---
title: Macdonald Dimensions Overview
url: https://www.emergentmind.com/topics/macdonald-dimensions
type: topic
---

# Macdonald Dimensions Overview

Searching arXiv for recent and foundational papers on Macdonald dimensions across representation theory, symmetric functions, and 4d/2d correspondences.
arXiv search query: "Macdonald dimensions Vogel universality Macdonald index refined character chiral algebra".
“Macdonald dimensions” denotes several closely related but context-dependent refined quantities built from Macdonald theory. In representation-theoretic and symmetric-function settings, the term refers to special evaluations of Macdonald polynomials at refined Weyl or topological loci, often yielding $q,t$-deformations of quantum or classical dimensions. In four-dimensional supersymmetric quantum field theory, it refers instead to refined graded degeneracies extracted from the Macdonald limit of the superconformal index and, via the $4\mathrm d$–$2\mathrm d$ correspondence, from refined vacuum characters of vertex operator algebras (VOAs). In bisymmetric and multi-symmetric extensions, the same phrase is used for $q,t$-Kostka coefficients that specialize to dimensions of irreducible representations of hyperoctahedral or wreath-product groups. The literature therefore uses a common label for structurally analogous, but not identical, notions of refined dimension [2507.11414] [2103.00985] [1211.3186] [1909.09354].

## 1. Terminological scope and common structure

Across the cited literature, Macdonald dimensions are always tied to the passage from Schur-theoretic or ordinary character data to Macdonald-theoretic $(q,t)$-refinements. What changes from one context to another is the underlying object being counted or evaluated: highest-weight representations, symmetric polynomials, protected operator spaces in SCFTs, or Schur-expansion coefficients in bisymmetric and multi-symmetric theories.

| Context | Object | Meaning of “Macdonald dimensions” |
|---|---|---|
| Root systems and Lie theory | $P^{(R,S)}_\lambda$ | Evaluation at refined Weyl points |
| Topological-locus symmetric functions | $M_{q,t}(\lambda;A)$ | $(q,t)$-refined dimension-like specialization |
| 4d SCFT / VOA | Macdonald index or refined vacuum character | Graded degeneracies at fixed level and charge |
| Double and multi-Macdonald theory | $q,t$-Kostka coefficients | $q,t$-analogs of group-representation dimensions |

A recurring structural feature is factorization. In the simply laced root-system setting, ordinary and dual Macdonald dimensions coincide and factorize [2507.11414]. At topological loci for ordinary Macdonald polynomials, the specialization also factorizes over boxes of a Young diagram [1607.00615]. In generalized Macdonald-polynomial settings, full factorization fails generically but survives on codimension-one slices [1607.00615]. In 4d/2d applications, the relevant refined series are not dimension formulas in the Lie-theoretic sense, but graded traces whose coefficients function as refined dimensions of protected state spaces [2103.00985] [1909.04074].

This suggests that the term is best understood as a family resemblance: each usage encodes a refined, graded, or specialized notion of dimension, but the ambient category differs.

## 2. Root-system definition: Macdonald, dual Macdonald, and mixed Macdonald dimensions

For a crystallographic root system $R$ with Weyl group $W$ and an admissible pair $(R,S)$ of root systems sharing the same $W$, Macdonald–Cherednik theory provides $W$-invariant symmetric polynomials $P^{(R,S)}_\lambda(x|t_\alpha;q,t)$ indexed by dominant weights $\lambda$, characterized by triangularity and orthogonality with respect to the Macdonald scalar product determined by the Macdonald density [2507.11414]. With Weyl and refined Weyl vectors
\[
\rho=\frac12\sum_{\alpha>0}\alpha,\qquad \rho_k=\frac12\sum_{\alpha>0}k_\alpha\alpha,
\]
and the corresponding coroot data
\[
r=\frac12\sum_{\alpha>0}\alpha^\vee,\qquad r_k=\frac12\sum_{\alpha>0}k_\alpha\alpha^\vee,
\]
the Macdonald dimension of highest weight $\lambda$ is defined by evaluation at the refined Weyl point
\[
Md^{(R,S)}_\lambda:=P^{(R,S)}_\lambda\bigl(x=q^{2\rho_k}\,\big|\,t_\alpha^2\,\big|\,q^2,t^2\bigr).
\]
In general these do not factorize [2507.11414].

