.

## The Ends of Algebras

### November 27, 2006

Abstract
We introduce the notion of ends for algebras. The definition is analogous to the one in geometric group theory. We establish some relations to growth conditions and cyclic cohomology.
Keywords: ends of algebras, ends of groups, Gelfand-Kirillov dimension, cyclic cohomology

1 Introduction and preliminaries

The goal of this paper is to introduce and study the notion of ends for algebras. First let us recall the definition of ends of groups (see e.g. Freudenthal (1945), Epstein (1962)).
Let $\Gamma$  be a group, then we call $P\subset \Gamma$  an almost invariant set if for any $g\in \Gamma$  , $Pg$  and $P$  differ only by finitely many elements. That is $\left\{\left(Pg\cup P\right)\P\right\}$  is a finite set. Obviously, if $P$  is almost invariant, then $\Gamma \P$  is almost invariant as well. If two almost invariant sets differ only by finitely many elements then we call them equivalent, $P\simeq Q$  . The equivalence classes of almost invariant sets are the endsets. We say that the group $\Gamma$  has $k$  ends if it can be partitioned into the disjoint union of $k$  infinite almost invariant sets, but it can not be partitioned into the disjoint union of $k+1$  infinite almost invariant sets. If for any $k\ge 1$  , the group has such partitions, then $\Gamma$  has infinitely many ends. By the theorem of Hopf (1944), a group may have $0,1,2$  or $\infty$  many ends. Finite groups have zero ends, the group ${\mathbb{Z}}^{2}$  has one end, the group $\mathbb{Z}$  has two ends and the free group of two generators has infinitely many ends.
Now let $K$  be a commutative field and $\text{}\mathcal{A}\text{}$  be a unital $K$  -algebra. We say that a $K$  -linear subspace $V\subset \text{}\mathcal{A}\text{}$  is an almost invariant subspace if for any $x\in \text{}\mathcal{A}\text{}$  , $\left(Vx+V\right)/V$  is finite dimensional.
Definition 1.1 The algebra $\text{}\mathcal{A}\text{}$  has $k$  ends if $\text{}\mathcal{A}\text{}={V}_{1}\oplus {V}_{2}\oplus ...\oplus {V}_{k}$  , where $\left\{{V}_{i}{\right\}}_{i=1}^{k}$  are infinite dimensional almost invariant subspaces, but $\text{}\mathcal{A}\text{}$  can not be written as the direct sum of $k+1$  infinite dimensional almost invariant subspaces. We say that $\text{}\mathcal{A}\text{}$  has infinitely many ends, if for any $k\ge 1$  , $\text{}\mathcal{A}\text{}$  has such direct sum decomposition. We apply the following convention; finite dimensional algebras have zero ends.
Let us see some examples.
Example 1 $\text{}\mathcal{A}\text{}=K\left[x\right]$  , the polynomial ring of one variable.
Let $S\subset \text{}\mathcal{A}\text{}$  be an infinite dimensional almost invariant subspace. Let $n>0$  , such that $Sx+S\subset S+\text{span}K\left\{1,x,{x}^{2},...,{x}^{n}\right\}$  . Suppose that ${p}_{m}\in S$  is a polynomial of degree $m>n$  . Then ${p}_{m}x\in S+\text{span}K\left\{1,x,{x}^{2},...,{x}^{n}\right\}$  . That is $S$  contains a polynomial of degree $m+1$  .
Inductively, $S$  contains a polynomial of degree $k$  for any $k\ge m$  . Hence $K\left[x\right]=S+\text{span}K\left\{1,x,{x}^{2},...,{x}^{m-1}\right\}$  . Therefore $K\left[x\right]$  has only one end.
Example 2 $\text{}\mathcal{A}\text{}=K\left[x,{x}^{-1}\right]=K\mathbb{Z}$  .
If $S\subset \text{}\mathcal{A}\text{}$  is an infinite dimensional almost invariant subspace, then our previous example shows that $K\left[x\right]\cap S$  is either finite or cofinite dimensional, similarly, $K\left[{x}^{-1}\right]\cap S$  is either finite or cofinite dimensional. Hence $K\left[x,{x}^{-1}\right]$  has two ends.
Example 3 The $*$  -algebra ${\Sigma }_{n}\left(K\right)={\oplus }_{1\le i\le n}K\left[{x}_{i}\right]$
One can easily prove that if $\text{}\mathcal{A}\text{}$  and $\text{}\mathcal{ℬ}\text{}$  are algebras, having respectively $k$  and $l$  ends then $\text{}\mathcal{A}\text{}\oplus \text{}\mathcal{ℬ}\text{}$  has $k+l$  ends. Thus ${\Sigma }_{n}\left(K\right)$  has exactly $n$  ends.
Example 4 The algebra $\text{}\mathcal{A}\text{}=K\left[x,y\right]/J$  , where the ideal $J$  is generated by the monomials of $x$  and $y$  containing at least $2$  $y$  's.
Note that ${S}_{k}=\text{s}pan\left\{{x}^{k},{x}^{k}y,{x}^{k}yx,{x}^{k}y{x}^{2},...\right\}$  is an almost invariant subspace. Then $\text{}\mathcal{A}\text{}={S}_{1}\oplus {S}_{2}\oplus ...{S}_{n}\oplus \left\{K\oplus {S}_{n+1}\oplus {S}_{n+2}...\right\}$  is a decomposition of $\text{}\mathcal{A}\text{}$  into the direct sum of $n+1$  infinite dimensional almost invariant subspaces. Hence $\text{}\mathcal{A}\text{}$  has infinitely many ends.
We shall define endclasses of algebras, similarly to the endsets of groups, and identify the endclasses with the idempotents of a certain natural extension of our algebras. Thus one-endedness of a certain algebra is equivalent to the fact that this extension has no non-trivial idempotents. Houghton (1972) proved that if a group is infinite then the number of ends is exactly the dimension of the first cohomology group ${H}^{1}\left(\Gamma ,\mathbb{C}\Gamma \right)$  plus one. Using a Fredholm-module construction we associate to the endclasses of $\text{}\mathcal{A}\text{}$  elements of the first cyclic cohomology group $H{C}^{1}\left(\text{}\mathcal{A}\text{}\right)$  . We also prove that algebras of Gelfand-Kirillov dimension $1$  have only finitely many ends. Finally, we study the case of group algebras.

