Zaichen Lu October 2009

. denoted A ∼ B. .j = hj (Yi. such that A ≈G B and B ≈G C. . . B ⊆ X. . B= i=1 Bi with Ai ∩ Aj = Bi ∩ Bj = Ø whenever i = j. hn ∈ G Let Yi. B. B1 . such that ∀g. h ∈ G and x ∈ X. . . . such such that Cj = hj (Bj ). j Yi. where 1 denotes the identity element in G. Proposition 1. Deﬁnition 1. Furthermore. . ∃g1 .the special orthogonal group. Deﬁnition 2. βn ⊆ B and C1 .j = (h ◦ g)(Xi. . I shall present some of the deﬁnitions and theorems I shall be using in the proof of the Banach-Tarski Paradox. . . Clearly the collection of Yi.j = Bi ∩ βj . .The group of all isometries. . . . . if A and B can be partitioned into ﬁnite disjoint unions of pairwise congruent Ai . By deﬁnition. this means that Bi = gi (Ai ).Background In this section.j } partitions C.j = gi (Yi. Am ⊆ A. and Zi.j also partitions B. A= i=1 n n G-equidecomposable. We shall be using the geometric concept of equidecomposability to represent this idea more technically. That is. if there is some isometry = n n f : R → R such that f (A) = B. we have Over the course of this paper. I shall be using the following notational conventions for groups acting on Rn : • En . gm ∈ G. j. Let G be a group acting on a set X and suppose A. . gn ∈ G such that Bi = gi (Ai ). Let X be a set. and transitivity doesn’t require much more eﬀort.j = βj for each j. the crux of the paradox is to generate two sets of identical ‘size’ as the original. . We deﬁne A and B to be subsets. there corresponds a bijection. and 2 . . Two sets A. .j = Bi for Zi. Bm ⊆ B and rigid motions g1 . • SOn . each i. . As mentioned in the introduction. The relation ≈G is an equivalence relation. . Deﬁnition 3. also denoted g : X ↔ X. B ⊆ Rn are congruent. Cn ⊆ C and h1 . C ⊆ Rn . . We denote this A ≈G B. and let G be a group. . Proof. g(h(x)) = (gh)(x) and 1(x) = x. i Yi. . . and similarly Transitivity: Suppose A.j ).j } partitions A. and that ∀ 1 ≤ i ≤ n. that there exists partitions A1 . and {Zi.j ). Then. −1 Let Xi. {Xi. . There are also partitions β1 .j ) for each i. G is said to act on X if to every g ∈ G. Reﬂexivity and symmetry are trivial. In R3 this consists of the rotations around the origin.

The identity of F will be denoted 1. s. we can understand BTP in the more general setting of paradoxical sets. and G a subgroup of En . if for some A. it is of particular interest when G is En or SOn . We can now phrase the Banach-Tarski Paradox using this terminology: BTP: S 2 is SO3 -paradoxical. no two reduced words are Figure 1: Cayley graph of free group on two generators. and called the empty word: clearly σσ −1 = σ −1 σ = 1. and A ≈G E ≈G B With this deﬁnition. Over the course of this paper we shall investigate certain particular cases when X is some Euclidean space. We deﬁne E to be Gparadoxical if E is G-equidecomposable with two disjoint copies of itself. Clearly. A ∩ B = Ø. since only rigid motions are then allowed. A word is reduced if equivalent. ∀σ ∈ M all adjacent pairs of letters of the form σσ −1 or σ −1 σ are removed. The center represents the empty word. That is. and suppose E ⊆ X.t. B ⊆ E. Let G be a group acting on a set X. The free group F with generating set M is the group of all ﬁnite reduced words using letters from {σ. σ −1 : σ ∈ M } under composition. each vertex representing a unique reduced word. Deﬁnition 5. τ ±1 . and moving along a branch represents multiplying by one of σ ±1 . Let M be a set. Thus. 3 .Deﬁnition 4.