The dual Macdonald dimensions are defined by evaluation at the dual refined point
\[
r_k^*:=\frac12\sum_{\alpha>0}k_\alpha \alpha^*=\frac12\sum_{\alpha>0}k_\alpha u_\alpha \alpha^\vee,
\]
with $\alpha^*=u_\alpha\alpha^\vee$, and they do factorize:
\[
\widehat{Md}{}^{(R,S)}_\lambda
=
\prod_{\alpha\in R_+}\prod_{j=1}^{(\alpha^\vee,\lambda)}
\frac{\{t_{\alpha/2}t_\alpha q_\alpha^{(\rho_k,\alpha^\vee)+j-1}\}}
{\{t_{\alpha/2}q_\alpha^{(\rho_k,\alpha^\vee)+j-1}\}}.
\]
Here $q_\alpha=q^{u_\alpha}$ and $t_\alpha=q_\alpha^{k_\alpha}$, with $k_\alpha$ depending only on root length [2507.11414].

For simply laced types $A,D,E$, one has $\rho_k=r_k$ and $u_\alpha=1$, hence $Md=\widehat{Md}$; factorization then already holds at $x=q^{2\rho_k}$ [2507.11414]. This is the setting in which the 2025 literature formulates a Vogel-universal adjoint expression. With Vogel parameters $(\mathfrak a,\mathfrak b,\mathfrak c)$ and $\mathfrak t=\mathfrak a+\mathfrak b+\mathfrak c$, the adjoint Macdonald dimension is given universally for simply laced series by
\[
Md_{Adj}(\mathfrak a,\mathfrak b,\mathfrak c;q,t)
=
-
\frac{
\{t^{\,\mathfrak a+\frac{\mathfrak b}{2}+\mathfrak c}\}
\{t^{\,\mathfrak a+\mathfrak b+\frac{\mathfrak c}{2}}\}
\{t^{\,\mathfrak a+\mathfrak b+\mathfrak c}\}
\{q\,t^{\,\mathfrak a+\mathfrak b+\mathfrak c}\}
}{
\{t^{\,\frac{\mathfrak a}{2}}\}
\{t^{\,\frac{\mathfrak b}{2}}\}
\{t^{\,\frac{\mathfrak c}{2}}\}
\{q\,t^{\,\mathfrak a+\mathfrak b+\mathfrak c-1}\}
},
\]
equivalently in the longer five-factor form stated in the paper [2507.11414].

The same paper also defines mixed Macdonald dimensions for pairs $(R,S)$ with $R\neq S$, such as $(B_n,C_n)$ or $(C_n,B_n)$, where length-dependent parameters split and only the dual quantity enjoys a product formula [2507.11414]. A central limitation is explicit: no universal Vogel expression is presently known beyond simply laced types, and for non-simply-laced root systems only the dual quantity factorizes in general [2507.11414].

## 3. Topological loci, hook formulas, and generalized polynomial settings

In the symmetric-function literature, Macdonald dimensions are also realized as principal or character-like specializations of Macdonald polynomials. For ordinary Macdonald polynomials, the topological locus is
\[
p_k=\frac{1-A^k}{1-t^k},
\]
and the specialization factorizes as
\[
M_{q,t}\Bigl(\lambda;A\Bigr)
=
M_{q,t}\Bigl\{p_k=\frac{1-A^k}{1-t^k}\Bigr\}
=
\prod_{\Box\in\lambda}
\frac{1-A\,q^{a'(\Box)}t^{l'(\Box)}}{1-q^{a(\Box)}t^{l(\Box)+1}}.
\]
When $A=t^N$, this specialization is referred to as the Macdonald dimension of $\lambda$ [1607.00615].

This is the direct $(q,t)$-refinement of the Schur-level hook formula for quantum dimensions of $U_q(SL_N)$ representations. For example,
\[
M_{q,t}((2);A)=\frac{(1-A)(1-Aq)}{(1-t)(1-qt)},\qquad
M_{q,t}((1,1);A)=\frac{(1-A)(1-At)}{(1-t)(1-t^2)},
\]
and setting $A=t^N$ produces the corresponding Macdonald dimensions [1607.00615].

The same source distinguishes this ordinary factorization from the more subtle behavior of generalized Macdonald polynomials associated with the toroidal Ding–Iohara–Miki algebra. For first-coproduct generalized Macdonald polynomials $M_{Y_1,Y_2}\{p,\bar p\}$ depending on two sets of times and an additional deformation parameter $Q$, full topological-locus factorization does not persist generically. Instead, the paper identifies a weak factorization on codimension-one slices, notably
\[
P_k^{**}=t^{-k},\qquad \bar p_k=0,
\]
where the plethystic logarithm becomes linear in $Q$ and factorized in a nontrivial sense; the conjecture is checked up to $|Y_1|+|Y_2|\le 5$ [1607.00615]. The paper is explicit that this weak factorization is conjectural and that a full two-parameter topological-locus factorization fails away from these slices [1607.00615].

