An optimal bound for embedding linear spaces into projective planes (Q1822758): Difference between revisions

The author considers finite linear spaces with v points and b lines. He continues studies in this field of \textit{R. G. Stanton} and \textit{J. G. Kalbfleisch [Proc. Conf. Comb., Chapel Hill, 451-462 (1972)], \textit{S. A. Vanstone}} [Proc. Conf. Numer. Math. Manitoba, 409-418 (1973; Zbl 0317.05015)] and himself and decides a conjecture of \textit{P. Erdős}, \textit{R. C. Mullin}, \textit{V. T. Sós} and \textit{D. R. Stinson} [Discrete Math. 47, 49-62 (1983; Zbl 0521.51005)] in the negative sense. A linear space is defined such that every two distinct points occur in a unique line and every line contains at least two points. A linear space is called a near pencil, if it has a line which contains all but one of the points of that space. Then classifying linear spaces with \(n^ 2- n+1<vv\leq b\leq n^ 2+n+1\) he receives that there is essentially one class of such linear spaces with no constant point degree which are not near pencils and which can not be embedded into any projective plane of order n. He formulates three theorems for a linear space being a near pencil or being not with \(v>n^ 2-n+1,\) \(v>n^ 2-(2/3)n+1\) and \(v>n^ 2-n+1\) points. Let A(n) be the least positive integer such that every linear space with \(A(n)=v\) \((\leq n^ 2+n+1)\) is a near pencil or is embeddable into a projective plane of order n. For small n he can show \(A(1)=A(2)=1\), \(A(3)=9\), and \(A(4)=16\). Generally he obtains the optimal bound \(A(n)=n^ 2-\sqrt{n}+2,\) if n is the order of a projective plane which has a Baer-subplane. In every other case it is \(A(n)\leq n^ 2- (1/6)n+1).\) His methods which are developed to prove his theorems also give a result characterizing blocking sets (of projective planes) which contain the points of a Baer-subplane.