2 The Fredholm extension

Let $K$  be a field and ${K}^{\infty }$  be the countable dimensional vector space over $K$  . Consider the endomorphism ring ${\text{End}}_{K}\left({K}^{\infty }\right)$  . Let $I\subset {\text{End}}_{K}\left({K}^{\infty }\right)$  be the ideal of endomorphisms of finite rank. If $T\in I$  then let $\stackrel{~}{T}$  be the restriction of $T$  onto the range of $T$  . Then $\stackrel{~}{T}$  is a finite dimensional linear transformation hence $Trace\left(\stackrel{~}{T}\right)$  is well-defined. If ${T}^{\prime }$  is the restriction of $T$  onto a finite dimensional space containing the range of $T$  , then of course $Trace\left(\stackrel{~}{T}\right)=Trace\left({T}^{\prime }\right)$  . We shall denote $Trace\left(\stackrel{~}{T}\right)$  by $\text{Tr}\left(T\right)$  . Obviously, $\text{Tr}:I\to K$  is a linear functional.
Lemma 2.1 If $A\in {\text{End}}_{K}\left({K}^{\infty }\right)$  and $B\in I$  , then $\text{Tr}\left(AB\right)=\text{Tr}\left(BA\right)$  .
Proof. Let $\left\{{e}_{i}{\right\}}_{i\ge 1}$  be a basis of ${K}^{\infty }$  such that ${e}_{1},{e}_{2},...,{e}_{n}$  span the range of $B$  and the span of $\left\{{e}_{1},{e}_{2},...,{e}_{m}\right\}$  contains the range of $AB$  , where $m>n$  . Let $B{e}_{i}={\sum }_{j=1}^{\infty }{b}_{ij}{e}_{j}$  , $A{e}_{i}={\sum }_{j=1}^{\infty }{a}_{ij}{e}_{j}$  . Then ${b}_{ij}=0$  if $j>n$  and ${a}_{ij}=0$  , if $j>m$  and $i\le n$  . Hence ${a}_{ij}{b}_{ji}=0$  , if either $i>m$  or $j>m$  . Thus $\text{Tr}\left(AB\right)={\sum }_{i=1}^{m}{\sum }_{j=1}^{\infty }{a}_{ij}{b}_{ji}={\sum }_{i=1}^{m}{\sum }_{j=1}^{m}{a}_{ij}{b}_{ji}={\sum }_{i=1}^{m}{\sum }_{j=1}^{\infty }{a}_{ij}{b}_{ji}=\text{Tr}\left(BA\right)$  Now let us suppose that $\text{}\mathcal{A}\text{}$  is a countable dimensional unital $K$  -algebra. For $x\in \text{}\mathcal{A}\text{}$  , denote by ${l}_{x}$  (resp. ${r}_{x}$  ) the left (resp. right) multiplications by $x$  . Obviously, if $T\in {\text{End}}_{K}\left(\text{}\mathcal{A}\text{}\right)$  and $T$  commutes with ${r}_{x}$  for any $x\in \text{}\mathcal{A}\text{}$  , then $T={l}_{T\left(1\right)}$  . We can identify the set $\left\{{l}_{x}{\right\}}_{x\in \text{}\mathcal{A}\text{}}$  with $\text{}\mathcal{A}\text{}$  itself and the set $\left\{{r}_{x}{\right\}}_{x\in \text{}\mathcal{A}\text{}}$  with the opposite algebra $\text{}\mathcal{A}{\text{}}^{o}$  . Let $\stackrel{~}{\text{}\mathcal{A}\text{}}=\left\{T\in {\text{End}}_{K}\left(\text{}\mathcal{A}\text{}\right)|T{r}_{x}-{r}_{x}T\in I,\text{if}x\in \text{}\mathcal{A}\text{}\right\}.$  Obviously $\text{}\mathcal{A}\text{}\subset \stackrel{~}{A},I\subset \stackrel{~}{A}$  and $\stackrel{~}{A}$  is a unital subalgebra of ${\text{End}}_{K}\left(\text{}\mathcal{A}\text{}\right)$  .
Definition 2.1 The Fredholm extension of $\text{}\mathcal{A}\text{}$  is defined as $\text{Fred}\left(\text{}\mathcal{A}\text{}\right)=\stackrel{~}{\text{}\mathcal{A}\text{}}/I$  .
Note that if for any $0\ne x\in \text{}\mathcal{A}\text{}$  , ${l}_{x}$  is of infinite dimensional range then $\pi :\text{}\mathcal{A}\text{}\to \text{Fred}\left(\text{}\mathcal{A}\text{}\right)$  is injective. If $\text{}\mathcal{A}\text{}$  is the polynomial ring $K\left[x\right]$  , then $\text{Fred}\left(\text{}\mathcal{A}\text{}\right)$  is just the Toeplitz-algebra.
Proposition 2.1 Suppose that $x$  is a Fredholm element,that is both $\text{Ker}\left({l}_{x}\right)$  and $\text{Coker}\left({l}_{x}\right)$  is finite dimensional, then $\left[x\right]$  is invertible in $\text{Fred}\left(\text{}\mathcal{A}\text{}\right)$  .
Proof. First of all note that if $W\subseteq \text{}\mathcal{A}\text{}$  is a finite codimensional subspace and $T\in {\text{End}}_{K}\left(\text{}\mathcal{A}\text{}\right)$  , then there exists a finite codimensional subspace ${W}_{T}\subset W$  such that $T\left({W}_{T}\right)\subseteq W$  . Indeed, consider the linear map ${\pi }_{W}\circ T:W\to \text{}\mathcal{A}\text{}/W.$  Then ${W}_{T}=\text{Ker}\left({\pi }_{W}\circ T\right)$  is clearly finite codimensional and $T\left({W}_{T}\right)\subseteq W$  . Now let ${W}^{x}\subset \text{}\mathcal{A}\text{}$  be a subspace, $\text{Ker}\left({l}_{x}\right)\oplus {W}^{x}=\text{}\mathcal{A}\text{}$  . Then ${l}_{x}{|}_{{W}^{x}}$  is injective.