A related but distinct framework appears in generalized Macdonald polynomials for 5d AGT. There, “dimension-like” quantities arise from character-like specializations or Selberg averages that reduce to product-over-box expressions built from the Nekrasov factors
\[
f_A^\pm(x)=\prod_{(i,j)\in A}\bigl[1-q^{\pm x}t^{\pm(i-1)}q^{\mp(j-1)}\bigr]
\]
and
\[
G_{AB}(x)=\prod_{s\in A}(1-Qq^{a_A(s)}t^{l_B(s)+1})\prod_{t\in B}(1-Qq^{-a_B(t)-1}t^{-l_A(t)}),
\]
with $Q=q^x$ [1412.8592]. In that literature, Macdonald dimensions are the specialization values that function as refined character or instanton weights rather than root-theoretic dimensions proper [1412.8592].

## 4. Double and multi-Macdonald theories: dimensions of $B_n$ and $C_n\sim S_d$

In the stable limit of Macdonald superpolynomials, double Macdonald polynomials indexed by bipartitions $(\lambda,\mu)$ arise as bisymmetric polynomials and obey a factorization theorem:
\[
P_{\lambda,\mu}(x,y;q,t)
=
P_\lambda(q,qt)\Bigl[X+\frac{q(1-t)}{1-qt}Y\Bigr]\cdot P_\mu(qt,t)[Y].
\]
This factorization immediately yields norms, kernels, duality, and evaluation formulas [1211.3186].

The term “Macdonald dimensions” enters here through the modified double Macdonald polynomials and their double $q,t$-Kostka coefficients. Writing
\[
H_{\lambda,\mu}(x,y;q,t)=\sum_{\kappa,\gamma}K_{\kappa,\gamma;\lambda,\mu}(q,t)\,s_{\kappa,\gamma}(x,y),
\]
the coefficients $K_{\kappa,\gamma;\lambda,\mu}(q,t)$ are positive and specialize at $(q,t)=(1,1)$ to the dimensions of irreducible representations of the hyperoctahedral group $B_n$:
\[
K_{\kappa,\gamma;\lambda,\mu}(1,1)=\dim \chi_{\kappa,\gamma}.
\]
Equivalently,
\[
\deg_{B_n}(\kappa,\gamma)=\binom{n}{|\gamma|}f^\kappa f^\gamma,
\]
where $f^\nu$ is the hook-length dimension of the irreducible $S_{|\nu|}$-module indexed by $\nu$ [1211.3186].

The paper also defines a type-$B$ Nabla operator whose action on a certain bisymmetric Schur function yields a Frobenius series with total dimension $(2n+1)^n$ [1211.3186]. This does not redefine Macdonald dimensions themselves, but it locates them inside a broader type-$B$ combinatorial and representation-theoretic apparatus.

An $n$-fold generalization is developed in the theory of multi-Macdonald polynomials indexed by multipartitions
\[
\pmb\lambda=(\lambda^{(1)},\dots,\lambda^{(n)}).
\]
These polynomials factor as products of ordinary Macdonald polynomials evaluated at recursively defined alphabets:
\[
P_{\pmb\lambda}^{(q,t)}[X^{(1)},\dots,X^{(n)}]
=
P_{\lambda^{(1)}}^{(q,q^{n-1}t)}[A_{n-1}]
\cdots
P_{\lambda^{(n)}}^{(qt,t)}[A_0],
\]
with
\[
A_0=X^{(n)},\qquad
A_i=X^{(n-i)}+\frac{q(1-q^{i-1}t)}{1-q^it}\,A_{i-1}.
\]
Their modified forms define multi $q,t$-Kostka coefficients
\[
H_{\pmb\lambda}^{(q,t)}=\sum_{\pmb\mu}K_{\pmb\mu,\pmb\lambda}(q,t)\,s_{\pmb\mu},
\]
and these satisfy
\[
K_{\pmb\mu,\pmb\lambda}(1,1)=\chi^{\pmb\mu}_{\mathrm{Id}}=
\dim\bigl(\mathrm{Irr}_{\pmb\mu}(C_n\sim S_d)\bigr)
=
\frac{d!}{\prod_i d_i!}\prod_i f^{\mu^{(i)}}.
\]
In this context, the paper explicitly states that “Macdonald Dimensions” refers to the coefficients $K_{\pmb\mu,\pmb\lambda}(q,t)$ as $q,t$-analogs of the dimensions of irreducible representations of the wreath product $C_n\sim S_d$ [1909.09354].

