Proposition 3.1.
We have the formulae
$$\mu (A\otimes B,u\oplus v)=\mu (A,u)*\mu (B,v)$$
$$\mu (A\otimes B,u\otimes v)=\mu (A,u)\times \mu (B,v)$$
$$\mu (A*B,u\oplus v)=\mu (A,u)\u229e\mu (B,v)$$
where
$*,\times ,\u229e$
are the additive, multiplicative, and free additive convolutions.

Proof.
The corresponding characters are given by the following formulae.
$$\chi (u\oplus v)=\chi \left(u\right)+\chi \left(v\right)$$
$$\chi (u\otimes v)=\chi \left(u\right)\chi \left(v\right)$$
The Haar functional of
$A\times B$
being the
$\times $
product of those of
$A$
and
$B$
, the variables
$\chi \left(u\right)\in A\times B$
and
$\chi \left(v\right)\in A\times B$
are independent or free, and have same spectral measures as
$\chi \left(u\right)\in A$
and
$\chi \left(v\right)\in B$
. The result follows from the definition of convolutions. □
The free wreath product
$A{*}_{w}B$
is constructed in [
4]
. This is the algebra generated by
$n$
copies of
$A$
and a copy of
$B$
, with the
$a$
th copy of
$A$
commuting with the
$a$
th row of
$v$
, for any
$a$
.
The maps
$\Delta $
and
$\varepsilon $
are constructed by using the universal property of
$A{*}_{w}B$
.
Definition 3.2.
The free wreath product of
$(A,u)$
and
$(B,v)$
is given by
$$A{*}_{w}B=({A}^{*n}*B)/<[{u}_{ij}^{\left(a\right)},{v}_{ab}]=0>$$
where
$n\times n$
is the size of
$v$
. This has fundamental magic biunitary matrix
$${w}_{ia,jb}={u}_{ij}^{\left(a\right)}{v}_{ab}$$
and Hopf algebra structure making
$w$
a corepresentation.
There are several approaches to free wreath products, to be discussed in what follows. A very first thing to be noticed is the following diagram.
$$\begin{array}{ccccccccccccccccc}{A}^{*n}*B& & \to & & A{*}_{w}B& & \downarrow & & & & \downarrow & & A*B& & \to & & A\otimes B\end{array}$$  
Here all maps are surjective morphisms of
${\mathbb{C}}^{*}$
algebras, coming from definitions. Vertical maps are obtained by collapsing the
$n$
copies of
$A$
to a single one, and horizontal maps correspond to certain commutation relations with rows of
$v$
.
This diagram gives different ways of mixing diagonal elements
${u}_{ii}$
and
${v}_{aa}$
.
$$\begin{array}{ccccccccccccccccc}\sum {u}_{ii}^{\left(a\right)}{v}_{aa}& & \to & & \sum {u}_{ii}^{\left(a\right)}{v}_{aa}& & \downarrow & & & & \downarrow & & \sum {u}_{ii}{v}_{aa}& & \to & & \sum {u}_{ii}\otimes {v}_{aa}\end{array}$$  
Here in the upper right corner we have the unknown, namely the character of
$w$
.
In the lower left corner we have a familiar object, namely the character of
$u\otimes v$
. This corepresentation
$u\otimes v$
is not a magic biunitary, but the corresponding spectral measure can be constructed as in Definition 1.4, and is given by the missing formula in Proposition 3.1.
The spectral measures of these two elements can be computed in several cases of interest, and turn to be equal. However, equality doesn't hold in general, and a quite natural assumption here is that all diagonal elements
${u}_{ii}$
, as well as all diagonal elements
${v}_{aa}$
, should have same spectral measure. In terms of Hopf algebras, this leads to the following statement.
Conjecture 3.1.
If
$A$
and
$B$
have homogeneous skeletons we have the formula
$$\mu (A{*}_{w}B,w)=\mu (A,u)\u22a0\mu (B,v)$$
where
$\u22a0$
is the free multiplicative convolution.
The notion of skeleton is introduced as follows. The idea is to regard each pair
$(B,v)$
as corresponding to the quantum symmetry group of a graphlike object
$X=({X}_{n},D)$
. This is definitely possible for all known examples of
$(B,v)$
, because they appear naturally in this way.
In the general case one can take for instance the combinatorial data
$D$
to be the planar algebra associated to
$(B,v)$
, by using the TannakaGalois type correspondence in [
2]
.
A skeleton
$X=({X}_{n},D)$
is called homogeneous if for any
$i,j\in {X}_{n}$
there is a permutation
$g\in {S}_{n}$
preserving the combinatorial data
$D$
, such that
$g\left(i\right)=j$
.
All this is probably a bit too heuristic, but finding the precise assumptions in Conjecture 3.1 is rather a matter of proving the conjecture, and this is not among purposes of this paper.