Consider the finite codimensional subspace ${l}_{x}\left({W}^{x}\right)$  . All elements in ${l}_{x}\left({W}^{x}\right)$  can be uniquely written as $xw$  , where $w\in {W}^{x}$  . Let $\text{}\mathcal{A}\text{}={l}_{x}\left({W}^{x}\right)\oplus B$  , where $B$  is finite dimensional. Define $T\in {\text{End}}_{K}\left(\text{}\mathcal{A}\text{}\right)$  to be $0$  on $B$  and $T\left(xw\right)=w$  on ${l}_{x}\left({W}^{x}\right)$  . Then $T\in \stackrel{~}{\text{}\mathcal{A}\text{}}$  . Indeed, let $s\in \text{}\mathcal{A}\text{}$  and let ${W}_{{r}_{s}}^{x}\subset {W}^{x}$  be a finite codimensional space as above, that is ${r}_{s}\left({W}_{{r}_{s}}^{x}\right)\subset {W}^{x}$  . If $w\in {W}_{{r}_{s}}^{x}$  , then $T{r}_{s}\left(xw\right)=T\left(xws\right)=ws\text{and}{r}_{s}T\left(xw\right)=ws.$  Hence, $\left[T\right]$  is the inverse of $\left[x\right]$  in $\text{Fred}\text{}\mathcal{A}\text{}$  Let $\text{}\mathcal{A}\text{}=V\oplus W$  be a decomposition of $\text{}\mathcal{A}\text{}$  into the direct sum of two almost invariant subspaces, where $V$  is infinite dimensional . A decomposition $\text{}\mathcal{A}\text{}={V}^{\prime }\oplus {W}^{\prime }$  is equivalent to the decomposition above, if $V$  and ${V}^{\prime }$  respectively $W$  and ${W}^{\prime }$  differ only by a finite dimensional subspace. That is $\left(V+{V}^{\prime }\right)/V$  , $\left(V+{V}^{\prime }\right)/{V}^{\prime }$  , $\left(W+{W}^{\prime }\right)/W$  and $\left(W+{W}^{\prime }\right)/{W}^{\prime }$  are all finite dimensional. We call the equivalence classes of such decompositions the endclasses of the algebra $\text{}\mathcal{A}\text{}$  . Note that $\text{}\mathcal{A}\text{}=V\oplus W$  and $\text{}\mathcal{A}\text{}=W\oplus V$  are determining different endclasses. Of course, if $V$  is an almost invariant subspace it might not have a complementary almost invariant subspace at all. The following example shows that in some pathological situations an almost invariant subspace can represent more than one endclasses.
Example 5 $\text{}\mathcal{A}\text{}={\text{End}}_{K}\left({K}^{\infty }\right)$  .
Let $\left\{{e}_{i}{\right\}}_{i=1}^{\infty }$  be a basis of ${K}^{\infty }$  . Let $A=\text{span}K\left\{{e}_{1},{e}_{3},{e}_{5},...\right\},B=\text{span}K\left\{{e}_{2},{e}_{4},{e}_{6},...\right\},C=\text{span}K\left\{{e}_{1}+{e}_{2},{e}_{3}+{e}_{4},{e}_{5}+{e}_{6},...\right\}$  .
For $X\subset {K}^{\infty }$  , let ${N}_{X}=\left\{T\in {\text{End}}_{K}{K}^{\infty }|T\left(v\right)=0\text{for any}v\in X\text{}\right\}.$  Then clearly, ${N}_{A},{N}_{B}$  and ${N}_{C}$  are pairwise inequivalent almost invariant subspaces, and ${\text{End}}_{K}{K}^{\infty }={N}_{A}\oplus {N}_{B}={N}_{A}\oplus {N}_{C}={N}_{B}\oplus {N}_{C}.$
Theorem 1 The nonzero idempotents of $\text{Fred}\left(\text{}\mathcal{A}\text{}\right)$  are in bijective correspondence with the endclasses of $\text{}\mathcal{A}\text{}$  .
Proof. Let $\text{}\mathcal{A}\text{}=V\oplus W$  be a decomposition of our algebra into the direct sum of almost invariant subspaces. Let ${\pi }_{V}$  (resp. ${\pi }_{W}$  ) be the projection onto $V$  (resp. $W$  ). Then ${\pi }_{V}$  is an idempotent in $\text{}\mathcal{A}\text{}$  . Indeed if $x\in \text{}\mathcal{A}\text{}$  , then by the observation in Proposition  2.1 , there exist ${R}_{x}^{V}\subseteq V$  , ${R}_{x}^{W}\subseteq W$  finite codimensional subspaces such that ${r}_{x}{R}_{x}^{V}\subseteq V$  and ${r}_{x}{R}_{x}^{W}\subseteq W$  . Hence if $a\in {R}_{x}^{V}$  , then ${r}_{x}{\pi }_{V}\left(a\right)={\pi }_{V}\left(ax\right)=ax$  , ${r}_{x}{\pi }_{W}\left(a\right)={\pi }_{W}\left(ax\right)=0$  , thus $\left[{\pi }_{V},{r}_{x}\right]\in I$  . If $A={V}^{\prime }\oplus {W}^{\prime }$  an equivalent decomposition then clearly $\left[{\pi }_{V}\right]=\left[{\pi }_{{V}^{\prime }}\right]$  in $\text{Fred}\left(\text{}\mathcal{A}\text{}\right)$  . Now, if $P\in \text{Fred}\left(\text{}\mathcal{A}\text{}\right)$  is a non-zero idempotent, then let $\stackrel{~}{P}\in \stackrel{~}{\text{}\mathcal{A}\text{}}$  be a representative of $P$  and $\stackrel{~}{Q}\in \stackrel{~}{\text{}\mathcal{A}\text{}}$  be a representative of $1-P$  . It is easy to check that $\text{Ran}\left(\stackrel{~}{P}\right)$  and $\text{Ran}\left(\stackrel{~}{Q}\right)$  are almost invariant subspaces, such that $\text{Ran}\left(\stackrel{~}{P}\right)+\text{Ran}\left(\stackrel{~}{Q}\right)$  is finite codimensional in $\text{}\mathcal{A}\text{}$  and $\text{Ran}\left(\stackrel{~}{P}\right)\cap \text{Ran}\left(\stackrel{~}{Q}\right)$  is finite dimensional. Hence there exist almost invariant subspaces $V,W$  , such that $\text{}\mathcal{A}\text{}=V\oplus W$  and $V$  is equivalent to $\text{Ran}\left(\stackrel{~}{P}\right)$  , $W$  is equivalent to $\text{Ran}\left(\stackrel{~}{P}\right)$  , where $V$  is infinite dimensional. Then $\left[{\pi }_{V}\right]=P$  , $\left[{\pi }_{W}\right]=1-P$  , thus the Theorem follows.