A common misconception is that these coefficients are merely evaluation formulas. The paper distinguishes them from principal specializations: in the multi-Macdonald setting, “Macdonald Dimensions” specifically refers to the $q,t$-Kostka coefficients, not to the evaluation map itself [1909.09354].

## 5. Four-dimensional SCFTs and VOAs: Macdonald dimensions as refined graded operator counts

In the 4d $\mathcal N=2/\mathcal N=3$ SCFT literature, Macdonald dimensions are not evaluations of symmetric functions but refined graded degeneracies of protected operator spaces. For a 4d $\mathcal N=2$ SCFT, the superconformal index is
\[
\mathcal I(p,q,t;z)=
\mathrm{Tr}\,(-1)^F\,
p^{\frac{E-2j_2-2R-r}{2}}
q^{\frac{E+2j_2-2R-r}{2}}
t^{R+r}z^m,
\]
and the Macdonald limit is obtained by $p\to 0$, yielding
\[
\mathcal I^{\mathcal T}_{\mathrm{Macdonald}}(q,t;z)
=
\mathrm{Tr}\,(-1)^F\,q^{E-2R-r}t^{R+r}z^m
\]
on the subsector obeying $E-2j_2-2R-r=0$ [2103.00985].

Via the 4d–2d correspondence, the Macdonald-limit index is identified with a refined vacuum character of the associated VOA. In the conventions used for $\mathcal N=3$ S-fold theories, one sets $t\to \xi q$, so that
\[
\mathcal I^{\mathcal T}_{\mathrm{Macdonald}}(q,\xi;z)
=
\mathrm{Tr}\,(-1)^F\,q^{E-R}\xi^{R+r}z^m,
\]
and
\[
\mathcal I^{\mathcal T}_{\mathrm{Macdonald}}(q,\xi;z)
=
\chi_{\mathcal W_{\mathsf G}}(q,\xi,z),
\qquad
\chi_{\mathcal W_{\mathsf G}}(q,\xi,z)
=
\mathrm{Tr}\,(-1)^F q^h \xi^{R+r} z^m,
\]
with $h=E-R$ [2103.00985].

In this setting, the coefficients in the expansion
\[
\mathcal I^{\mathcal T}_{\mathrm{Macdonald}}(q,\xi;z)
=
\sum_{h\in \frac12\mathbb Z_{\ge 0}}q^h\Bigl(\sum_Q d_h(Q)\,\xi^Q\Bigr)
\]
are the Macdonald dimensions: graded degeneracies of Schur operators at fixed level $h=E-R$ and charge $Q=R+r$ (and flavor $m$ if $z$ is retained) [2103.00985]. In the Schur limit, $\xi\to 1$ and $z\to 1$, they reduce to
\[
\mathcal I_{\mathrm{Schur}}(q)=
\sum_h q^h\bigl(\#\text{ bosons at }h-\#\text{ fermions at }h\bigr).
\]

The paper computes these quantities by brute force for $\mathcal N=2$ VOAs labeled by crystallographic complex reflection groups. For $\mathsf G=\mathbb Z_3,\mathbb Z_4,\mathbb Z_6$ and $G(3,1,2)$, the VOAs are realized as subalgebras of free $\beta\gamma bc$ ghost systems and identified as kernels of screening operators [2103.00985]. The resulting refined vacuum characters encode the Macdonald-limit data. Representative results include:

- For $\mathbb Z_3$, with central charge $c=-15$, the Schur expansion begins
  \[
  1+q+q^2+2q^3-2q^{7/2}+3q^4-2q^{9/2}+4q^5+\cdots,
  \]
  so low-order Macdonald dimensions include $d_0=1$, $d_1=1$, $d_2=1$, $d_3=2$, $d_{7/2}=-2$ [2103.00985].

- For $\mathbb Z_4$, with central charge $c=-21$, bosonic states appear at integer $h$ and fermionic states at half-integer $h$, implying no boson–fermion cancellation within a fixed level; the Schur series begins
  \[
  1+q-2q^{3/2}+5q^2-6q^{5/2}+10q^3+\cdots
  \]
  [2103.00985].

