---
title: Coset Free-Field Resolution in CFT
url: https://www.emergentmind.com/topics/coset-free-field-resolution
type: topic
---

# Coset Free-Field Resolution in CFT

Coset free-field resolution denotes a class of constructions in which a coset conformal field theory is realized through free fields, screening operators, and BRST projection. In the most explicit formulation available here, the starting point is a general coset model $\widehat g(k)/\widehat h(jk)$ with stress tensor $T(z)=T^{\widehat g(k)}(z)-T^{\widehat h(jk)}(z)$, where the coset fields are characterized as $\widehat h(jk)$-singular operators; if $L_{\widehat\Lambda}^{\widehat g}$ is a $\widehat g(k)$-module, then it decomposes as
\[
L_{\widehat{\Lambda}}^{\widehat g}
=
\bigoplus_{\widehat\lambda}
\Big(
L_{(\widehat\Lambda,\widehat\lambda)}^{\widehat g/\widehat h}
\otimes
L_{\widehat\lambda}^{\widehat h}
\Big),
\]
and the coset characters are the branching functions $b_{(\widehat\Lambda,\widehat\lambda)}$ [2509.01949]. In this sense, a free-field resolution isolates the branching space $L_{(\widehat\Lambda,\widehat\lambda)}^{\widehat g/\widehat h}$ inside a larger affine free-field complex. Closely related literature also uses boson–parafermion realizations, perturbative free-field expansions, and distinguished free-field-like bases, but these are not always full resolutions in the Felder, Wakimoto, or BRST sense [1411.3313], [2004.10216], [1111.2803].

## 1. Algebraic definition and branching-space viewpoint

The defining algebraic datum is the decomposition of an affine $\widehat g(k)$-module into denominator and coset sectors. In the formulation used for diagonal ADE cosets, the branching rule identifies the physical coset module with the multiplicity space in the decomposition under the denominator algebra [2509.01949]. This makes the branching function the natural character of the coset sector and fixes the target of any free-field construction.

In this framework, a free-field resolution is not merely a bosonization of currents. It is a construction of a complex whose zeroth cohomology reproduces the irreducible coset module. The relevant subcomplex is obtained by selecting $\widehat h$-singular vectors of fixed $\widehat h$-weight $\widehat\lambda$ inside a free-field resolution of $L_{\widehat\Lambda}^{\widehat g}$; the result is the free-field resolution of $L_{(\widehat\Lambda,\widehat\lambda)}^{\widehat g/\widehat h}$ [2509.01949]. The paper further states that the universal method for implementing this projection is a BRST procedure using a $\mathrm{rank}(h)$-dimensional constraint and $\mathrm{rank}(h)$ $bc$-ghost systems [2509.01949].

This branching-space viewpoint also clarifies why coset free-field resolution is more rigid than a generic free-field reformulation. The objective is not only to rewrite the theory in simpler variables, but to enforce the denominator-current constraint cohomologically and thereby extract the physical coset state space.

## 2. Free bosons, screenings, and affine-to-coset complexes

The explicit free-field machinery described for principal QDS $\mathcal W[\widehat g]$ models uses $r=\mathrm{rank}(g)$ background-charged bosons $\phi^i$ with stress tensor
\[
T(z)
=
-\frac12 :\partial\phi\cdot\partial\phi:(z)
-
(\alpha_+\rho+\alpha_-\rho^\vee)\cdot i\partial^2\phi(z),
\]
and screening operators
\[
s_i^\pm(z)=:e^{-i\alpha_\pm \alpha_i\cdot\phi}:(z)
\]
[2509.01949]. On the affine side, the realization also includes $\beta^\alpha,\gamma^\alpha$ systems indexed by positive roots, with affine screenings
\[
\widetilde s_i^+(z)=:e^{-i\alpha_+\alpha_i\cdot\phi}:(z)\,\widetilde P,
\]
where $\widetilde P$ is a polynomial in $\beta,\gamma$ [2509.01949].

