um HAE TWARSACTIONS ON INFORMATION THEORY, OK, 3, NO. 5, SEPFEMBER 1988 Coding Theory Introduction JH. VAN LINT. ASSOCIATE EDITOR HERE ARE fourteen papers on algebraic and combi- nnatorial coding theory in this special issue. Theit subject matter mix attests that there are at least two hot topics currently, namely, covering radius problems (four papers) and algebraic geometric codes (\hree papers) moreover, several other papers on these two areas are in the pipeline, OF course, there are also a few papers on decoding algorithms. If-a binary code C has covering radius R. then the spheres of radius R around codewords cover the space F. By expressing this fact in terms of the group algebra QF, Calderbank and Sloane find linear inequalities involving the Lloyd polynomials via the Delsarte-MacWilliams in- equalities, This leads to two new values of rm, k), the smallest covering radius of any {n,k] code, namely 117,10] = 3 and 1[23,15]= 5. MacWilliams’ equation also plays a role in the paper by Simonis, who shows that 1[15,6]=4 using a combination of this equation and argu: ments from finite geometry, Previous upper bounds on. K(ni,R), the minimal cardinality of a binary code of Tength n and covering radius R, are improved by a new family of normal noalinear codes given by Honkala and Hamilinen. Their idea is first to construct the code consisting of all words with weights in a set 17 satisfying carefully chosen conditions, then to double the length by the rules 0-»00 and 111 or 0->10 and 1-01, and finally t0 delete a coordinate. The paper by Pach and Spencer is ina different vein, Probabilistic and asymptotic iiethods are used to find a series of linear codes with both low covering radius and a rapid decoding algorithm. For a survey of results on covering radius problems that ap- peared either as a paper or as a preprint in 1986-1988, see [2] Recently several papers have appeared extending Goppa’s construction of codes using algebraic curves. For an introduction to the subject, see (3]- Quebbeman consid fers what he calls a “eyclotomic extension” E of the ratio- nal function field F,(X). He shows that E has small genus ‘compared to its number of F,rational points. This leads to ‘a sequence of codes with a transitive automorphism group. [Although the codes are not “good” in the usual asymptotic sense, they are not bad either. Wirtz found a gap in the original proof of Goppa on the parameters of subfield subcodes of generalized Goppa codes. This is repaired by fan ingenious induetion argument. Stichtenoth considers codes defined using Hermitian curves. The results extend those of Tiersma [5]. As was the case for Tiersma’s paper, this one is useful for those readers who are stil gaining familiarity with this area, since it is not too deep and it ‘explicitly finds generator matrices and results on weight distribution ‘There are two papers related to Berlekamp-Massey decoding of BCH codes. Stevens decodes beyond the BCH bound by using not only the known values of power sums bbut also a number of them that can take on only a few (unknown) values, The method seems useful for parallel ‘computing. The paper also contains several important ta- bles. Shiozaki's correspondence is essentially about decod- ing Reed-Solomon codes. By a slight variation on the standard method, involving an error-locator and ertor- evaluator, the final step (which is again a ged calculation) produces the information polynomial directly instead of the error evaluator. The amount of computation seems to be the same as in Berlekamp-Massey decoding, The paper by Boly and van Gils was inspired by the (N, K concept fault-tolerant computers (ef, []]). The codes involve as symbols elements of Fm represented as mr tuples over F,. The codes are used 10 handle single symbol ertors, multiple bit errors, and combinations of symbol erasures and bit errors, Extensive tables are given. Etzion’s paper concerns two-dimensional generalizations of De Bruijn sequences, namely, rv doubly periodic arrays (where ro = 2") with the so-called “n x m window property," ie, each mm binary matrix appears exactly ‘once in one period of the array. Some generalizations are also considered. Weber, de Vroedt, and Boekee consider codes for the Z-channel, a binary channel where the 1-0 error occurs With probability p and the 0-1 error essentially does not ‘occur. There have been several papers in the IT TRANSAC~ TIONS on this subject plus others by Delsarte and iret, Klove, Shiozaki, and Varshamov. Inthe present paper new upper bounds on the maximum number of codewords in & linear code of length and asymmetric distance 4 are given, Construction methods for such codes with small length ate discussed and several tables are included In [4] McEliece applied some sophisticated methods from number theory to the study of eyelic codes. These ideas are used by Remijn and Tiersma to prove that the weight enumerators of two seemingly unrelated codes,namely, a [q” +1,2m] irreducible cyclic code and a certain Ig” —1.2m] eyelic code over F, are connected by a simple formula, The case g=2 had been shown by Dillon and later by Dir. In the meantime an extension of these results| has been submitted to the IT TRANSACTIONS by Lachaud and Wolfman, Finally, the correspondence by Leon deserves special mention. A combined effort of several editors, referees, and of the author himself has resulted in a period of four and a half years between submission and publication! By ‘means of a probabilistic algorithm, the minimum weight of very large codes is determined with a certain probability of error (usually very small). Similar ideas have been used recently by other authors when the codes were too difficult to handle completely. 1 feel privileged to have served as the Associate Editor for Coding Theory, and on balance I enjoyed the experi- ams ence. Among the benefits is that, even if I had not been editor, T would have intended to read at least ren of the above fourteen papers, but itis not certain that this would really have happened. Now, I have read all of them and hhope that others will do the samme. REFERENCES (0) T Krol, "(,) conept fale tolerance,” IEEE Tran. Comput, Vol C35, pp 389-349, 186 (2) IH wae Lin, “Recent reals on covering problems,” Pre AAECC, Rome, 1988 (3) 1H van Lint and TA. Springer, “Generaiond Reed-Solomon fades fom algebralcseometty” FEE Tran Inform. Thecry. vo. 1733, pp 308-309, 1987 Ta) RoI, Metiece, "ietucible cyte codes and Gauss sams” in Combinatorics (4. Hal JH. van Liat Bas, Math Cone ‘Trace 38, Mathematch Centr, Amsterdass, 1924, pp. 179-196 [5] 1 Tersma, "Remarks on codes from Herman cuss,” IEEE Trans. Inform. Thay, vl TT.33 pp. 605-€8, ly 1987