---
title: Markov Traces in Iwahori–Hecke Algebras
url: https://www.emergentmind.com/topics/markov-traces-for-iwahori-hecke-algebras
type: topic
---

# Markov Traces in Iwahori–Hecke Algebras

Searching arXiv for recent and foundational papers on Markov traces and Iwahori–Hecke algebras, especially type \(B\), generalized constructions, and related categorical or representation-theoretic refinements.
Markov traces for Iwahori–Hecke algebras are trace functionals on towers of Hecke algebras that satisfy cyclicity together with stabilization rules adapted to braid generators, and they form the algebraic core of braid-closure constructions of link invariants. In the classical type \(A\) setting, the Markov trace recovers the HOMFLY–PT polynomial; in type \(B\), analogous traces yield invariants of links in the solid torus [2507.19896]. Beyond the classical theory, recent work has emphasized two complementary directions: generalized Markov traces encoded by central elements built from multiplicative Jucys–Murphy elements in ordinary Iwahori–Hecke algebras of types \(A\), \(B\), and \(D\) [2507.19896], and Markov traces on framized Hecke-type algebras, notably the type \(B\) framization \({\rm Y}_{d,n}^{\mathtt B}(u,v)\), where framing idempotents enter both the quadratic relations and the trace recursion [1603.08487]. These developments clarify that “Markov trace” in the Hecke context is not a single rigid object, but a family of related constructions whose precise form depends on the tower, the parameter regime, and the algebraic enlargement under consideration.

## 1. Definition and basic framework

A Markov trace on a tower of Iwahori–Hecke algebras is a family of linear maps
\[
\phi_n:H(X_n)\to \mathcal R
\]
such that there exist \(\mu,\rho\in\mathcal R\) with
\[
\phi_n(hh')=\phi_n(h'h),
\]
\[
\phi_n(\iota(h))=\rho\,\phi_{n-1}(h),
\]
and
\[
\phi_n(\iota(h)t_n)=\mu\,\phi_{n-1}(h),
\]
where \(\iota_n:H(X_n)\hookrightarrow H(X_{n+1})\) is the natural embedding in the tower [2507.19896]. In this formulation, the trace property is accompanied by an unlink or removal property and a Markov stabilization rule for the last generator.

This notion is the algebraic counterpart of the Markov-move formalism in braid-theoretic link invariants. In type \(A\), the Jones–Ocneanu trace on \(H(A_{n-1})\) recovers the HOMFLY–PT polynomial, while in type \(B\) analogous traces produce invariants of links in the solid torus [2507.19896]. The same source emphasizes that in types \(B\) and \(D\), classifications had previously been obtained by Geck and Lambropoulou.

A useful perspective is to represent a trace by a central element. If \(z\in H(X_n)\otimes_{\mathcal A}\mathcal R\) and
\[
\phi(h)=\langle z,h\rangle,
\]
then the trace property is equivalent to centrality of \(z\) [2507.19896]. This shifts the problem from functional equations on \(\phi_n\) to the construction of explicit central elements with the required inductive behavior.

Not every trace appearing in the Hecke literature is a Markov trace in this sense. Work on irreducible character values on Coxeter basis elements computes ordinary characters \(\phi(T_w)\) of finite Hecke algebras, but does not impose tower recursion or stabilization rules [2412.00540]. Likewise, the regular-representation trace
\[
\mathfrak T(w,q)=\operatorname{tr}(L_{T_w}:H\to H)
\]
encodes point-counting on varieties attached to reductive groups, but again is not a braid-theoretic Markov trace [2105.04061]. These trace theories are relevant as spectral or geometric input, but they are distinct from the Markov-trace framework proper.

