Is there an easy direction for the Higgs correspondence?

73 Views Asked by At

There is a deep famous correspondence between analytic and algebraic properties.

For a complex curve $X$, representations $Hom(\pi_1(X), U(n))$ correspond to degree $0$ semistable bundles. This is a hard theorem but one direction is easy; given a representation you can easily build the bundle with flat connection and holomorphic metric and thus see it's semistable.

It turns out that representations $Hom(\pi_1(X), GL(n))$ correspond to semistable Higgs bundle.

I want whatever easy part of this correspondence you can tell me; given a representation we can build a vector bundle with a flat connection, but how do we build the map $V \to V \otimes(K)$? Note that $H^0(V,V \otimes(K)) = H^1(End(V))^*$ is the cotangent space of $V \in Bund(X)$.

From this story, we see that we expect there to be a map from the cotangent space of representations $T^*Hom(\pi_1,U(n)) \to Hom(\pi_1(X), GL(n))$! For a representation $\rho$, we can describe the former cotangent space as $$H^1(\pi_1 , Ad(\rho)$$. Given $e \in H^1(\pi_1 , Ad(\rho)$, how do we get an element of $Hom(\pi_1(X), GL(n))$?

Warning; I've heard that these equivalent spaces have different complex structure, so maybe what I'm asking is unreasonable because the cotangent spaces aren't canonically identified?

1

There are 1 best solutions below

0
On BEST ANSWER

One direction which makes the theorem vaguely make sense is:

Given a representation $\rho:\pi_1(X) \to GL_n$ we get a bundle $V$ with a connection $\nabla$. Let's suppose this $V$ is stable, so that we can find a unique unitary representation $\rho'$ giving a flat unitary connection $\nabla'$ (Whatever that means). Then $\nabla - \nabla'$ is the desired twisted endomorphism $V \to V \otimes K$! Thus if we believe the (hard) theorem that says semistable bundles come from unitary representations we can easily define one direction of the Higgs bundle.