3 The noncommutative geometry of the ends

In this section we apply the Fredholm module construction of Connes (1994). First recall the notion of a graded differential algebra of length $1$  . Let $\text{}\mathcal{A}\text{}$  be a unital $K$  -algebra and $\Omega$  be an $\text{}\mathcal{A}\text{}$  -bimodule with a $K$  -linear map $\int :\Omega \to K$  and a $K$  -homomorphism $d:\text{}\mathcal{A}\text{}\to \Omega$  satisfying the following conditions
• $d\left(ab\right)=a\cdot db+da\cdot b$  , if $a,b\in \text{}\mathcal{A}\text{}$  .
• $\int a\omega =\int \omega a$  , if $a\in \text{}\mathcal{A}\text{}$  , $\omega \in \Omega$  .
• $\int da=0$  , if $a\in \text{}\mathcal{A}\text{}$  .
Then $\tau :\text{}\mathcal{A}\text{}{\otimes }_{K}\text{}\mathcal{A}\text{}\to K$  , $\tau \left(a,b\right)=\int a\cdot db$  defines a cyclic cohomology of $\text{}\mathcal{A}\text{}$  , $\tau \in H{C}^{1}\left(\text{}\mathcal{A}\text{}\right)$  .
Indeed, one can easily check that $\tau \left(a,b\right)=-\tau \left(b,a\right)$  and $\tau \left(ab,c\right)-\tau \left(a,bc\right)+\tau \left(ca,b\right)=0.$  Also, $\tau$  defines a character ${\phi }_{\tau }:G{L}_{1}\left(\text{}\mathcal{A}\text{}\right)\to K$  , by ${\phi }_{\tau }\left(a\right)=\tau \left({a}^{-1},a\right)$  .
The classical example is of course the algebra of smooth complex valued functions on the unit circle. Then, $\tau \left(f,g\right)=\frac{1}{2\pi i}\int f\left(z\right)dg\left(z\right).$  ${\phi }_{\tau }\left(f\right)=\frac{1}{2\pi i}\int \frac{{f}^{\prime }\left(z\right)}{f\left(z\right)}dz$  is just the winding number of $f$  .
Now let $\text{}\mathcal{A}\text{}$  be a finitely generated $K$  -algebra and let $\text{}\mathcal{A}\text{}=V\oplus W$  be a decomposition of $\text{}\mathcal{A}\text{}$  into the direct sum of two infinite dimensional almost invariant subspaces. Let $F$  be defined the following way, $F\left(v\right)=v$  , if $v\in V$  , $F\left(w\right)=w$  , if $w\in W$  . That is $F\in \stackrel{~}{\text{}\mathcal{A}\text{}}$  is an endomorphism such that ${F}^{2}=1$  . Then $\left(\text{}\mathcal{A}\text{},I,F\right)$  is a Fredholm-module, where $I\in {\text{End}}_{K}\left(\text{}\mathcal{A}\text{}\right)$  is the trace ideal. The coboundary operator $d:\text{}\mathcal{A}\text{}\to I$  is given by $dx=\left[F,{r}_{x}\right]$  , $\int :I\to K$  is defined as $\int \omega =\text{Tr}\left(\omega \right)$  .
Clearly, $d\left(xy\right)=dx\cdot y+x\cdot dy$  and $\int a\omega =\int \omega a$  . One should only check that $\int da=0$  , if $a\in \text{}\mathcal{A}\text{}$  . Pick a basis $\left\{{e}_{i}{\right\}}_{i=1}^{\infty }$  for $V$  and a basis $\left\{{f}_{i}{\right\}}_{i=1}^{\infty }$  for $W$  such that the span of $\left\{{e}_{1},{e}_{2},...,{e}_{n},{f}_{1},{f}_{2},...,{f}_{n}\right\}$  contains the range of $\left[F,{r}_{a}\right]$  . Let ${e}_{i}a={v}_{i}+{w}_{i}$  , ${f}_{i}a={v}_{i}^{\prime }+{w}_{i}^{\prime }$  , where ${v}_{i},{v}_{i}^{\prime }\in V$  and ${w}_{i},{w}_{i}^{\prime }\in W$  . Then $\left[F,{r}_{a}\right]{e}_{i}=-2{w}_{i}$  and $\left[F,{r}_{a}\right]{f}_{i}=2{v}_{i}$  , hence $\text{Tr}\left[F,{r}_{a}\right]$  must be zero. Therefore we associated a first cyclic cohomology class in $H{C}^{1}\left(\text{}\mathcal{A}\text{}\right)$  to the decomposition. Now if $\text{}\mathcal{A}\text{}={V}^{\prime }\oplus {W}^{\prime }$  is an equivalent decomposition then $F-{F}^{\prime }\in I$  . Thus, $\int a\left(Fb-bF\right)-\int a\left({F}^{\prime }b-b{F}^{\prime }\right)=\int \left(a\left(F-{F}^{\prime }\right)b-ab\left(F-{F}^{\prime }\right)\right)=\int \left(ab-ba\right)\left(F-{F}^{\prime }\right).$  This shows that $\tau -{\tau }^{\prime }$  is a cyclic coboundary. Hence we proved the following theorem.
Theorem 2 If $\left\{A,I,F\right\}$  is a triple as above associated to a decomposition $\text{}\mathcal{A}\text{}=V\oplus W$  , then $p\left(a,b\right)=\text{Tr}\left(a\left[F,b\right]\right)$  is a cyclic cocycle. The cyclic cohomology element represented by $p$  in $H{C}^{1}\left(\text{}\mathcal{A}\text{}\right)$  depends only on the endclass associated to the decomposition.
Finally, let us show that at least in some cases one can obtain non-trivial cyclic cohomologies.
Let $K\left[x,{x}^{-1}\right]=\text{}\mathcal{A}\text{}=V\oplus W$  , where $V=\text{span}K\left\{1,x,{x}^{2},...\right\}$  , $W=\left\{{x}^{-1},{x}^{-2},...\right\}$  . Let us calculate ${\phi }_{\tau }\left(x\right)=\text{Tr}\left({x}^{-1}\left[F,{r}_{x}\right]\right)$  . Clearly $\left(F{r}_{x}-{r}_{x}F\right)\left({x}^{k}\right)=0$  if $k\ne -1$  . On the other hand $\left(F{r}_{x}-{r}_{x}F\right)\left({x}^{-1}\right)=2$  thus $\text{Tr}\left({x}^{-1}\left[F,{r}_{x}\right]\right)=2.$  That is the associated cyclic cohomology is non-trivial.

