nLab homotopy inverse

Definition

Let $\sim$ be the relation of being homotopic (for example between morphisms in the category Top). Let $f:X\to Y$ and $g:Y\to X$ be two morphisms. We say that $g$ is a left homotopy inverse to $f$ or that $f$ is a right homotopy inverse to $g$ if $g\circ f\sim id_X$. A homotopy inverse of $f$ is a map which is simultaneously a left and a right homotopy inverse to $f$.

$f$ is said to be a homotopy equivalence if it has a left and a right homotopy inverse. In that case we can choose the left and right homotopy inverses of $f$ to be equal. To show this denote by $g_L$ the left and by $g_R$ the right homotopy inverse of $f$. Then

$$g_L \sim g_L\circ (f\circ g_R) = (g_L\circ f)\circ g_R \sim g_R.$$

Hence

$$f\circ g_L\sim f\circ g_R\sim id_X,$$

therefore $g_L$ is not only a left, but also a right, homotopy inverse to $f$.

This makes sense in any category equipped with an equivalence relation $\sim$, which is compatible with the composition (and with the equality of morphisms).

Examples

• A functor which is an equivalence of categories is/has a homotopy inverse with respect to the notion of homotopy given by natural isomorphisms.

• A weak homotopy equivalence of topological spaces between CW-complexes is a homotopy equivalence and hence has a homotopy inverse.

• A weak homotopy equivalence of simplicial sets between Kan complexes is a homotopy equivalence and hence has a homotopy inverse.