The associated free-field complexes are organized by affine Weyl-group degree. For dominant affine modules,
\[
\widetilde F_{\widehat\Lambda}^{(n)}
=
\bigoplus_{\widehat w:\,l(\widehat w)=n}
\Big[
F_{\widehat w*\widehat\Lambda}^{\phi^i}
\otimes
F^{\beta^\alpha\gamma^\alpha}
\Big],
\]
and for admissible modules,
\[
\widetilde F_{(\widehat\Lambda^{(+)},\widehat\Lambda^{(-)})}^{(n)}
=
\bigoplus_{\widehat w:\,l(\widehat w)=n}
\Big[
F_{(\widehat w*\widehat\Lambda^{(+)},\widehat\Lambda^{(-)})}^{\phi^i}
\otimes
F^{\beta^\alpha\gamma^\alpha}
\Big]
\]
[2509.01949]. Under QDS reduction, these descend to a bosonic complex
\[
C_{(\Lambda^{(+)},\Lambda^{(-)})}
=
\big(F_{(\Lambda^{(+)},\Lambda^{(-)})}^{\phi^i},d\big),
\]
whose zeroth cohomology is either $0$ or the irreducible $\mathcal W$-module $L_{(\Lambda^{(+)},\Lambda^{(-)})}$ [2509.01949].

A central simplifying statement is the lemma
\[
H_Q\big(F_{\beta\gamma}\otimes F_{bc}\big)\cong \mathbb C,
\]
which explains why only the bosonic $\phi^i$ remain nontrivial after QDS reduction [2509.01949]. In the coset setting, this same mechanism underlies the projection from an affine free-field complex to the physical branching space.

## 3. Diagonal ADE cosets, principal $\mathcal W$ models, and parafermionic backgrounds

The most explicit coset application concerns diagonal ADE cosets
\[
\widehat g^{(1)}(k)\oplus \widehat g^{(2)}(1)\big/\widehat g^{(3)}(k+1),
\qquad
J^{a(3)}(z)=J^{a(1)}(z)+J^{a(2)}(z),
\]
denoted in the source as $\big[\widehat g\oplus \widehat g/\widehat g,(k,1)\big]$ [2509.01949]. Their chiral characters are stated to be identical to those of the principal QDS minimal models $\mathcal W[\widehat g](p,p')$ at
\[
k+h^\vee=\frac{p}{p'-p}:=\frac{p}{u}.
\]
For simply-laced $g$, the paper cites a rigorous proof that the principal QDS $\mathcal W$-algebra is equivalent to the diagonal coset $\mathcal W$-algebra [2509.01949]. This identification is fundamental: it allows the same irreducible $\mathcal W$-minimal module to be reached either from QDS free bosons or from a coset branching-space resolution.

The branching rule is written as
\[
L_{\widehat\Lambda^{(1)}}^{(1)}\otimes L_{\widehat\Lambda^{(2)}}^{(2)}
=
\bigoplus
\Big(
L^{\mathcal W}_{(\Lambda^{(+)},\Lambda^{(-)})}
\otimes
L_{\widehat\Lambda^{(3)}}^{(3)}
\Big),
\]
subject to
\[
\Lambda^{(1)}+\Lambda^{(2)}-\Lambda^{(3)}\in Q
\]
[2509.01949]. The root-lattice condition is then enforced by BRST projection in the free-field realization.

A broader coset background appears in the study of the GKO family
\[
\mathcal{M}(r,p,n)
=
\frac{\widehat{\mathfrak{sl}(r)}_{p}\times\widehat{\mathfrak{sl}(r)}_{n-p}}
{\widehat{\mathfrak{sl}(r)}_{n}},
\]
with chiral algebra $\mathcal A(r,p)$ and central charge
\[
c=\frac{p(r^2-1)}{p+r}+\frac{r(r^2-1)}{p}Q^2,
\qquad
Q=b+\frac{1}{b}
\]
[1411.3313]. That paper does not construct a full generic coset free-field resolution, but it does describe $\mathcal A(r,p)$ through parafermionic Toda theory with Lagrangian
\[
\mathcal{L}
=
\mathcal{L}_{\textrm{PF}}
+\frac{p}{8\pi}\left(\partial_a\varphi\right)^2
+\mu\sum_{k=1}^{r-1}\Psi_{e_k}\bar{\Psi}_{e_k}e^{b(e_k,\varphi)},
\]
and it characterizes local integrals of motion as commuting with the integrated fields
\[
\int_{\mathcal C}\Psi_{e_k}\bar{\Psi}_{e_k}e^{b(e_k,\varphi)}\,dx
\]
[1411.3313]. This establishes a screening-type, boson-plus-parafermion background for coset algebras, even where a full BRST resolution is absent.