4 Algebras of linear growth

In this section $\text{}\mathcal{A}\text{}$  shall be a finitely generated $K$  -algebra. Let ${t}_{1},{t}_{2},...{t}_{n}$  be a generator system of $\text{}\mathcal{A}\text{}$  containing the unit. Let ${S}_{1}$  be the vector space spanned by $\left\{{t}_{1},{t}_{2},...{t}_{n}\right\}$  .
Set ${S}_{2}={S}_{1}{S}_{1},{S}_{3}={S}_{2}{S}_{1}$  , inductively ${S}_{n+1}={S}_{n}{S}_{1}$  . Then ${S}_{1}\subset {S}_{2}\subset ...$  are finite dimensional vector spaces and ${\cup }_{n=1}^{\infty }{S}_{n}=\text{}\mathcal{A}\text{}$  . Now we briefly review some basic notions on the Gelfand-Kirillov dimension (see Krause and Lenagan (2000)). The growth function of $\text{}\mathcal{A}\text{}$  , relative to the generator system is the function given by $f\left(n\right):=dim\left({S}_{n}\right)$  , and the Gelfand-Kirillov dimension is $GKdim\left(\text{}\mathcal{A}\text{}\right)=limsup\frac{log\left(f\left(n\right)\right)}{logn}.$  Note that the Gelfand-Kirillov dimension does not depend on the choice of the generators.
If $GKdim\left(\text{}\mathcal{A}\text{}\right)=1$  , then by Bergman's theorem, $\text{}\mathcal{A}\text{}$  in fact has linear growth, that is $f\left(n\right)\le cn$  , for some real number $c>0$  . Bergman also showed that the linear growth of an algebra is equivalent to the existence of a constant $K>0$  , such that $f\left(n+1\right)-f\left(n\right)  .
By definition, if $V\subseteq A$  is an almost invariant subspace then there exists ${k}_{V}>0$  such that $V{S}_{1}\subseteq V+{S}_{{k}_{V}}$  .
Lemma 4.1 Let $\text{}\mathcal{A}\text{}=V\oplus W$  be a direct sum decomposition into almost invariant subspaces, then for some $l>0$  , ${S}_{n}\subseteq \left(V\cap {S}_{n+l}\right)\oplus \left(W\cap {S}_{n+l}\right)$  , holds for any $n>0$  .
Proof. Let $k>0$  be an integer such that $V{S}_{1}\subseteq V+{S}_{k}$  , $W{S}_{1}\subseteq W+{S}_{k}$  and let $j>k$  be an integer such that ${S}_{k}\subseteq \left(V\cap {S}_{j}\right)\oplus \left(W\cap {S}_{j}\right)$  . If $z\in {S}_{k}$  , $z=v+w,v\in V\cap {S}_{j},w\in W\cap {S}_{j}$  , then $z{t}_{i}=v{t}_{i}+w{t}_{i}$  , where $v{t}_{i}\in \left(V\cap {S}_{j+1}\right)+{S}_{k}$  and ${t}_{i}w\in \left(W\cap {S}_{j+1}\right)+{S}_{k}$  . Hence ${S}_{k+1}\subseteq \left(V\cap {S}_{j+1}\right)\oplus \left(V\cap {S}_{j+1}\right)$  . Inductively, ${S}_{k+n}\subseteq \left(V\cap {S}_{j+n}\right)\oplus \left(V\cap {S}_{j+n}\right)$  .
We call a subspace $V\subset \text{}\mathcal{A}\text{}$  sparse if for infinitely many integer $m$  , $V\cap {S}_{m}=V\cap {S}_{m+1}$  .
Lemma 4.2 If $\text{}\mathcal{A}\text{}$  is a finitely generated algebra and $V\subset \text{}\mathcal{A}\text{}$  is a sparse almost invariant subspace, then $V$  is finite dimensional.
Proof. Let $\text{}\mathcal{A}\text{}=V\oplus W$  as above , $m>j$  and $V\cap {S}_{m}=V\cap {S}_{m+1}$  . Suppose that $z\in {S}_{k}$  , $z=v+w$  , $v\in V\cap {S}_{m}$  , $w\in W\cap {S}_{m+1}$  . Then $z{t}_{i}\in \left(V\cap {S}_{m+1}\right)\oplus \left(W\cap {S}_{m+2}\right)=\left(V\cap {S}_{m}\right)\oplus \left(W\cap {S}_{m+2}\right).$  Inductively, ${S}_{n}\subseteq \left(V\cap {S}_{m}\right)\oplus \left(W\cap {S}_{m+n}\right)$  .
Hence, $\left(V\cap {S}_{m}\right)\oplus W=\text{}\mathcal{A}\text{}$  , thus $V\subset {S}_{m}$  .
Theorem 3 If $\text{}\mathcal{A}\text{}$  is a finitely generated algebra of Gelfand-Kirillov dimension $1$  , then $\text{}\mathcal{A}\text{}$  has only finitely many ends. Also if the growth function satisfies $f\left(n\right)\le Cn$  , then $\text{}\mathcal{A}\text{}$  has at most $C$  ends.
Proof. Suppose that for some $C>0$  , ${dim}_{K}{S}_{n}\le Cn$  and $\text{}\mathcal{A}\text{}={V}_{1}\oplus {V}_{2}\oplus ...{V}_{k}$  and $\left\{{V}_{i}{\right\}}_{i=1}^{k}$  are infinite dimensional almost invariant subspaces. Then none of the ${V}_{i}$  's are sparse. Hence, for some $m>0$  , ${V}_{i}\cap \left({S}_{m+1}\{S}_{m}\right)\ne \varnothing$  , for all $1\le i\le k$  .
Thus ${dim}_{K}{S}_{n}\ge k\left(n-m\right)$  , therefore $k\le C$  .
Groups of linear growth must have two ends, however it follows from Example 3. that the number of ends of algebras of linear growth may reach any positive integer. Groups of non-exponential growth have only finitely many ends, nevertheless the algebra of Example 4. in the Introduction, has quadratic growth and infinitely many ends.
Question 1 If $GKdim\left(\text{}\mathcal{A}\text{}\right)$  is finite and $\text{}\mathcal{A}\text{}$  is finitely presented, is it possible that $\text{}\mathcal{A}\text{}$  has infinitely many ends ?