We should mention here that finding a truly delinearised definition for skeletons is quite a subtle problem, beyond the stateofart of the subject. At level of general planar algebras this is known to be possible, as shown by Kodiyalam and Sunder in [
14]
.
In the rest of this section we will just explain why some assumptions are needed, and why these should correspond to homogeneity of skeletons of
$A$
and
$B$
.
Our first task is to construct a counterexample. We use the following simple fact.
Proposition 3.2.
We have a canonical isomorphism
$$(A,u){*}_{w}({B}_{1}*{B}_{2},{v}_{1}\oplus {v}_{2})=\left(A{*}_{w}{B}_{1}\right)*\left(A{*}_{w}{B}_{2}\right)$$
for any three pairs
$(A,u)$
,
$({B}_{1},{v}_{1})$
and
$({B}_{2},{v}_{2})$
.

Proof.
The algebra on the left is generated by
${n}_{1}+{n}_{2}$
copies of
$A$
and a copy of
${B}_{1}*{B}_{2}$
, subject to certain commutation relations. These relations say that the first
${n}_{1}$
copies of
$A$
commute with corresponding rows of the matrix
${v}_{1}$
, and that the remaining
${n}_{2}$
copies of
$A$
commute with corresponding rows of the matrix
${v}_{2}$
. In other words, when moving the copy of
${B}_{2}$
from the
${n}_{1}+{n}_{2}+2$
th position to the
${n}_{1}+1$
th position, we get the algebra on the right. □
Now if we assume that Conjecture 3.1 holds without assumptions on
$(B,v)$
, we would get from Propositions 3.1 and 3.2 a kind of distributivity formula for
$\u22a0$
with respect to
$\u229e$
.
$$\mu \u22a0({\mu}_{1}\u229e{\mu}_{2})=(\mu \u22a0{\mu}_{1})\u229e(\mu \u22a0{\mu}_{2})$$
This formula is known to be wrong, so some assumptions on
$(B,v)$
are needed in Conjecture 3.1. We believe that homogeneity of the skeleton, not satisfied by
$({B}_{1}*{B}_{2},{v}_{1}\oplus {v}_{2})$
, but satisfied in the two cases where we know how to prove the conjecture, is the good one.
As for the assumption on
$(A,u)$
, we are less confident about it. Let us mention here that Conjecture 3.1 should be regarded as an analogue of the planar algebra formula
$\mu (P*Q)=\mu \left(P\right)\u22a0\mu \left(Q\right)$
of Bisch and Jones ([
10]
). In fact, we have the following computation.
$$\begin{array}{ccc}\mu \left(A{*}_{w}B\right)& =& \mu \left(P\right(A{*}_{w}B\left)\right)\end{array}$$  
$$\begin{array}{ccc}& =& \mu \left(P\right(A)*P(B\left)\right)\end{array}$$  
$$\begin{array}{ccc}& =& \mu \left(P\right(A\left)\right)\u22a0\mu \left(P\right(B\left)\right)\end{array}$$  
$$\begin{array}{ccc}& =& \mu \left(A\right)\u22a0\mu \left(B\right)\end{array}$$  
Here the first and fourth equality come from the TannakaGalois type correspondence
$A\to P\left(A\right)$
from [
2]
. The third equality comes from the formula of Bisch and Jones. The second equality comes from the following conjectural formula.
$$P\left(A{*}_{w}B\right)=P\left(A\right)*P\left(B\right)$$
This formula should be regarded as a version of Conjecture 3.1. The above computation shows that the formula is stronger than the conjecture, but one can probably go in the other sense as well, by using the following basic lemma.
Lemma 3.1.
Assume that
$(A,u)$
and
$(B,v)$
correspond to quantum permutation groups of
${X}_{n}$
, and let
$f:A\to B$
be a morphism of
${\mathbb{C}}^{*}$
algebras such that
$f\left({u}_{ij}\right)={v}_{ij}$
.
(i)
$f$
is a surjective morphism of Hopf
${\mathbb{C}}^{*}$
algebras.
(ii) The
$k$
th moment of
$\mu \left(A\right)$
is smaller than the
$k$
th moment of
$\mu \left(B\right)$
, for any
$k$
.
(iii)
$f$
is an isomorphism if and only if
$\mu \left(A\right)=\mu \left(B\right)$
.

Proof.
The first two assertions follows from definitions. The third one is wellknown, and follows by taking a basis of coefficients of irreducible corepresentations, as in [26] . □
As a conclusion, what is behind the conjecture is that the free wreath product operation
${*}_{w}$
should correspond to a kind of free product operation
$*$
at level of skeletons. This approach suggests that
$(A,u)$
and
$(B,v)$
will have to play at some point symmetric roles, and this is why homogeneity of the skeleton of
$(A,u)$
is included in Conjecture 3.1.
