Characterization of \(\{2(q+1)+2,2;t,q\}\)-\(\min \cdot hypers\) in PG(t,q) (t\(\geq 3,q\geq 5)\) and its applications to error-correcting codes (Q1124583): Difference between revisions

The Griesmer bound for a linear (n,k)-code over GF(q) with minimum distance d has the form \(n\geq \sum^{k-1}_{\ell =0}\lceil d/q^{\ell}\rceil,\) where \(\lceil x\rceil\) denotes the smallest integer \(\geq x\). Among the linear codes with fixed parameters k, d and q it is of interest to determine those which satisfy the Griesmer bound with equality and have thus minimum length n. Earlier work by the former author proved that the solution of the above problem for a wide range of parameters amounts to the determination of certain configurations of points in a projective space called min\(\cdot hypers\). A non-empty set F of points of a projective space PG(t,q) with \(t\geq 2\) is said to be an \(\{\) \(| F|,m;t,q\}\)-min\(\cdot hyper\) for a non-negative integer m if it meets each hyperplane of PG(t,q) in at least m points and there exists one hyperplane of PG(t,q) having exactly m points in common with F. The former paper characterizes all \(\{2(q+1)+2,2;t,q\}\)-min\(\cdot hypers\) for any integer \(t\geq 3\) nd any prime power \(q\geq 5:\) they consist namely of two skew lines and two further points outside them. Given a prime power q define \(v_{\ell}:=(q^{\ell}-1)/(q-1)\) for each non-negative integer \(\ell\). Let t, q and \(\epsilon\) be integers with \(1\leq \mu <t,\) \(0\leq \epsilon <\sqrt{q}.\) The latter paper (see the paper below) characterizes all \(\{v_{\mu +1},v_ 1;t,q\}\)-min\(\cdot hypers:\) they consist namely of a \(\mu\)-dimensional subspace and \(\epsilon\) points not on it (if instead q is a square, \(\mu =1\) and \(\sqrt{q}\leq \epsilon <q\) then \(\{v_{\mu +1},v_ 1;t,q\}\)-min\(\cdot hypers\) exist which are not of the described type: for \(t=2\) one can simply take a Baer subplane of PG(2,q)). The two papers are quite similar in the structure, the proofs have to distinguish many cases, but they involve basically only counting arguments in finite projective spaces. Characterizations of all linear codes satisfying the Griesmer bound with equality are thus obtained in the former paper for \(k\geq 3\), \(d=q^{k- 1}-(2+2q),\) \(q\geq 5\) and in the latter one for \(1\leq \mu <k-1,\) \(d=q^{k-1}-(\epsilon +q^{\mu})\) and \(0\leq \epsilon <\sqrt{q}.\)