Perfect Obstruction Theory Defn

definition

[GP99, pg. 4].

See also appendix C of Ricolfi.

Definition

Let \(f:X\to Y\) be a morphism of algebraic stacks over some base stack \(S\). A perfect obstruction theory for \(f\), or as it is more commonly denoted, a perfect obstruction theory \(X/Y\), is a two term complex \(E^\bullet = [E^{-1}\xrightarrow{d} E^0]\) together with a morphism

\begin{align*} \phi:E^\bullet \to L_{X/Y} \end{align*}

to the relative cotangent complex of \(f\) such that

  1. \(h^0(f)\) is an isomorphism
  2. \(h^{-1}(f)\) is surjective.

We call \((\phi, E^\bullet)\) an absolute perfect obstruction theory if \(Y = S\). We will sometimes call \((\phi,E^\bullet)\) a relative perfect obstruction theory for \(f:X\to Y\) when \(Y\neq S\) to contrast against the absolute situation.

When a perfect obstruction theory for \(f\) exists, we say that \(f\) is virtually smooth, or that \(X\) is virtually smooth over \(Y\).

Perfect obstruction theories are precisely the data needed to define virtual fundamental classes and virtual pullbacks.

See also obstruction-theory-defn.

References 1