4 Finite groups
The simplest example of a free wreath product is with
$B=\mathbb{C}\left(G\right)$
, where
$G$
is a finite group acting on itself. That is, we consider
$G$
as being a subgroup of
${S}_{n}$
, where
$n$
is the order of
$G$
.
The matrix
$v$
has a particularly simple form.
$$v=\left(\begin{array}{cccccccccccccccc}{\delta}_{1}& {\delta}_{12}& ...& {\delta}_{1n}& {\delta}_{21}& {\delta}_{1}& ...& {\delta}_{2n}& ...& ...& ...& ...& {\delta}_{n1}& {\delta}_{n2}& ...& {\delta}_{1}\end{array}\right)$$
Here
${\delta}_{1}$
is the Dirac mass at the unit element
$1\in G$
, and for
$i,j\in {X}_{n}$
we denote by
${\delta}_{ij}$
the Dirac mass at the unique element
$g\in G$
satisfying
$g\left(j\right)=i$
. We see that each row of
$v$
consists of Dirac masses at elements of
$G$
, so in particular its entries generate
$\mathbb{C}\left(G\right)$
. This shows that a free wreath product by
$\mathbb{C}\left(G\right)$
has the following decomposition.
$$A{*}_{w}\mathbb{C}\left(G\right)={A}^{*n}\otimes \mathbb{C}\left(G\right)$$
This is pointed out in [
5]
, where the case of
$G={\mathbb{Z}}_{2}$
is worked out explicitely, with a complete discussion of corepresentations of the free wreath product. The fusion rules found there allow one to compute the spectral measure of the free wreath product. In what follows we present this computation of spectral measure, along with a generalisation to arbitrary groups
$G$
.
We use Voiculescu's
$S$
transform, whose log linearises
$\u22a0$
.
$$S(\mu \u22a0\nu )=S\left(\mu \right)S\left(\nu \right)$$
The
$S$
transform of a measure
$\mu $
is constructed by introducing series
$\psi ,\chi $
as follows ([
21]
).
$$\psi \left(z\right)=f\left(z\right)1\chi \circ \psi =\psi \circ \chi =idS\left(z\right)=\frac{1+z}{z}\chi \left(z\right)$$
Here
$f$
is the series having as coefficients the moments of
$\mu $
. In case
$\mu $
is the spectral measure associated to a Hopf
${\mathbb{C}}^{*}$
algebra
$A$
, this series
$f$
is the Poincaré series of
$A$
.
It is useful to compute both the spectral measure of
$\mathbb{C}\left(G\right)$
, and its
$S$
transform.
Proposition 4.1.
For a group
$G$
acting on itself the corresponding spectral measure of
$\mathbb{C}\left(G\right)$
and its
$S$
transform are given by
$$\mu =\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\delta}_{n}$$
$$S\left(z\right)=\frac{1+z}{1+nz}$$
where
$n$
is the number of elements of
$G$
.

Proof.
The identity of
$G$
has
$n$
fixed points, and the other
$n1$
elements, none. Together with Proposition 2.1 this gives the formula of
$\mu $
, then we get
$f,\psi ,\chi $
.
$$\begin{array}{ccc}\mu =\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\delta}_{n}& \Rightarrow & f\left(z\right)=1+\frac{z}{1nz}\end{array}$$  
$$\begin{array}{ccc}& \Rightarrow & \psi \left(z\right)=\frac{z}{1nz}\end{array}$$  
$$\begin{array}{ccc}& \Rightarrow & z=\frac{\chi \left(z\right)}{1n\chi \left(z\right)}\end{array}$$  
$$\begin{array}{ccc}& \Rightarrow & \chi \left(z\right)=\frac{z}{1+nz}\end{array}$$  
Multiplying by
$1+{z}^{1}$
gives the formula in the statement. □
Theorem 4.1.
If
$G$
is a finite group acting on itself then
$$\mu (A{*}_{w}\mathbb{C}(G\left)\right)=\mu \left(A\right)\u22a0\mu (\mathbb{C}(G\left)\right)$$
where
$\u22a0$
is the free multiplicative convolution.

Proof.
We denote by
$\chi ,{\chi}_{l}$
the characters of
$u,w$
, and by
$\mu ,{\mu}_{l}$
the associated spectral measures.