## 2. Classical type \(A\) and its extensions

In type \(A\), the Markov trace is classically unique after normalization. This uniqueness is recalled explicitly in the study of Yokonuma–Hecke algebras: there exists a unique normalized Markov trace \(\{\tau_n\}_{n\ge1}\) on the tower of type \(A\) Iwahori–Hecke algebras \(\mathcal H_n\), where \(\tau_n\) satisfies
\[
\tau_n(xy)=\tau_n(yx)
\]
and
\[
\tau_{n+1}(xT_n)=\tau_{n+1}(xT_n^{-1})=\tau_n(x)
\qquad (x\in\mathcal H_n)
\]
[1501.06389]. The same source records the induced normalization formula
\[
\tau_n(1)=\bigl(v^{-1}(1-u^2)\bigr)^{n-1},
\]
as well as the factorization relation on tensor-product Hecke subalgebras
\[
\tau_n(x_1\otimes \cdots\otimes x_d) = \bigl(v^{-1}(1-u^2)\bigr)^{[\mu]-1} \tau_{\mu_1}(x_1)\cdots \tau_{\mu_d}(x_d)
\]
for \(\mathcal H^\mu\cong \mathcal H_{\mu_1}\otimes\cdots\otimes \mathcal H_{\mu_d}\) [1501.06389].

The rigidity of the ordinary type \(A\) theory has motivated the study of generalized settings where extra trace parameters appear only after enlarging the algebra. A notable example is the central extension of the Iwahori–Hecke algebra at \(q=-1\). For an arbitrary Coxeter system \((W,S)\), a non-split central extension of the Hecke algebra defined by
\[
(s-a)^2=0,\qquad a^2=1
\]
is constructed, and in type \(A\) this extension admits a unique exotic Markov trace
\[
t_n:\overline F_n\to A_0
\]
such that
\[
t_{n+1}(x s_n^{\pm1})=t_n(x)
\qquad\text{and}\qquad
t_2(C)=1
\]
[1403.4021]. This trace does not descend to the ordinary Hecke algebra; it detects the nilpotent central class \(C\). The construction shows that extra Markov traces may arise at singular parameter values only after passing to a larger algebraic object.

A different kind of extension appears in the Yokonuma–Hecke setting. The isomorphism
\[
Y_{d,n}\cong \bigoplus_{\mu\in \mathrm{Comp}_d(n)} \mathrm{Mat}_{m_\mu}(\mathcal H^\mu)
\]
reduces the classification of Markov traces on \(Y_{d,n}\) to the already understood type \(A\) theory [1501.06389]. Concretely, every Markov trace on the Yokonuma tower is blockwise a scalar multiple of tensor products of the unique normalized type \(A\) traces:
\[
\rho^\mu=x_{[\mu]}(\tau_{\mu_1}\otimes \cdots \otimes \tau_{\mu_d})
\]
[1501.06389]. This suggests a general structural principle: once a Hecke-type algebra decomposes into matrix blocks over ordinary Iwahori–Hecke factors, its Markov-trace theory is often inherited from the classical type \(A\) case.

## 3. Generalized Markov traces via Jucys–Murphy elements

A recent development is the construction of generalized Markov traces on ordinary Iwahori–Hecke towers of classical types directly from multiplicative Jucys–Murphy elements [2507.19896]. The setting includes the infinite crystallographic series:
- type \(A\): \(H(A_{n-1})\),
- type \(B\): \(H_{v,v_0}(B_n)\),
- type \(D\): \(H(D_n)\).

For \(X\in\{A_{-1},B,D\}\), the multiplicative Jucys–Murphy elements \(J_i^X\) are defined from the full twist \(S(X_n)=t_{w_0}^{-2}\) by
\[
J_n^X=S(X_n)^{-1}S(X_{n-1}),
\]
with initial values
\[
J_1^A=1,\qquad J_1^B=t_0^2,\qquad J_1^D=1
\]
[2507.19896]. These elements commute pairwise, and \(J_n^X\) centralizes \(H(X_{n-1})\). Symmetric polynomials in them are central, which makes them natural trace representatives.

The basic uniform construction is
\[
\zeta_n=\prod_{i=1}^n\bigl(1+a^{-1}J_i^X\bigr),\qquad \zeta_0=1.
\]
The associated family \((tr_{\zeta_n})_{n\ge0}\) is a Markov trace with constants
\[
\rho=1+a,\qquad \mu=v-v^{-1}
\]
in all three types \(A\), \(B\), and \(D\) [2507.19896]. The proof uses the decomposition
\[
\zeta_n=\zeta_{n-1}(1+a^{-1}J_n^X),
\]
the Serre property of the full twist, and orthogonality with respect to parabolic cosets.

