows to simplify derivations and proofs and also simplifies finding the error vector after solving the key equation.     

Generalized Balanced Tournament Packings and Optimal Equitable Symbol Weight Codes for Power Line Communications

Generalized balanced tournament packings (GBTPs) extend the concept of generalized balanced tournament designs introduced by Lamken and Vanstone (1989). In this paper, we establish the connection between GBTPs and a class of codes called equitable symbol weight codes (ESWCs). The latter were recently demonstrated to optimize the performance against narrowband noise in a general coded modulation scheme for power line communications. By constructing classes of GBTPs, we establish infinite families of optimal ESWCs with code lengths greater than alphabet size and whose narrowband noise error‐correcting capability to code length ratios do not diminish to zero as the length grows.