The tensor product decomposition of the free wreath product gives the following formula.
$${\chi}_{l}=\left({\chi}^{\left(1\right)}+\dots +{\chi}^{\left(n\right)}\right)\otimes {\delta}_{1}$$
We raise both terms to the power
$k\ge 1$
.
$${\chi}_{l}^{k}={\left({\chi}^{\left(1\right)}+\dots +{\chi}^{\left(n\right)}\right)}^{k}\otimes {\delta}_{1}$$
The Haar functional of the tensor product being the tensor product of Haar functionals, we get the following formula.
$$\int {\chi}_{l}^{k}=\frac{1}{n}\int {\left({\chi}^{\left(1\right)}+\dots +{\chi}^{\left(n\right)}\right)}^{k}$$
We get a formula which is valid for any polynomial
$\phi $
.
$$\int \phi \left({\chi}_{l}\right)=\frac{n1}{n}\phi \left(0\right)+\frac{1}{n}\int \phi \left({\chi}^{\left(1\right)}+\dots +{\chi}^{\left(n\right)}\right)$$
Now remember that elements
${\chi}^{\left(a\right)}$
are free, and each of them has spectral measure
$\mu $
. This gives the following identity.
$$\int \phi \left(x\right)d{\mu}_{l}\left(x\right)=\frac{n1}{n}\phi \left(0\right)+\frac{1}{n}\int \phi \left(x\right)d{\mu}^{\u229en}\left(x\right)$$
This finishes computation of the spectral measure on the left.
$${\mu}_{l}=\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\mu}^{\u229en}$$
The spectral measure on the right is given by the following formula.
$${\mu}_{r}=\mu \u22a0\left(\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\delta}_{n}\right)$$
We can use here the following general formula, discussed below.
$$\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\mu}^{\u229en}=\mu \u22a0\left(\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\delta}_{n}\right)$$
Together with the above descriptions of
${\mu}_{l}$
and
${\mu}_{r}$
, this finishes the proof. □
The free probability formula used in the end of proof of Theorem 4.1 is proved by Nica and Speicher in [
17]
by using noncrossing partitions. A second proof is given by Voiculescu in the appendix of [
17]
, by using random matrices.
The proof below was already written when we learned about [
17]
, and this is the main reason why we include it. The other reason is that we hope that a suitable analytic function procedure can be used in order to simplify the operation
$A\to \mu \left(A\right)$
, so a complete proof of Theorem 4.1 using analytic functions is probably useful.
We use Voiculescu's
$R$
transform, which linearises
$\u229e$
.
$$R(\mu \u229e\nu )=R\left(\mu \right)+R\left(\nu \right)$$
The
$R$
transform of a measure
$\mu $
is defined by introducing series
$G,K$
as follows ([
20]
).
$$G\left(\xi \right)={\xi}^{1}f\left({\xi}^{1}\right)K\circ G=G\circ K=idR\left(z\right)=K\left(z\right)\frac{1}{z}$$
Here, as usual,
$f$
denotes the series having as coefficients the moments of
$\mu $
.
A first remark is that both
$R$
and
$S$
are given, up to some normalisations, by an inversion of series. We have the following formulae connecting
$R$
and
$S$
.
Lemma 4.1.
We have the formulae
$$R\left(zS\right(z\left)\right)=\frac{1}{S\left(z\right)}S\left(zR\right(z\left)\right)=\frac{1}{R\left(z\right)}$$
where
$R$
and
$S$
are the
$R$
and
$S$
transforms of the same measure
$\mu $
.

Proof.
The starting point is the following formula, relating
$G$
to
$\psi $
.
$$G\left(\xi \right)=\frac{1}{\xi}f\left(\frac{1}{\xi}\right)=\frac{1}{\xi}+\frac{1}{\xi}\psi \left(\frac{1}{\xi}\right)$$
We make the replacement
$\xi \to 1/\chi \left(z\right)$
.
$$G\left(\frac{1}{\chi \left(z\right)}\right)=\chi \left(z\right)+\chi \left(z\right)z$$
By applying
$K$
we get a formula relating
$K$
and
$\chi $
.
$$\frac{1}{\chi \left(z\right)}=K\left(\chi \right(z\left)\right(1+z\left)\right)$$
In terms of
$R$
and
$S$
, we get the following equality.
$$\frac{1+z}{z}\cdot \frac{1}{S\left(z\right)}=K\left(zS\right(z\left)\right)=R\left(zS\right(z\left)\right)+\frac{1}{zS\left(z\right)}$$
This gives the first formula. For the second one, we start again with the equality relating
$G$
to
$\psi $
, and we make the replacement
$\xi \to K\left(z\right)$
.
$$z=\frac{1}{K\left(z\right)}+\frac{1}{K\left(z\right)}\psi \left(\frac{1}{K\left(z\right)}\right)$$
This can be written in the following way.
$$\psi \left(\frac{1}{K\left(z\right)}\right)=zK\left(z\right)1$$
By applying
$\chi $
we get a second formula relating
$K$
and
$\chi $
.
$$\frac{1}{K\left(z\right)}=\chi \left(zK\right(z)1)$$
On the other hand, we have the following equality.
$$S\left(zR\right(z\left)\right)=\frac{1+zR\left(z\right)}{zR\left(z\right)}\cdot \chi \left(zR\right(z\left)\right)=\frac{K\left(z\right)}{R\left(z\right)}\cdot \chi \left(zR\right(z\left)\right)$$
Together with
$zR\left(z\right)=zK\left(z\right)1$
, this gives the second formula. □
Theorem 4.2.
