How is a connection the same as Morphism between different pullbacks to the first infinitesimal neighbourhood

54 Views Asked by At

This question arose from the first pages of Delignes Equations Differentielles.

There he defines a connection on a vector bundle $V$ on an complex analytic space $X$ as the following Data:

For all analytic spaces $S$ and pairs $(x,y)$ of $S$-points of $X$ (i.e. Morphisms from $S$ to $X$), such that the morphism $(x,y):S\longrightarrow X\times X$ factors through the first infinitesimal neighbourhood $X_1$ of the diagonal $X_0$: a morphism $\gamma_{x,y}:x^*V\longrightarrow y^*V$ s.t.:

  • For all morphisms $f:T\longrightarrow S$ we have $f^*\gamma_{x,y}=\gamma_{xf,yf}$
  • $\gamma_{x,x}=\operatorname{id}$

He then claims a little bit further that this is the same as a homomorphism $\gamma:p_1^*V\longrightarrow p_2^*V$ which induces the identity on $X_0$ and that this is automatically an isomorphism. (Here the $p_i$ denote the projections of $X_1$ to $X$).

For me this sounds like he uses the Yoneda Lemma here to derive this claim, somehow interpreting the definition of a connection as a natural transformation between a representable functor (probably represented by $X_1$?) and another one. I played around with this idea but wasn't able to make it precise, so my question is: Can we, and if yes how, interpret the claim he makes as an application of the Yoneda Lemma?