for example. τ W (τ −1 ). . Suppose therefore. for n ≥ 3. where F acts on itself by left multiplication. Proof. and transformations which witness that G is paradoxical. that is. . . which is based on the realisation of the free group on two generators as a group of isometries in Rn .Theorem 1. Let G be a paradoxical group acting on a set X without nontrivial ﬁxed points. . no non-identity element of the group ﬁxes a point of the set. It is also useful to talk of groups having the property of being paradoxical. . and g1 . gives a word beginning with ρ. Hence SOn has a free subgroup of rank 2. τ are the two generators of F. For the following theorem. . σ and τ . . then they are independent. α and β generate a free group of rank 2. gm . Here. . Note that as G acts without nontrivial ﬁxed 4 . we obtain a word beginning with σσ −1 ρ . We deﬁne the orbit of p under G by o(p) := {g(p) : g ∈ G}. . τ ±1 . we need the extra condition that the group acts without nontrivial ﬁxed points.e of rank 2) is F-paradoxical. . Proof. that A1 . and also uses the idea of transferring a paradox from a group to a set upon which it acts. Am . . hn ∈ G are subsets Let p ∈ X. was shown to be paradoxical. Recall that a group is paradoxical if it is equidecomposable with two disjoint copies of itself. . which upon reduction. Suppose σ. . Proof. where the group acts on itself by left multiplication. where ρ is one of σ ±1 . note that if we pre-multiply an element in W (σ −1 ) by σ. A free group F on two generators (i. if α and β are two rotations of the same irrational angle θ (in degrees). This result is important to our dealing with BTP. Theorem 2 (AC).e. but through perpendicular axes. B1 . . it is the set of set for a free group H ≤ G. such that no non-trivial sequence of compositions results in the identity rotation. We denote the set of words in F whose ﬁrst letter is ρ as W (ρ). i. Bn ⊆ G. about axes through the origin in R3 . . . h1 . we deﬁne a subset S of a group G to be independent if S is a generating to the empty word. which by Theorem 1. We can then partition F into ﬁve sets: F = {1} W (σ) W (σ −1 ) W (τ ) W (τ −1 ) However. There exists two independent rotations. It is clear to see that. To transfer the paradoxical quality from a group to a set upon which it acts. That is: σW (σ −1 ) = {1} W (σ −1 ) W (τ ) We therefore obtain that: W (σ) σ(W σ −1 ) = F = W (τ ) W (τ −1 ) = F \ W (σ). . Proposition 2. Consequently. we shall be considering rotations in R3 . . This means that no non-trivial word composed of elements of S corresponds all points that can be reached from p by action from G. Then X is G-paradoxical. where ρ is one of σ −1 . τ ±1 .

5 . Then {g(M ) : g ∈ G} partitions X.points. without nontrivial ﬁxed points. by the paradoxical decomposition i m i=1 gi (Ai ) of G. Then G is paradoxical. Hence. n m gi (A∗ ) = X = i i=1 j=1 ∗ hj (Bj ) Corollary. any two orbits are either identical or disjoint. Then. if we consider the group of rotations about the origin in R2 . where we restrict attention to exclude nontrivial ﬁxed points. because G acts without nontrivial ﬁxed points. G is H -paradoxical. we have that =G= n j=1 hj (Bj ). we can see that all rotations leave the origin as ﬁxed. An immediate problem we encounter is that groups often have non-trivial ﬁxed points. X partitions into a disjoint union of orbits. Proof. Hausdorﬀ’s Paradox presents a precursor to BTP. Suppose H is a paradoxical subgroup of G. G is G-paradoxical. Then. Hence. For example. Formally. A subgroup H of a group G acts by left multiplication on the whole group. there exists a choice set M containing one element from each G-orbit in X. ∗ Let A∗ = {g(M ) : g ∈ Ai } and Bj = {g(M ) : g ∈ Bj }. any group with a free subgroup of rank 2 is paradoxical. Therefore. Let G be a group. By Theorem 3.