In type \(B\), the construction goes further. The Geck–Lambropoulou classification yields a universal Markov trace with parameters \(y_1,y_2,\dots\), and the specialization \(y_n=y^n\) is realized by the central element
\[
\beta_n(y)=\prod_{i=1}^n\Bigl(1+(\overline y+\alpha_0)j_i^B+a^{-1}J_i^B\Bigr),
\]
where
\[
\alpha=v-v^{-1},\qquad \alpha_0=v_0-v_0^{-1},\qquad J_i^B=(j_i^B)^2
\]
[2507.19896]. The resulting trace satisfies
\[
tr_{n,y}^B=tr_{\beta_n(y)}.
\]
A key strengthening is the identity
\[
tr_{\beta_n(y)}(\iota(h)T_n)=y\,tr_{\beta_{n-1}(y)}(h),
\]
with
\[
T_n=t_{n-1}\cdots t_1t_0t_1^{-1}\cdots t_{n-1}^{-1},
\]
which is precisely how the parameter \(y\) is encoded [2507.19896].

Type \(D\) is obtained by restriction along \(H(D_n)\hookrightarrow H_{v,1}(B_n)\). If
\[
P_n(x_1,\dots,x_n;y)=\prod_{i=1}^n\bigl(1+\overline y x_i+a^{-1}x_i^2\bigr)
\]
and \(P_n^+\) denotes its even-degree part, then the type \(D\) trace is represented by
\[
\delta_n(y)=P_n^+(j_1^D,\dots,j_n^D;y),
\qquad
tr_{n,y}^D=tr_{\delta_n(y)}
\]
[2507.19896]. The even-part extraction reflects the parity obstruction specific to the \(D\)-embedding.

This Jucys–Murphy approach is significant because it represents generalized Markov traces by explicit central elements rather than only abstract classification parameters. A plausible implication is that it renders the inductive behavior of the traces more transparent, especially in types \(B\) and \(D\), where the classical theory is structurally richer than in type \(A\).

## 4. Framization and the type \(B\) algebra \({\rm Y}_{d,n}^{\mathtt B}(u,v)\)

A different enlargement of the Hecke framework is the framization of the type \(B\) Iwahori–Hecke algebra introduced in “A Framization of the Hecke algebra of Type B” [1603.08487]. The algebra
\[
{\rm Y}_{d,n}^{\mathtt B}(u,v)
\]
is defined over
\[
{\Bbb K}={\Bbb C}(u,v)
\]
from the \(d\)-modular framed braid group of type \(B\), \(\mathcal F_{d,n}^{\mathtt B}\), with generators \(\rho_1,\sigma_1,\dots,\sigma_{n-1},t_1,\dots,t_n\). Inside \({\Bbb K}[\mathcal F_{d,n}^{\mathtt B}]\), one defines the idempotents
\[
f_1:= \frac{1}{d}\sum_{m=0}^{d-1} t_1^{m}, \qquad e_i:= \frac{1}{d}\sum_{m=0}^{d-1} t_i^{m}t_{i+1}^{d-m}.
\]
The quotient is determined by the framized quadratic relations
\[
\rho_1^2 - 1 - (v - v^{-1})f_1\rho_1,
\qquad
\sigma_i^2 - 1 -(u - u^{-1})e_i\sigma_i.
\]

Writing \(g_i\) for the image of \(\sigma_i\) and \(b_1\) for the image of \(\rho_1\), the key quadratic relations become
\[
g_i^2 =  1+ (u-u^{-1})e_ig_i,
\qquad
b_1^2 =  1 + (v-v^{-1})f_1b_1
\]
[1603.08487]. This is the essential novelty: the scalar Hecke coefficients are replaced by coefficients involving framing idempotents, and the loop generator \(b_1\) is framized as well.

The relation with the ordinary type \(B\) Iwahori–Hecke algebra
\[
{\rm H}_n(u,v)
\]
is explicit. When \(d=1\), one has
\[
{\rm Y}_{1,n}^{\mathtt B}(u,v)\cong {\rm H}_n(u,v),
\]
and for general \(d\), the map
\[
g_i\mapsto h_i,\qquad b_1\mapsto h_0,\qquad t_j\mapsto 1
\]
induces an epimorphism
\[
{\rm Y}_{d,n}^{\mathtt B}(u,v)\twoheadrightarrow {\rm H}_n(u,v)
\]
[1603.08487]. There is also an epimorphism to \({\rm H}_n(u,1)\) obtained by sending all \(t_i\) to a fixed nontrivial \(d\)-th root of unity.