## 4. Ishibashi states and boundary-sector realization

The most direct application of coset free-field resolution in the present corpus is the construction of diagonal ADE coset $\mathcal W$-minimal Ishibashi states [2509.01949]. The starting point is the tensor product of two affine dominant-module resolutions,
\[
\big(
\widetilde F_{\widehat\Lambda^{(1)}}\otimes
\widetilde F_{\widehat\Lambda^{(2)}},
\widetilde d^{(1)},\widetilde d^{(2)}
\big),
\qquad
[\widetilde d^{(1)},\widetilde d^{(2)}]=0,
\]
followed by projection to fixed diagonal $\widehat g^{(3)}$-weight באמצעות the BRST-invariant operator
\[
\widehat P_{\Lambda^{(3)}}
=
\int_0^1 d\theta\;
e^{2\pi i\theta\cdot(\widehat h_0-\Lambda^{(3)})},
\qquad
\widehat h_0
=
h_0^{(1)}+h_0^{(2)}+\text{ghost terms}
\]
[2509.01949].

The projected resolution complex is
\[
\Big[
\widehat P_{\,w^{(1)}*\Lambda^{(1)}+w^{(2)}*\Lambda^{(2)}-\Lambda^{(3)}}
\big(
F_{\widehat w^{(1)}*\widehat\Lambda^{(1)}}^\phi
\otimes
F_{\widehat w^{(2)}*\widehat\Lambda^{(2)}}^\phi
\otimes
F^{\beta^\alpha\gamma^\alpha}
\big),
\widetilde d^{(1)},\widetilde d^{(2)}
\Big],
\]
whose zeroth cohomology is conjectured to be the irreducible $\mathcal W$-module $L_{(\Lambda^{(+)},\Lambda^{(-)})}$ [2509.01949].

The corresponding Ishibashi state is expressed as
\[
\big|L_{(\Lambda^{(+)},\Lambda^{(-)})}\big\rangle\!\big\rangle_\Omega
=
\sum_{\widehat w^{(1,2)}}
\widehat P_{\,w^{(1)}*\Lambda^{(1)}+w^{(2)}*\Lambda^{(2)}-\Lambda^{(3)}}
\,
\theta_{\widehat w_1,\widehat w_2}
\,
\big|F_{\widehat w^{(1)}*\widehat\Lambda^{(1)}}\big\rangle\!\big\rangle_\Omega
\otimes
\big|F_{\widehat w^{(2)}*\widehat\Lambda^{(2)}}\big\rangle\!\big\rangle_\Omega
\otimes
\big|F^{\beta^\alpha\gamma^\alpha}\big\rangle\!\big\rangle_I.
\]
Each bosonic block is
\[
\big|F_{\widehat w^{(i)}*\widehat\Lambda^{(i)}}\big\rangle\!\big\rangle_\Omega
=
\exp\!\Big[
\sum_{n\in\mathbb Z_+}
\frac{-1}{n}\,
a_{-n}\cdot V_\Omega\cdot \bar a_{-n}
\Big]
|p\otimes \bar p\rangle,
\qquad
p=\alpha_+\widehat w^{(i)}*\widehat\Lambda^{(i)},
\]
while the $\beta\gamma$ Ishibashi state is
\[
\big|F^{\beta^\alpha\gamma^\alpha}\big\rangle\!\big\rangle_I
=
\exp\!\Big[
-\sum_{\alpha\in\Delta_+}\sum_{n\in\mathbb Z_+}
(\beta_{-n}\bar\gamma_{-n}+\gamma_{-n}\bar\beta_{-n})
\Big]
|0\otimes \bar 0\rangle
\]
[2509.01949]. The gluing data are encoded by
\[
V_\Omega=I,U,
\qquad
\bar p=-V_\Omega\cdot(p+\alpha_0\rho)-\alpha_0\rho.
\]

In this construction, the BRST projector is the mechanism that turns a tensor product of numerator free-field blocks into a physical coset boundary state. The free-field resolution thereby functions not only as a statement about irreducible modules, but also as a practical boundary-CFT tool.