5 The ends of group algebras

Let $\Gamma$  be a group and $K$  is a commutative field. If $V\subset \Gamma$  is an almost invariant subset then $KV\subset K\Gamma$  is an almost invariant subspace. Therefore if $\Gamma$  has at least $k$  ends, then $K\Gamma$  has at least $k$  ends as well. On the other hand let us suppose that $\Gamma$  is a one-ended group, then $\Gamma ×{C}_{2}$  is one-ended as well, where ${C}_{2}$  is the cyclic group of two elements. Then $\frac{1+a}{2}$  is an idempotent in $\mathbb{C}\left(\Gamma ×{C}_{2}\right)$  , where $a$  is the generator of ${C}_{2}$  . In general, complex group algebras of infinite groups with torsion always has more than one ends. Of course finite groups have zero ends and its group algebras are finite dimensional, so they have zero ends as well.
According to the Idempotent Conjecture, if $\Gamma$  is torsion-free, then $\mathbb{C}\Gamma$  has no non-trivial idempotent. The following question can be viewed as a version of the Idempotent Conjecture.
Question 2 If $\Gamma$  is torsion-free and has only one end, is it true that $\mathbb{C}\Gamma$  has one end as well ?
Note that the answer is yes if and only if $\text{Fred}\left(\mathbb{C}\Gamma \right)$  has no non-trivial idempotent.
In the Abelian case we can answer the Question.
Theorem 4 If $\Gamma \simeq {\mathbb{Z}}^{n}$  , $n\ge 2$  , then the group algebra $K\Gamma$  has only one end.
We introduce some notations. If $P\in K{\mathbb{Z}}^{n}$  and $P={\sum }_{\underline{v}\in {\mathbb{Z}}^{n}}{k}_{\underline{v}}\underline{v}$  , ${k}_{\underline{v}}\in K$  , where $\left[\underline{v}\right]$  is the representative of the element $\underline{v}\in {\mathbb{Z}}^{n}$  in $K{\mathbb{Z}}^{n}$  , thus $\left[\underline{v}\right]\left[\underline{w}\right]=\left[\underline{v}+\underline{w}\right]$  . Then let $T\left(P\right)=\left\{\underline{v}\in {\mathbb{Z}}^{n}|{k}_{\underline{v}}\ne 0\right\}.$  If $\underline{v}=\left({k}_{1},{k}_{2},...,{k}_{n}\right)\in {\mathbb{Z}}^{n}$  , then $s\left(\underline{v}\right)={sup}_{1\le j\le n}{k}_{j}$  , $i\left(\underline{v}\right)={inf}_{1\le j\le n}{k}_{j}$  . If $P\in K{\mathbb{Z}}^{n}$  , then $s\left(P\right)={sup}_{\underline{v}\in T\left(P\right)}s\left(\underline{v}\right),$  $i\left(P\right)={inf}_{\underline{v}\in T\left(P\right)}i\left(\underline{v}\right).$  If $m\le t$  are integers, then ${A}_{m,t}=\left\{\left({k}_{1},{k}_{2},...,{k}_{n}\right)\in {\mathbb{Z}}^{n}|m\le {k}_{j}\le t,\text{if}1\le j\le n\right\},{A}_{m}={\cup }_{t=m}^{\infty }{A}_{m,t}.$  The leading term of $P\in {\mathbb{Z}}^{n}$  , $l\left(P\right)$  is the greatest element of $T\left(P\right)$  in the lexicographic ordering. Furthermore, let ${x}_{1}=\left[\left(1,0,...,0\right)\right],{x}_{2}=\left[\left(0,1,0...,0\right)\right],...,{x}_{n}=\left[\left(0,0,...,0,1\right)\right]\in K{\mathbb{Z}}^{n}$  .
Lemma 5.1 Let $V\subset K{\mathbb{Z}}^{n}$  be an infinite dimensional almost invariant subspace.
Then if $d\in \mathbb{N}$  is large enough, there exists $Q\in V$  such that $l\left(Q\right)=\left(d,d,...,d\right)$  .
