nLab proper morphism


For the concept in topology see at proper map.



Between topological spaces

While every continuous map sends compact subsets to compact subsets (see at continuous images of compact spaces are compact), it is not true in general that the preimage of a compact set along a continuous map is compact.

A continuous function f:XYf : X \to Y between topological spaces is a proper map if it is closed and the inverse image of any compact subset is itself compact.

A notion of proper homotopy between proper maps leads to proper homotopy theory.

Similarly, one can consider the conditions on morphisms in other geometric situations, like algebraic geometry, and properness often either reflects or implies good behaviour with respect to the compact objects (cf. also proper push-forward).

Between schemes

A proper morphism of schemes is by definition a morphism f:XYf:X\to Y which is

  1. separated,

  2. of finite type

  3. universally closed (the latter means that for every h:ZYh: Z\to Y the pullback h *(f):Z× YXZh^*(f): Z\times_Y X\to Z is closed).

There is a classical and very practical valuative criterion of properness due Chevalley.

We say that a scheme XX is proper if the canonical map XSpecX \to \operatorname{Spec} \mathbb{Z} to the terminal object is proper.

Proper schemes are analogous to compact topological spaces. This is one reason why one uses the terminology “quasi-compact” when referring to schemes whose underlying topological space is compact.

The base change formulas for cohomology for proper and for smooth morphisms of schemes motivated Grothendieck (in Pursuing Stacks) to define abstract proper and smooth functors in the setting of fibered categories; this is further expanded on in (Maltsiniotis).

Between locales

Recall that a locale LL is given by a frame O(L)O(L), its frame of opens, and that a continuous map ff from KK to LL is given by an adjunction f *f *:O(K)O(L)f^* \dashv f_* \colon O(K) \rightleftarrows O(L) such that the inverse image function f *f^* preserves finitary meets (or equivalently is a frame homomorphism, since it must preserve all joins).

Such a map ff is proper iff the direct image function f *f_* preserves directed joins (or equivalently is Scott-continuous, or equivalently is a morphism of preframes), and also satisfies the Frobenius reciprocity-like condition that f *(Uf *(V))=f *(U)Vf_*(U\cup f^*(V)) = f_*(U) \cup V (which by itself states that the map is closed).

In particular, the map L1L\to 1 is proper iff LL is both compact and covert. But in this case the second condition is redundant, since every compact locale is automatically covert; see covert space for a proof.

Proper maps of locales can also be characterized as those that are universally closed, i.e. every pullback of them (along any map of locales) is closed.

Between toposes

Proper maps of locales can be generalized to geometric morphisms of Grothendieck toposes; see proper geometric morphism.

The topos-theoretic condition refers only to directed unions of subterminal objects, suggesting a stronger condition that it preserve all filtered colimits. This is a strictly stronger condition even for locales (i.e. localic toposes), called being tidy. In fact properness and tidiness are the first two rungs on an infinite ladder of higher properness for higher toposes.


(Inconsistent terminology)
In the world of 1-toposes, the concept of proper map splits into two subconcepts

tidyproper \mathrm{tidy} \Rightarrow \mathrm{proper}

with the stronger concept being called “tidy”.

More generally in \infty-toposes, one gets an infinite tower of related definitions

proper...1proper0proper \infty-\mathrm{proper} \Rightarrow ... \Rightarrow 1-proper \Rightarrow 0-proper

This is where the inconsistency arises: in \infty-topos theory it is customary to call “proper morphism” the strongest notion while in 1-topos theory it is customary to use it for the weaker notion. Hence, a proper map of spaces f:XYf\,:\, X \to Y will induce a proper map of 1-toposes but not a proper map of \infty-toposes (without extra assumptions).


  • Georges Maltsiniotis, Structures d’asphéricité, foncteurs lisses, et fibrations, Ann. Math. Blaise Pascal 12, pp. 1-39 (2005) (ps).

(TO ADD: The definition of a proper dg algebra, proper dg category, proper A-inf cat ???)

Last revised on November 15, 2023 at 09:46:48. See the history of this page for a list of all contributions to it.