We have the NicaSpeicherVoiculescu formula
$$\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\mu}^{\u229en}=\mu \u22a0\left(\frac{n1}{n}{\delta}_{0}+\frac{1}{n}{\delta}_{n}\right)$$
for any probability measure
$\mu $
on the real line.

Proof.
We denote by
${\mu}_{l}$
and
${\mu}_{r}$
the measures on the left and on the right. Let also
${\mu}_{+}$
be the measure
${\mu}^{\u229en}$
, and
${\mu}_{\delta}$
be the measure in the bracket.
For any measure of type
${\mu}_{x}$
we denote by
${f}_{x},{G}_{x},{K}_{x},{R}_{x},{\psi}_{x},{\chi}_{x},{S}_{x}$
the series involved in the construction of
$R$
and
$S$
transforms.
The starting point is the following equality, coming from the fact that
$R$
linearises
$\u229e$
.
$${R}_{+}\left(z\right)=nR\left(z\right)$$
Together with the above lemma, this gives the following formula.
$$\begin{array}{ccc}{S}_{+}\left(nz\right)& =& {S}_{+}\left(nzS\right(z\left)R\right(zS\left(z\right)\left)\right)\end{array}$$  
$$\begin{array}{ccc}& =& {S}_{+}\left(zS\right(z\left){R}_{+}\right(zS\left(z\right)\left)\right)\end{array}$$  
$$\begin{array}{ccc}& =& \frac{1}{{R}_{+}\left(zS\right(z\left)\right)}\end{array}$$  
$$\begin{array}{ccc}& =& \frac{1}{nR\left(zS\right(z\left)\right)}\end{array}$$  
$$\begin{array}{ccc}& =& \frac{S\left(z\right)}{n}\end{array}$$  
We use now the following identity.
$${\psi}_{+}\left(z\right)={f}_{+}\left(z\right)1=n\left({f}_{l}\right(z)1)=n{\psi}_{l}\left(z\right)$$
At level of
$\chi $
functions, this translates as follows.
$${\chi}_{l}\left(z\right)={\chi}_{+}{\psi}_{+}{\chi}_{l}\left(z\right)={\chi}_{+}\left(n{\psi}_{l}{\chi}_{l}\right(z\left)\right)={\chi}_{+}\left(nz\right)$$
We have now all ingredients for computing
${S}_{l}$
.
$$\begin{array}{ccc}{S}_{l}\left(z\right)& =& \frac{1+z}{z}{\chi}_{l}\left(z\right)\end{array}$$  
$$\begin{array}{ccc}& =& \frac{1+z}{z}{\chi}_{+}\left(nz\right)\end{array}$$  
$$\begin{array}{ccc}& =& \frac{1+z}{z}\cdot \frac{nz}{1+nz}{S}_{+}\left(nz\right)\end{array}$$  
$$\begin{array}{ccc}& =& \frac{(1+z)n}{1+nz}\cdot \frac{S\left(z\right)}{n}\end{array}$$  
$$\begin{array}{ccc}& =& \frac{1+z}{1+nz}S\left(z\right)\end{array}$$  
Proposition 4.1 shows that the right term is
${S}_{\delta}\left(z\right)S\left(z\right)={S}_{r}\left(z\right)$
, and we are done. □
5 Colored graphs
Just as in the classical case, a natural way to get quantum permutation groups is to consider simple algebraic structures like graphs and construct their quantum automorphism groups.
Given a finite graph
$X$
, the idea is to label its vertices
$1,\dots ,n$
, then to consider an algebra of form
$A\left(X\right)=A\left({X}_{n}\right)/J$
, where
$J$
is an ideal expressing preservation of edges.
This will be explained later on. For the moment we have to fix some notations regarding
$X$
.
In previous papers [
1]
, [
2]
, [
4]
, [
5]
we use several kinds of finite graphs with edges oriented or not, colored or not plus finite metric spaces.
For the purposes of this paper, most convenient is to start with the following definition.
Definition 5.1.
In this paper a colored graph
$X=(V,E)$
is a finite set
$V={X}_{n}$
together with a partition
$E$
of the form
$$(V\times V){\Delta}_{V}={E}_{1}\bigsqcup \dots \bigsqcup {E}_{p}$$
where
${\Delta}_{V}$
is the diagonal of
$V\times V$
.
The terminology is of course not standard, we just use it in order to simplify presentation.
In fact colors are obviously missing, so
$X$
should be rather called “colorable” graph. We use “colored” because most interesting examples come along with a natural coloring.