The algebra is accompanied by a faithful tensorial representation
\[
\Phi:{\rm Y}_{d,n}^{\mathtt B}(u,v)\to {\rm End}(V^{\otimes n}),
\]
extending Green’s tensor representation of the type \(B\) Hecke algebra [1603.08487]. This representation supports two basis theorems. The first basis,
\[
\mathsf{D}_n=\{g_w t_1^{m_1}\cdots t_n^{m_n}\, ;\, w\in W_n,\ (m_1,\dots, m_n)\in (\mathbb{Z}/d\mathbb{Z})^n\},
\]
implies
\[
\dim {\rm Y}_{d,n}^{\mathtt B}(u,v)=2^n d^n n!.
\]
The second basis, \(\mathsf C_n\), is tailored to the inductive trace construction [1603.08487].

This framized type \(B\) algebra is not the same as the cyclotomic Yokonuma–Hecke algebra \({\rm Y}(d,2,n)\) at \(m=2\). The difference lies precisely in the loop-generator quadratic relation:
\[
b_1^2 = 1 + (v-v^{-1})f_1 b_1
\]
in \({\rm Y}_{d,n}^{\mathtt B}(u,v)\), whereas in \({\rm Y}(d,2,n)\) the quadratic relation for the loop generator does not involve framing idempotents [1603.08487]. This difference propagates to trace values and link invariants.

## 5. Relative traces, \(E\)- and \(F\)-systems, and the framized Markov trace

The Markov trace on \({\rm Y}_{d,n}^{\mathtt B}(u,v)\) is constructed on the tower
\[
{\rm Y}_{d,0}^{\mathtt B}\subseteq {\rm Y}_{d,1}^{\mathtt B}\subseteq \cdots \subseteq {\rm Y}_{d,n}^{\mathtt B}\subseteq {\rm Y}_{d,n+1}^{\mathtt B}\subseteq \cdots,
\]
with \({\rm Y}_{d,0}^{\mathtt B}={\Bbb L}\) and
\[
{\Bbb L}={\Bbb C}(u,v,z)
\]
[1603.08487]. The trace depends on parameters
\[
x_1,\dots,x_{d-1},\ y_0,\dots,y_{d-1}\in {\Bbb L},
\qquad x_0:=1,
\]
together with the Markov parameter \(z\).

The construction proceeds by relative traces
\[
{\rm tr}_n:{\rm Y}_{d,n}^{\mathtt B}\to {\rm Y}_{d,n-1}^{\mathtt B}.
\]
For \(n=1\),
\[
{\rm tr}_1(t_1^{a_1}) = x_{a_1}, \qquad {\rm tr}_1 (b_1t_1^{a_1}) = y_{a_1}.
\]
For \(n\ge 2\), on the basis \(\mathsf C_n\),
\[
{\rm tr}_n(w \mathfrak{m}_n) =\left\{
\begin{array}{ll}
x_{m}w & \mbox{for}\quad \mathfrak{m}_n=t_{n}^{m} \\
y_{m}w & \mbox{for}\quad \mathfrak{m}_n=b_{n}t_{n}^{m} \\
z\, w \mathfrak{m}_{n-1,k,m}^{\pm} & \mbox{for} \quad \mathfrak{m}_n=\mathfrak{m}_{n,k, m}^{\pm}
\end{array}
\right.
\]
[1603.08487]. The absolute trace is then
\[
{\rm Tr}_1={\rm tr}_1,\qquad {\rm Tr}_n={\rm Tr}_{n-1}\circ {\rm tr}_n.
\]