Proof. By definition, if $V$  is an infinite dimensional almost invariant subspace of $K{\mathbb{Z}}^{n}$  , then there exists a finite set $\left\{{\underline{b}}_{1},{\underline{b}}_{2},...,{\underline{b}}_{s}\right\}$  such that for any $R\in V,R{x}_{i}={R}^{\prime }+L$  , where ${R}^{\prime }\in V,L\in \text{span}K\left\{{\underline{b}}_{1},{\underline{b}}_{2},...,{\underline{b}}_{s}\right\}$  . Let $i\left(b\right)={inf}_{1\le j\le s}i\left({\underline{b}}_{j}\right),s\left(b\right)={sup}_{1\le j\le s}s\left({\underline{b}}_{j}\right).$  so the symmetric difference of $T\left({R}^{\prime }\right)$  and $T\left(R\right)+\left(0,0,...,1,...,0\right)$  is in the hypercube ${A}_{i\left(b\right),s\left(b\right)}.$  Naively speaking we have control on the terms only outside this hypercube. Start with an element ${P}_{1}\in V$  with a term ${\underline{v}}^{1}/\in {A}_{i\left(b\right),s\left(b\right)}.$  Then there exists ${P}_{2}\in V$  with a term ${\underline{v}}^{2}/\in {A}_{i\left(b\right),s\left(b\right)}$  and recursively we can obtain ${P}_{l}\in V$  with a term ${\underline{v}}^{l}$  , such that ${\underline{v}}^{l}>{\underline{b}}_{j}$  in the lexicographic ordering, for any $1\le j\le s$  . Then repeating the process, we can obtain $Q\in V$  , $l\left(Q\right)=\left(d,d,...,d\right)$  , $s\left(b\right)  .
Now fix an integer $d>0$  such that for some $Q\in V$  , $l\left(Q\right)=\left(d,d,...,d\right)$  and $s\left(b\right)  . We also fix an integer $m\le 0$  such that $m  and $m\le {inf}_{\underline{v}\in T\left(Q\right)}i\left(\underline{v}\right)$  , for the same $Q\in V$  .
Let $r=s\left(Q\right)-d$  . Then $Q{x}_{1}={Q}^{\prime }+{L}^{\prime }$  , where ${Q}^{\prime }\in V$  and ${L}^{\prime }\in \text{span}K\left\{{\underline{b}}_{1},{\underline{b}}_{2},...,{\underline{b}}_{s}\right\}.$  Obviously, $l\left({Q}^{\prime }\right)=\left(d+1,d,...,d\right)$  and $s\left({Q}^{\prime }\right)-\left(d+1\right)=r$  . By induction, we can obtain the following lemma.
Lemma 5.2 For any $\underline{v}\in {\mathbb{Z}}^{n}$  such that $i\left(\underline{v}\right)\ge d$  , there exists $Q\in V$  such that $l\left(Q\right)=\underline{v}$  , $T\left(Q\right)\subset {A}_{m}$  and $s\left(Q\right)-s\left(\underline{v}\right)=r$  .
Lemma 5.3 ${lim}_{t\to \infty }\frac{{dim}_{K}\left(V\cap \text{span}K\left({A}_{m,t}\right)\right)}{|{A}_{m,t}|}=1$
Proof. First note that if ${P}_{1},{P}_{2},...,{P}_{l}\subset K{\mathbb{Z}}^{n}$  have different leading terms then they form an independent system over $K$  . By the previous lemma, if $t$  is large enough then for any $\underline{v}$  , $d\le i\left(\underline{v}\right)\le t-r$  , there exists $Q\in V$  such that $l\left(Q\right)=\underline{v}$  and $T\left(Q\right)\subset {A}_{m,t}$  . Hence, $\frac{{dim}_{K}\left(V\cap \text{span}K\left({A}_{m,t}\right)\right)}{|{A}_{m,t}|}\ge \frac{|{A}_{d}\cap {A}_{m,t-r}|}{|{A}_{m,t}|}.$  However, ${lim}_{t\to \infty }\frac{|{A}_{d}\cap {A}_{m,t-r}|}{|{A}_{m,t}|}=1$  Now the Theorem follows. Indeed, by the previous lemma, any two infinite dimensional almost invariant subspaces have non-zero intersections.

