Inmathematics, in particularalgebraic topology, ap-compact group is ahomotopical version of acompactLie group, but with all the local structure concentrated at a singleprimep. This concept was introduced inDwyer & Wilkerson (1994), making precise earlier notions of a mod p finite loop space. A p-compact group has many Lie-like properties likemaximal tori andWeyl groups, which are defined purely homotopically in terms of the classifying space, but with the important difference that theWeyl group, rather than being a finitereflection group over the integers, is now a finitep-adic reflection group. They admit a classification in terms of root data, which mirrors the classification of compact Lie groups, but with the integers replaced by thep-adic integers.
Ap-compact group is a pointed spaceBG, which is local with respect to modphomology, and such the pointedloop spaceG = ΩBG has finite modp homology. One sometimes also refer to thep-compact group byG, but then one needs to keep in mind that the loop space structure is part of the data (which then allows one to recoverBG).
Ap-compact group is said to beconnected if G is aconnected space (in general the group of components of G will be a finite p-group). Therank of ap-compact group is the rank of its maximal torus.
The classification ofp-compact groups fromAndersen & Grodal (2009) states that there is a 1-1 correspondence between connectedp-compact groups, up to homotopy equivalence, androot data over thep-adic integers, up to isomorphism. This is analogous to the classical classification of connected compact Lie groups, with thep-adic integers replacing therational integers.
It follows from the classification that anyp-compact group can be written asBG = BH × BK whereBH is thep-completion of a compact connected Lie group and BK is finite direct product of simpleexotic p-compact groups i.e., simple p-compact groups whose Weyl group group is not a-reflection groups. Simple exotic p-compact groups are again in 1-1-correspondence with irreducible complex reflection groups whose character field can be embedded in, but is not.
For instance, whenp=2 this implies that every connected 2-compact group can be writtenBG = BH × BDI(4)s, where BH is the 2-completion of the classifying space of a connected compact Lie group, and BDI(4)s denotes s copies of the "Dwyer-Wilkerson 2-compact group" BDI(4) of rank 3, constructed inDwyer & Wilkerson (1993) with Weyl group corresponding to group number 24 in theShepard-Todd enumeration ofcomplex reflection groups. Forp=3 a similar statement holds but the new exotic 3-compact group is now group number 12 on the Shepard-Todd list, of rank 2. For primes greater than 3, family 2 on the Shepard-Todd list will contain infinitely many exotic p-compact groups.
Afinite loop space is a pointed space BG such that the loop space ΩBG is homotopy equivalent to a finite CW-complex. The classification of connected p-compact groups implies aclassification of connected finite loop spaces: Given a connected p-compact group for each prime, all with the same rational type, there is an explicit double coset space of possible connected finite loop spaces with p-completion the give p-compact groups. As connected p-compact groups are classified combinatorially, this implies a classification of connected loop spaces as well.
Using the classification, one can identify the compact Lie groups inside finite loop spaces, giving ahomotopical characterisation of compact connected Lie groups: They are exactly those finite loop spaces that admit an integral maximal torus; this was the so-calledmaximal torus conjecture. (SeeAndersen & Grodal (2009) andGrodal (2010).)
The classification also implies a classification of which graded polynomial rings can occur as the cohomology ring of a space, the so-calledSteenrod problem. (SeeAndersen & Grodal (2008).)