- For $G(3,1,2)$, with central charge $c=-48$, the refined Macdonald character at $z\to 1$ begins
  \[
  1+q\,\xi
  +q^{3/2}(-\sqrt{\xi}+\xi^{3/2})
  +q^2(\xi+\xi^2)
  +q^{5/2}(-\sqrt{\xi}-\xi^{3/2}+2\xi^{5/2})
  +\cdots,
  \]
  so, for example, at $h=\frac32$ one has $d_{3/2}(\tfrac12)=-1$ and $d_{3/2}(\tfrac32)=+1$ [2103.00985].

The significance of these quantities is operational: null-state relations subtract from naive free-field degeneracies, producing the nontrivial signs and cancellations visible in the refined series [2103.00985]. In this usage, “Macdonald dimensions” is therefore a protected-sector counting notion rather than an evaluation formula.

## 6. Refined characters, Argyres–Douglas theories, universality, and open issues

A closely related SCFT usage appears in the study of $(A_{n-1},A_{m-1})$ Argyres–Douglas theories with $\gcd(n,m)=1$. There, Song’s proposal identifies the 4d Macdonald index with a refined character of the dual chiral algebra,
\[
\mathcal I^{\mathrm{4d}}_{\mathrm{Mac}}(q,t;\text{defects})
\stackrel{?}{=}
\chi^{\mathrm{VOA}}_{\mathrm{ref}}(q,T;\text{module}),
\qquad T=\frac{t}{q},
\]
where
\[
\chi_{\mathrm{ref}}(q,T)=\sum_{\text{states}}T^\# q^h
\]
counts the number of basic chiral-algebra generators in a PBW-like basis after removing nulls [1909.04074]. In this framework, “Macdonald dimensions” are the coefficients in the $q,T$ expansion of the refined character, equivalently the $q,t$-refined multiplicities of 2d states per level [1909.04074].

A key result is the large-$m$ product formula
\[
\lim_{m\to\infty}\mathcal I^{(A_{n-1},A_{m-1})}(q,t)
=
\prod_{s=2}^n \frac{1}{(T^{s-1}q^s;q)},
\qquad T=\frac{t}{q},
\]
which directly encodes refined generator multiplicities in the $W_n$ VOA [1909.04074]. Explicit low-rank examples show agreement between Macdonald indices and refined characters for vacuum modules and some simple defect modules, but the paper also reports mismatches for larger defect labels, especially for rank-1 defects with $s\ge 2$, where naive Higgsing strip-off factors lead to indices that cannot be interpreted cleanly as refined characters [1909.04074]. This is one of the clearest controversies in the SCFT usage: the proposal works well in many simple and large-$m$ cases, but not uniformly.

A different kind of limitation appears in the Vogel-universality literature. For simply laced ADE root systems, the adjoint Macdonald dimension admits a single universal Vogel expression [2507.11414]. However, the 2025 extension to link hyperpolynomials argues that Macdonald dimensions themselves are not universal on Vogel’s plane; rather, universality emerges for products of Macdonald dimensions with $(q,t)$-deformed Littlewood–Richardson coefficients in the adjoint sector [2505.16569]. In that setting, the universal objects are the six “uirreps” in
\[
Adj\otimes Adj=X_2\oplus Y_2(\mathfrak a)\oplus Y_2(\mathfrak b)\oplus Y_2(\mathfrak c)\oplus Adj\oplus \varnothing,
\]
and the relevant universal combinations are
\[
\mathcal U_\lambda(\mathfrak a,\mathfrak b,\mathfrak c;q,t)
=
\mathcal B^\lambda_{Adj,Adj}(q,t)\cdot Md^R(\lambda)
\]
rather than $Md^R(\lambda)$ alone [2505.16569]. The paper further uses these combinations to write universal refined Hopf and $T[2,2n]$ torus-link hyperpolynomials in the adjoint sector for ADE [2505.16569].

Taken together, these developments show that the phrase “Macdonald dimensions” has stabilized around a common theme—$(q,t)$-refined dimension data—but not around a single formal definition. In Lie theory it is an evaluation of Macdonald polynomials; in topological-locus and AGT settings it is a factorized specialization; in double and multi-Macdonald theory it is a $q,t$-Kostka refinement of representation-theoretic dimensions; and in SCFT/VOA applications it is a refined graded count of protected states [2507.11414] [1607.00615] [1211.3186] [1909.09354] [2103.00985] [1909.04074]. A plausible implication is that the durability of the term comes less from a single definition than from a shared algebraic pattern: Macdonald theory repeatedly turns ordinary dimensions, characters, or graded multiplicities into two-parameter refined invariants that preserve substantial factorization and representation-theoretic structure.

Source: https://www.emergentmind.com/topics/macdonald-dimensions