## 5. Free-field perspectives that stop short of a full resolution

A substantial adjacent literature adopts free-field methods for coset theories without producing an exact Felder or Wakimoto resolution. A notable example is the large-$k$, exact-in-$\lambda$ treatment of the $\lambda$-deformed $SU(2)/U(1)$ coset. There the construction is explicitly described as not being a full exact Coulomb-gas/Wakimoto-style free-field resolution of the undeformed coset CFT Hilbert space; instead it gives a perturbative free-field description around a “free-field point,” with two free bosons, $\lambda$-dependent interaction vertices, and $\lambda$-dressed parafermions carrying an essential Wilson-like phase [2004.10216]. The dressed parafermions are written as
\[
\Psi_+=\left(\partial x_- + F_+\right)e^{-i\phi_1},
\qquad
\Psi_-=\left(\partial x_+ - F_-\right)e^{i\phi_1},
\]
with analogous antiholomorphic expressions, and the framework reproduces the known $\beta$-function, anomalous dimensions, and four-point functions at leading order in $1/k$ [2004.10216]. This is a free-field realization in spirit, but not a cohomological resolution of the coset module category.

Another neighboring development is the study of distinguished bases in coset-related theories associated with instantons on $\mathbb C^2/\mathbb Z_p$. For the case $p=2$, $r=2$, the working algebra is
\[
\mathcal A=\mathcal H\oplus \mathcal H\oplus \mathcal F\oplus \mathsf{NSR},
\]
and the paper exhibits two commuting Virasoro algebras inside $\mathcal F\oplus\mathsf{NSR}$, together with the decomposition
\[
\pi_{\mathcal F\oplus\mathsf{NSR}}
=
\bigoplus_{k\in\mathbb Z}\pi^k_{\mathsf{Vir}\oplus\mathsf{Vir}}
\]
[1111.2803]. This leads to basis vectors built from two copies of the AFLT construction and to fully factorized matrix elements. The result is highly suggestive for coset free-field resolution, because it resolves a complicated coset-related representation into simpler Virasoro building blocks; however, the paper explicitly does not provide a BRST differential, screening complex, or Felder-type resolution [1111.2803].

These examples show that “free-field approach” can mean several distinct things: exact cohomological resolution, perturbative free-boson expansion, or algebraic decomposition into simpler commuting subalgebras.

## 6. Generalizations, neighboring uses, and recurrent misconceptions

A recurrent misconception is to identify any coset reformulation using simpler variables with a coset free-field resolution. Several recent works make the distinction explicit. The study of integrable auxiliary-field deformations of $\mathbb Z_N$-coset models is not a paper on coset free-field resolutions in the usual CFT sense; it studies classical two-dimensional integrable sigma models on coset target spaces and introduces an auxiliary-field enlargement with constrained Lax and Hamiltonian structures, but no Wakimoto representation, Coulomb-gas/Felder complex, BRST resolution, screening operators, or vertex-operator realization of coset primaries [2409.04523]. Likewise, the effective field theory for superfluid vortex lattices uses the coset construction method in the EFT sense to isolate invariant low-energy data and the Tkachenko mode; this is a symmetry-based reduction of Goldstone content, not a chiral free-field resolution of a coset CFT [2507.13352].

A different neighboring usage appears in the path-integral treatment of a driven semiconductor, where quartic fermion interactions are rewritten through anomalous-doubled self-energy matrices, then factorized into a subgroup background and a coset sector, and finally transformed from curved coset coordinates to locally Euclidean variables with flat integration measure [1004.1992]. This is a geometric “free-field” flattening of a coset matrix integral, not a BRST or screening resolution of affine-coset modules.

Generalizations of the GKO idea also remain relevant. The abstract of “On a Generalization of GKO Coset Construction of Conformal Field Theories” states that it introduces a generalization of the GKO construction based on a choice of a scaled affine subalgebra $\hat{\mathfrak h}^s$ of a given affine Lie algebra $\hat{\mathfrak h}$ and studies the Ising CFT as a generalized GKO coset of $\mathrm{su}(2)_1$ with a scaling factor $s=2$ [1512.08243]. This suggests a natural interface with free-field or BRST techniques, because a modified denominator algebra changes the branching-space problem; a precise statement about the corresponding free-field resolution would require the unavailable technical text.

In the strict encyclopedic sense, therefore, coset free-field resolution refers most precisely to the affine/QDS/BRST construction in which numerator free-field complexes, screenings, and ghost systems are combined to produce the physical branching spaces of coset theories, with diagonal ADE $\mathcal W$ minimal models providing the clearest fully articulated example [2509.01949]. Other uses of “free-field perspective” on cosets are mathematically and physically important, but they occupy adjacent rather than identical positions in the theory.

Source: https://www.emergentmind.com/topics/coset-free-field-resolution