Quadratic embeddings

Abstract

The quadratic Veronese embedding $\rho$ maps the point set $P$ of $\PG{n,F)$ into the point set of $PG((\genfrac{}{}{0ex}{}{n+2}{2})-1,F$ ($F$ a commutative field) and has the following well-known property: If $M\subset P$, then the intersection of all quadrics containing $M$ is the inverse image of the linear closure of $M^{\rho }$. In other words, $\rho$ transforms the closure from quadratic into inear. In this paper we use this property to define "quadratic embeddings". We shall prove that if $\nu$ is a quadratic embedding of $PG{n,F)$ into $PG(n^{\prime },F^{\prime })$ ($F$ a commutative field), then ${\rho }^{-1}\nu$ is dimension-preserving. Moreover, up to some exceptional cases, there is an injective homomorphism of $F$ into $F^{\prime }$. An additional regularity property for quadratic embeddings allows us to give a geometric characterization of the quadratic Veronese embedding.