The resulting family \({\rm Tr}=\{{\rm Tr}_n\}_{n\ge1}\) is a Markov trace in the Hecke sense:
\[
{\rm Tr}_n(1)=1,
\]
\[
{\rm Tr}_{n+1}(Xg_n)=z{\rm Tr}_n(X),
\]
\[
{\rm Tr}_{n+1}(Xb_{n+1}t_{n+1}^{m})=y_{m}{\rm Tr}_n(X),
\]
\[
{\rm Tr}_{n+1}(Xt_{n+1}^{m})=x_{m}{\rm Tr}_n(X),
\]
\[
{\rm Tr}_n(XY)={\rm Tr}_n(YX)
\]
for \(X,Y\in {\rm Y}_{d,n}^{\mathtt B}\) [1603.08487]. Compared with ordinary Hecke traces, the presence of framings and the loop generator forces a richer recursion with the extra parameters \(x_m\) and \(y_m\).

To obtain link invariants, the trace must factor through the relevant idempotents. The key requirement is
\[
{\rm Tr}_{n+1}(we_n)={\rm Tr}_n(w)\,{\rm Tr}_{n+1}(e_n)
\qquad \text{for all } w\in {\rm Y}_{d,n}^{\mathtt B}.
\]
This leads to the \(E\)- and \(F\)-systems. Define
\[
E^{(k)}:= \frac{1}{d}\sum_m x_{k+m}x_{d-m},
\qquad
F^{(k)}:=\frac{1}{d}\sum_m x_{d-m}y_{k+m},
\]
with indices taken modulo \(d\). The \(E\)-system is
\[
E^{(m)} = x_mE^{(0)} \qquad (1\le m\le d-1),
\]
and, assuming an \(E\)-solution, the \(F\)-system is
\[
F^{(m)} = y_mE^{(0)} \qquad (0\le m\le d-1)
\]
[1603.08487]. These are the compatibility constraints peculiar to the framized type \(B\) setting.

The \(E\)-solutions are parametrized by nonempty subsets \(S\subseteq \mathbb Z/d\mathbb Z\):
\[
x_S=\frac{1}{|S|}\sum_{s\in S}{\bf e}_s.
\]
For such a solution, the \(F\)-solutions are
\[
y_S= \sum_{s\in S}\alpha_s{\bf e}_{s},
\]
with arbitrary complex coefficients \(\alpha_s\) [1603.08487]. The paper interprets this by saying that \(\widehat y\) must be supported inside the same subset \(S\) as \(\widehat x\).

This machinery generalizes the classical Markov-trace picture in a precise sense. The trace is still a normalized cyclic functional on a braid-type tower with stabilization under the last braid generator, but it now carries framization-specific parameters and idempotent factorization constraints. This suggests that framization replaces the classical scalar stabilization data by a structured compatibility problem involving Fourier-analytic support conditions.

## 6. Topological realizations, refinements, and neighboring trace theories

The topological role of Markov traces remains central throughout the Hecke and Hecke-type literature. For the framized type \(B\) algebra, after specialization to \(E\)- and \(F\)-solutions, the trace yields invariants of framed links in the solid torus via the Jones recipe for braids of type \(B\). If
\[
\pi:\mathcal F_n^{\mathtt B}\to {\rm Y}_{d,n}^{\mathtt B}
\]
is the natural map and \(\alpha\) is a framed braid, then
\[
{\mathcal X}_S^{\mathtt B}(\widehat{\alpha}) := \Lambda_S^{\,n-1} (\sqrt{\lambda_S})^e\, {\rm Tr}(\pi(\alpha))
\]
depends only on the isotopy class of the framed link in the solid torus [1603.08487]. Restricting to zero framings gives invariants of classical links in \(ST\).

The type \(B\) invariants from \({\rm Y}_{d,n}^{\mathtt B}(u,v)\) differ from those coming from the cyclotomic Yokonuma–Hecke algebra \({\rm Y}(d,2,n)\). Already for \(b_1^2\),
\[
{\rm Tr}(\pi(b_1^2)) = 1+\frac{(v-v^{-1})}{d}\sum_s y_s,
\]
whereas in \({\rm Y}(d,2,n)\),
\[
{\mathtt Tr}(\overline{\pi}(b_1^2)) = 1+(v-v^{-1})y_0
\]
[1603.08487]. The invariants therefore differ on a basic example, showing that the loop-generator framization is not a superficial variation.

