11.07.2015 Views

International Congress of Mathematicians

International Congress of Mathematicians

International Congress of Mathematicians

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

ICM 2002 • Vol. II • 373-384The Topology <strong>of</strong> Out(F n )Mladen Bestvina*AbstractWe will survey the work on the topology <strong>of</strong> Out(F n ) in the last 20 yearsor so. Much <strong>of</strong> the development is driven by the tantalizing analogy withmapping class groups. Unfortunately, Out(F n ) is more complicated and lesswell-behaved.Culler and Vogtmann constructed Outer Space X n , the analog <strong>of</strong> Teichmüllerspace, a contractible complex on which Out(F n ) acts with finitestabilizers. Paths in X n can be generated using "foldings" <strong>of</strong> graphs, an operationintroduced by Stallings to give alternative solutions for many algorithmicquestions about free groups. The most conceptual pro<strong>of</strong> <strong>of</strong> the contractibility<strong>of</strong> X n involves folding.There is a normal form <strong>of</strong> an automorphism, analogous to Thurston's normalform for surface homeomorphisms. This normal form, called a "(relative)train track map", consists <strong>of</strong> a cellular map on a graph and has good propertieswith respect to iteration. One may think <strong>of</strong> building an automorphismin stages, adding to the previous stages a building block that either growsexponentially or polynomially. A complicating feature is that these blocks arenot "disjoint" as in Thurston's theory, but interact as upper stages can mapover the lower stages.Applications include the study <strong>of</strong> growth rates (a surprising feature <strong>of</strong> freegroup automorphisms is that the growth rate <strong>of</strong> / is generally different fromthe growth rate <strong>of</strong> / _1 ), <strong>of</strong> the fixed subgroup <strong>of</strong> a given automorphism, andthe pro<strong>of</strong> <strong>of</strong> the Tits alternative for Out(F n ). For the latter, in addition totrain track methods, one needs to consider an appropriate version <strong>of</strong> "attractinglaminations" to understand the dynamics <strong>of</strong> exponentially growingautomorphisms and run the "ping-pong" argument. The Tits alternative isthus reduced to groups consisting <strong>of</strong> polynomially growing automorphisms,and this is handled by the analog <strong>of</strong> Kolchin's theorem (this is one instancewhere Out(F n ) resembles GL„(Z) more than a mapping class group).Morse theory has made its appearance in the subject in several guises.The original pro<strong>of</strong> <strong>of</strong> the contractibility <strong>of</strong> X n used a kind <strong>of</strong> "combinatorial"Morse function (adding contractible subcomplexes one at a time and studyingthe intersections). Hatcher-Vogtmann developed a "Cerf theory" for graphs.This is a parametrized version <strong>of</strong> Morse theory and it allows them to provehomological stability results. One can "bordify" Outer Space (by analogywith the Borei-Serre construction for arithmetic groups) to make the action'Department <strong>of</strong> Mathematics, University <strong>of</strong> Utah, USA. E-mail: bestvina@math.utah.edu

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!