In general, we can use a color set
$C$
, and an injective map
$c$
assigning colors to edges.
$$c:\{1,\dots ,p\}\to C$$
We get a picture of
$X$
in the following way. We call points elements of
$V$
, we draw them in the plane, then we draw an oriented edge
$\stackrel{\u20d7}{i}j$
between any two points
$i\ne j$
and we color it
$c\left(k\right)$
, with
$k$
given by
$(i,j)\in {E}_{k}$
. This kind of picture is to be called colored graph as well.
As a first example, consider a finite metric space
$X$
. This can be regarded as a colored graph, with edges colored by their lenghts. For instance when all distances are equal, say to
$a\in (0,\infty )$
, we get a monocolored graph, called simplex and denoted
${X}_{n}^{a}$
.
It is convenient to call simplex any monocolored graph. In case the unique color is already specified, say it is the color
$a$
, the simplex is denoted
${X}_{n}^{a}$
. In case the unique color is not specified, our favorite choice will be the real number
$a=0$
.
The second example is with an oriented graph
$X=(V,E)$
. Here
$V={X}_{n}$
is a finite set, called set of vertices, and
$E$
is a set of oriented edges between pairs of distinct vertices.
$$E\subset V\times V{\Delta}_{V}$$
We get a partition in the following way, where
$\overline{E}$
is the set of missing edges.
$$V\times V{\Delta}_{V}=E\bigsqcup \overline{E}$$
We can regard
$X$
as a colored graph, with color
$1$
assigned to edges, and color
$0$
assigned to missing edges. For instance
$E=\varnothing $
gives the simplex
${X}_{n}^{0}$
and
$\overline{E}=\varnothing $
gives the simplex
${X}_{n}^{1}$
.
In both the metric space and the graph example colors are complex numbers, and the matrix
${d}_{ij}=c\left(k\right)$
is the Laplacian of
$X$
. This suggests the following definition.
Definition 5.2.
The Laplacian of
$X$
is the matrix given by
${d}_{ii}=0$
and
$${d}_{ij}=c\left(k\right)$$
with
$k$
given by
$(i,j)\in {E}_{k}$
, where
$c:\{1,\dots ,p\}\to \mathbb{C}$
is a fixed injective map.
If
$X$
comes from a metric space or from an oriented graph, or more generally if
$X$
comes along with a natural complex coloring map
$c$
, this map will be the one used for constructing the Laplacian, unless otherwise specified.
The fact that the Laplacian depends on the choice of the complex coloring map
$c$
is not a problem, because all constructions in this paper do not depend on this choice. Actually, one of the reasons of presenting things this way is that we want to use independence from the choice of colors, for instance by using our favorite color
$a=0$
, whenever needed.
We will be interested in commutation relations of type
$vd=dv$
, with
$v$
magic biunitary matrix. The StoneWeierstrass theorem shows that validity of such a commutation relation doesn't depend on the choice of
$c$
. See Theorem 2.2 in [
2]
.
In other words, we can consider the ideal
${J}_{min}\subset A\left({X}_{n}\right)$
generated by the relations
$vd=dv$
.
Consider also the ideal
${J}_{com}$
generated by all commutation relations between
${v}_{ij}$
's.
Proposition 5.1.
We have a canonical isomorphism
$$A\left({X}_{n}\right)/{J}_{max}\simeq \mathbb{C}\left(G\right)$$
where
${J}_{max}=<{J}_{min},{J}_{com}>$
, and where
$G$
is the symmetry group of
$X$
.

Proof.
The quotient
$A\left({X}_{n}\right)/{J}_{com}$
is canonically isomorphic to
$\mathbb{C}\left({S}_{n}\right)$
, then further quotienting by relations
$[v,d]=0$
gives
$\mathbb{C}\left(G\right)$
. See [2] for details. □
This result suggests to define
$A\left(X\right)$
to be the quotient of
$A\left({X}_{n}\right)$
by an ideal
$J$
satisfying
${J}_{min}\subset J\subset {J}_{max}$
. The choice
$J={J}_{max}$
is of course not very interesting. Another natural choice is
$J={J}_{min}$
, used in [
1]
, [
2]
. An intermediate choice is made in [
4]
, [
5]
.
In this paper we extend and generalise results in [
5]
to algebras defined using
${J}_{min}$
, in order to get a general formula of type
$A(X*Y)=A\left(X\right){*}_{w}A\left(Y\right)$
. Together with results in [
2]
, this will lead to a second verification of Conjecture 3.1.
Definition 5.3.
The algebra
$A\left(X\right)$
is the quotient of
$A\left({X}_{n}\right)$
by the relations
$vd=dv$
.
The pair
$\left(A\right(X),v)$
satisfies conditions in Definition 1.3. In view of Proposition 5.1, the underlying quantum permutation group is an analogue of the automorphism group of
$X$
.