6 Some questions on domains

Among the intensively studied “geometric” properties of groups, one has the growth, amenability and ends (see Gromov (1993)). As mentioned in the course of this paper, these concepts are closely related. The three concepts have been introduced in the context of algebras (See Elek (2003) and Samet (2000) for discussions on the amenability of algebras).
Groups of polynomial growth has integer degree of growth. Algebras of polynomial growth may have any real number greater or equal two as Gelfand-Kirillov dimension (Krause-Lenagan (2000)). It is an old problem, whether all domains of polynomial growth have integer Gelfand-Kirillov dimension.
By Hopf Theorem, a group may have zero, one, two or infinitely many ends. All the domains we can check has zero, one, two or infinitely many ends. So we can ask a similar question.
Question 3 Is it true, that a domain may have only zero, one, two or infinitely many ends ?
Finally, it is well known, that amenable groups have finitely many ends. The algebra in Example 3. has quadratic growth and so it is amenable. On the other hand, it has infinitely many ends. Again, we can ask about the similarity of domains to groups.
Question 4 Is it true, that any amenable domain has only finitely many ends ?
References Connes, A. (1994). Noncommutative Geometry. Academic Press Inc: San Diego. Elek, G. (2003). The amenability of affine algebras. J. Algebra 264:469-478.
Epstein, D. (1962). Ends. In: Topology of 3-manifolds and related results, Ford M. Ed., Prentice Hall, pp. 110-117.
Freudenthal, H. (1945). Über die Enden diskreter Räume und Gruppen Comment. Math. Helv. 17:1-38.
Gromov, M. (1993). Asymptotic invariants of infinite groups. in: Geometric Group Theory, Graham A. Niblo and Martin A. Roller, Eds., Vol. 2, London Mathematical Society Lecture Note Series 182, Cambridge Univ. Press, Cambridge, UK, pp. 1-295.
Hopf, H. (1944). Enden offener Räume und unendliche diskontinuierliche Gruppen. Comment. Math. Helv. 16:81-100.
Houghton, C. (1972). Ends of groups and the associated first cohomology group. J. London Math. Soc. 6:81-92.
Krause, G., Lenagan, T.(2000) Growth of algebras and the Gelfand-Kirillov dimension (Revised edition), Graduate Studies in Mathematics, 22, American Mathematical Society: Providence, RI. Samet-Vaillant, A. Y. (2000) ${C}^{*}$  -algebras, Gelfand-Kirillov dimension and Folner sets. J. Funct. Anal. 171:346-365.