A more categorical refinement is developed in the Hecke-category setting. A monoidal trace
\[
\mathsf{AH}:\mathsf H_W^{\op}\to Mod_2^b(\mathbb A_W)
\]
decategorifies to a trace
\[
\langle -\rangle:H_W\to R(W)\otimes \mathbb Q_W(q^{\frac12})
\]
valued in graded virtual characters of \(W\) [2106.07444]. This is not a scalar Markov trace, but it recovers Gomi’s Markov trace through the formula
\[
\mathsf{tr}(\beta) = (-q^{-\frac12})^{|\beta|} \left(\frac{1-q}{1-a^2}\right)^r \sum_i (-a^2)^i\, (\Alt^i(\mathbb{V}),\langle \beta\rangle^0)_W
\]
[2106.07444]. In type \(A\), this yields the HOMFLY polynomial of the braid closure; in the categorical enhancement, Khovanov–Rozansky homology appears as a summand. This shows that the classical Markov trace can arise as a scalar shadow of a richer character-valued or categorical trace theory.

Other neighboring trace theories remain relevant but should not be conflated with Markov traces. The regular trace
\[
\mathfrak T(w,q)=\operatorname{tr}(L_{T_w})
\]
controls point-counting and irreducibility statements for varieties attached to reductive groups [2105.04061]. Irreducible character values on Coxeter basis elements satisfy
\[
\phi(T_w)=\varepsilon_\phi (u^{1/2})^{m_\phi},
\qquad \varepsilon_\phi\in\{0,\pm1\},
\]
providing precise spectral data for central trace decompositions [2412.00540]. Trace functionals on the infinite-dimensional Hecke algebra \(\mathcal H_\infty(q)\) are classified by Vershik–Kerov parameters \((\alpha,\beta,\gamma)\), but these are positive indecomposable traces on a direct limit algebra, not Markov traces in the braid-theoretic sense [2101.02133]. Such distinctions are important because the word “trace” spans several adjacent but non-equivalent theories in the Hecke context.

## 7. Structural themes and significance

Several broad structural themes emerge from these developments. First, centrality is the unifying algebraic mechanism. In the Jucys–Murphy approach, traces are represented by explicit central symmetric polynomials in commuting elements \(J_i^X\) or \(j_i^X\) [2507.19896]. In the framized type \(B\) approach, the recursive relative traces are built on a basis tailored to the tower and then constrained by factorization through idempotents [1603.08487]. In categorical refinements, monoidal trace properties replace scalar cyclicity while still decategorifying to central trace functionals [2106.07444].

Second, type \(B\) is genuinely richer than type \(A\). In ordinary Iwahori–Hecke theory, generalized type \(B\) traces involve the extra parameter \(y\) encoded by the loop-type elements \(T_n\) and by the central elements \(\beta_n(y)\) [2507.19896]. In framized type \(B\), the loop generator itself is framized, producing additional parameters \(y_m\) and the new \(F\)-system [1603.08487]. This is not merely a higher-parameter reformulation of type \(A\); it reflects the interaction between braid generators, loop generators, and framings.

Third, the scope of Markov-trace theory extends well beyond ordinary scalar traces on standard Hecke towers. Some extensions remain algebraic, as in the central extension at \(q=-1\) or the Yokonuma and framized algebras [1403.4021], [1501.06389], [1603.08487]. Others are geometric or categorical, as in character-valued traces from Hecke categories [2106.07444]. A plausible implication is that the modern theory of Markov traces is best viewed as a hierarchy: scalar traces on classical towers at the base, generalized central-element constructions above them, and categorical or geometric lifts above those.

In this sense, Markov traces for Iwahori–Hecke algebras occupy a central position between algebraic representation theory and low-dimensional topology. The classical trace on \(H(A_{n-1})\) remains foundational, but current work shows that the concept admits substantial generalization: by changing Coxeter type, by introducing loop or framing data, by passing to central extensions, or by lifting to category-valued constructions. The resulting theory retains the basic Markov-trace paradigm—cyclicity plus stabilization—while exhibiting a much wider range of algebraic realizations and topological outputs [2507.19896], [1603.08487].

Source: https://www.emergentmind.com/topics/markov-traces-for-iwahori-hecke-algebras