As an example, for a simplex we can take the Laplacian to be
$d=0$
, so we get
$A\left({X}_{n}\right)$
itself.
We can see here the difference with the formalism in [
4]
, [
5]
, where the graph having
$n$
vertices and no edges gives a different
$A$
algebra from that of the graph having
$n$
vertices and all possible edges. In fact, all definitions in this section are motivated by the present choice of
$A\left(X\right)$
.
An interesting situation is when
$X$
is the
$n$
gon. For any
$n\ne 4$
we have
$A\left(X\right)=\mathbb{C}\left({D}_{n}\right)$
, but for
$n=4$
the algebra
$A\left(X\right)$
is not commutative, and infinite dimensional. See [
2]
.
The relations defining
$A\left(X\right)$
might be rewritten in several convenient manners.
Proposition 5.2.
The algebra
$A\left(X\right)$
is the quotient of
$A\left({X}_{n}\right)$
by the relations
$${\sum}_{k,(k,j)\in {E}_{r}}{v}_{ik}={\sum}_{k,(i,k)\in {E}_{r}}{v}_{kj}$$
with the number
$r$
ranging over the set
$\{1,\dots ,p\}$
.

Proof.
We choose a Laplacian
$d$
, corresponding to a complex coloring map
$c$
. If
${d}_{r}$
denotes the Laplacian of the oriented graph
${X}_{r}=(V,{E}_{r})$
, we have the following formula.
$$d={\sum}_{r=1}^{p}c\left(r\right){d}_{r}$$
The matrix
${d}_{r}$
multiplies by
$v$
in the following way.
$$(v{d}_{r}{)}_{ij}={\sum}_{k}{v}_{ik}({d}_{r}{)}_{kj}={\sum}_{k,(k,j)\in {E}_{r}}{v}_{ik}$$
$$({d}_{r}v{)}_{ij}={\sum}_{k}({d}_{r}{)}_{ik}{v}_{kj}={\sum}_{k,(i,k)\in {E}_{r}}{v}_{kj}$$
This shows that the
$r$
th relation in the statement is equivalent to the commutation relation
$v{d}_{r}={d}_{r}v$
. On the other hand independence of
$A\left(X\right)$
from the choice of the Laplacian shows that
$vd=dv$
is equivalent to
$v{d}_{r}={d}_{r}v$
for any
$r$
, and this gives the result. □
Given two colored graphs
$X,Y$
, we can consider the colored graph
$X*Y$
obtained by putting a copy of
$X$
at each vertex of
$Y$
. In other words, we have the following definition.
Definition 5.4.
Let
$X=(T,E)$
and
$Y=(Z,F)$
be two colored graphs, with
$E=\{{E}_{1},\dots ,{E}_{p}\}$
and
$F=\{{F}_{1}\dots ,{F}_{q}\}$
. We define subsets of
$T\times Z$
times itself in the following way.
$${E}_{r}^{\circ}=\left\{\right(i\alpha ,j\alpha \left)\right(i,j)\in {E}_{r},\alpha \in Z\}$$
$${F}_{s}^{\circ}=\left\{\right(i\alpha ,j\beta \left)\righti,j\in T,(\alpha ,\beta )\in {F}_{s}\}$$
These determine a colored graph
$X*Y$
, called free product of
$X$
and
$Y$
.
The terminology is of course not standard, the free product of colored graphs not being a coproduct in a categorical sense. We call it like this because it is expected to be compatible with the planar algebra free product
$*$
of Bisch and Jones [10] .
As a first example, consider a free product of
$s$
simplices.
$$Z={Y}_{1}^{{\varepsilon}_{1}}*\dots *{Y}_{s}^{{\varepsilon}_{s}}$$
As pointed out in [
2]
, we have a nice interpretation of
$Z$
if we choose the corresponding
$s$
colors to be an increasing sequence of infinitesimals.
$${\varepsilon}_{1}<<{\varepsilon}_{2}<<\dots <<{\varepsilon}_{s}$$
The free product is obtained by starting with
${Y}_{s}$
, then putting a copy of
${Y}_{s1}$
at each vertex of
${Y}_{s}$
, a copy of
${Y}_{s2}$
at each vertex of
${Y}_{s1}$
, and so on until
${Y}_{1}$
. What we get is a kind of metric space, by using the infinitesimal summing conventions
${\varepsilon}_{i}+{\varepsilon}_{j}={\varepsilon}_{i}$
for
$i>j$
.
A simplex is called generic if it has
$n\ge 4$
vertices. The terminology comes from the proof below, where numbers
$n$
become Jones indices, called generic when bigger than
$4$
.
Theorem 5.1.
If
$Z$
is a free product of
$s$
generic simplices then
$$\mu \left(A\right(Z\left)\right)={\eta}^{\u22a0s}$$
where
$\eta $
is the free Poisson law of parameter
$1$
.

