This dissertation consists of three papers in combinatorial Coxeter group theory.

A Coxeter group is a group W generated by a set S, where all relations can be derived from the relations s2 = e for all s ?? S, and (ss?)m(s,s?) = e for some pairs of generators s ? s? in S, where e ?? W is the identity element and m(s, s?) is an integer satisfying that m(s, s?) = m(s?, s) ? 2. Two prominent examples of Coxeter groups are provided by the symmetric group Sn (i.e., the set of permutations of {1, 2, . . . , n}) and finite reflection groups (i.e., finite groups generated by reflections in some real euclidean space). There are also important infinite Coxeter groups, e.g., affine reflection groups.

Every Coxeter group can be equipped with various natural partial orders, the most important of which is the Bruhat order. Any subset of a Coxeter group can then be viewed as an induced subposet.

In Paper A, we study certain posets of this kind, namely, unions of conjugacy classes of involutions in the symmetric group. We obtain a complete classification of the posets that are pure (i.e., all maximal chains have the same length). In particular, we prove that the set of involutions with exactly one fixed point is pure, which settles a conjecture of Hultman in the affirmative. When the posets are pure, we give their rank functions. We also give a short, new proof of the EL-shellability of the set of fixed-point-free involutions, established by Can, Cherniavsky, and Twelbeck.

Paper B also deals with involutions in Coxeter groups. Given an involutive automorphism ? of a Coxeter system (W, S), let

?(?) = {w ?? W | ?(w) = w?1}

be the set of twisted involutions. In particular, ?(id) is the set of ordinary involutions in W. It is known that twisted involutions can be represented by words in the alphabet = { | s ?? S}, called -expressions. If ss? has finite order m(s, s?), let a braid move be the replacement of ? ? by ? ? ?, both consisting of m(s, s?) letters. We prove a word property for ?(?), for any Coxeter system (W, S) with any ?. More precisely, we provide a minimal set of moves, easily determined from the Coxeter graph of (W, S), that can be added to the braid moves in order to connect all reduced -expressions for any given w ?? ?(?). This improves upon a result of Hamaker, Marberg, and Pawlowski, and generalises similar statements valid in certain types due to Hu, Zhang, Wu, and Marberg.

In Paper C, we investigate the topology of (the order complexes of) certain posets, called pircons. A special partial matching (SPM) on a poset is a matching of the Hasse diagram satisfying certain extra conditions. An SPM without fixed points is precisely a special matching as defined by Brenti. Let a pircon be a poset in which every non-trivial principal order ideal is finite and admits an SPM. Thus pircons generalise Marietti’s zircons. Our main result is that every open interval in a pircon is a PL ball or a PL sphere.

An important subset of ?(?) is the set ??(?) = {?(w?1)w | w ?? W} of twisted identities. We prove that if ? does not flip any edges with odd labels in the Coxeter graph, then ??(?), with the order induced by the Bruhat order on W, is a pircon. Hence, its open intervals are PL balls or spheres, which confirms a conjecture of Hultman. It is also demonstrated that Bruhat orders on Rains and Vazirani’s quasiparabolic W-sets (under a boundedness assumption) form pircons. In particular, this applies to all parabolic quotients of Coxeter groups.