Proof.
It is shown in [2] that the corresponding
$s$
Laplacians satisfy Landau's exchange relations in [15] . This shows that the planar algebra associated to
$A\left(Z\right)$
is a FussCatalan algebra on
$s$
colors, whose Poincaré series is computed in the generic case by Bisch and Jones in [6] .
$$f\left(z\right)={\sum}_{k=0}^{\infty}\frac{1}{sk+1}\left(\begin{array}{cc}(s+1)k& k\end{array}\right){z}^{k}$$
As pointed out in [
7]
, this series is a solution of the following equation.
$$f\left(z\right)=1+zf(z{)}^{s+1}$$
This can be used for computing the
$S$
transform of the corresponding spectral measure.
$$\begin{array}{ccc}f\left(z\right)=1+zf(z{)}^{s+1}& \Rightarrow & \psi \left(z\right)=z(1+\psi (z){)}^{s+1}\end{array}$$  
$$\begin{array}{ccc}& \Rightarrow & z=\chi \left(z\right)(1+z{)}^{s+1}\end{array}$$  
$$\begin{array}{ccc}& \Rightarrow & \chi \left(z\right)=\frac{z}{(1+z{)}^{s+1}}\end{array}$$  
$$\begin{array}{ccc}& \Rightarrow & S\left(z\right)=\frac{1}{(1+z{)}^{s}}\end{array}$$  
With
$s=1$
this gives the
$S$
transform of
${\eta}_{4+}=\eta $
. Now back to arbitrary
$s$
, we get the following equality.
$$S\left(z\right)={\left(\frac{1}{1+z}\right)}^{s}=\left(S\eta \right(z){)}^{s}$$
The result follows from the fact that
$logS$
linearises
$\u22a0$
. □
6 Free products
The purpose of this section is to establish the formula
$A(X*Y)=A\left(X\right){*}_{w}A\left(Y\right)$
. This will be done via a modification plus generalisation of a proof in [
5]
, where free wreath products are constructed, and where a first such decomposition result is found.
The idea is to construct a pair of inverse morphisms, by using universal properties of various algebras involved. The morphism from left to right is easy to construct, and this is done in proof of Theorem 6.1 below. In the other sense, we have first to construct inside
$A(X*Y)$
analogues
$U,V$
of matrices
$u,v$
, and this is the purpose of next two lemmas.
The magic biunitary matrix associated to
$A(X*Y)$
is denoted
${W}_{i\alpha ,j\beta}$
. Here double indices
$i\alpha ,j\beta \in T\times Z$
are produced by indices
$i,j\in T$
and
$\alpha ,\beta \in Z$
, coming from Definition 5.4.
Lemma 6.1.
We can define an element
${V}_{\alpha \beta}$
by the formula
$${V}_{\alpha \beta}={\sum}_{k}{W}_{i\alpha ,k\beta}={\sum}_{k}{W}_{k\alpha ,j\beta}$$
for any
$i,j$
, and the resulting matrix
$V$
is a magic biunitary.

Proof.
The relations in Proposition 5.2 are written in the following way.
$${\sum}_{k,(k,j)\in {E}_{r}}{W}_{i\alpha ,k\beta}={\sum}_{k,(i,k)\in {E}_{r}}{W}_{k\alpha ,j\beta}$$
$${\sum}_{k,\gamma ,(\gamma ,\beta )\in {F}_{s}}{W}_{i\alpha ,k\gamma}={\sum}_{k,\gamma ,(\alpha ,\gamma )\in {F}_{s}}{W}_{k\gamma ,j\beta}$$
The equality in the statement is obtained from these relations.
$${\sum}_{k}{W}_{i\alpha ,k\beta}={\sum}_{r}{\sum}_{k,(k,j)\in {E}_{r}}{W}_{i\alpha ,k\beta}={\sum}_{r}{\sum}_{k,(i,k)\in {E}_{r}}{W}_{k\alpha ,j\beta}={\sum}_{k}{W}_{k\alpha ,j\beta}$$
The entries of
$V$
are easily seen to be selfadjoint.
$${V}_{\alpha \beta}^{*}={\sum}_{i}{\left({W}_{i\alpha ,j\beta}\right)}^{*}={\sum}_{i}{W}_{i\alpha ,j\beta}={V}_{\alpha \beta}$$
We check now the conditions regarding sums on rows and columns.
$${\sum}_{\alpha}{V}_{\alpha \beta}={\sum}_{i\alpha}{W}_{i\alpha ,j\beta}=1$$
$${\sum}_{\beta}{V}_{\alpha \beta}={\sum}_{j\beta}{W}_{i\alpha ,j\beta}=1$$
The fact that rows of
$V$
are partitions of unity is proved as follows.