---
title: Equivariant Fried Conjecture
url: https://www.emergentmind.com/topics/equivariant-fried-conjecture
type: topic
---

# Equivariant Fried Conjecture

The Equivariant Fried Conjecture is the problem of relating a symmetry-refined dynamical zeta function of a flow to a symmetry-refined analytic torsion invariant. In the modern formulation developed for proper cocompact group actions, ordinary periodicity is replaced by periodicity up to a fixed group element \(g\), so that one studies flow trajectories satisfying \(\gamma(l)=g\gamma(0)\), together with a \(g\)-delocalized trace on the torsion side. This viewpoint was formalized for unimodular locally compact groups acting properly and cocompactly on smooth manifolds, and it was later proved for a substantial class of suspension flows. A broader surrounding literature studies closely related “twisted” or representation-theoretic Fried statements for locally symmetric spaces, Anosov flows, and pseudo-Anosov flows; these results are not uniformly equivariant in the same sense, but they supply much of the analytic, harmonic-analytic, and torsion-theoretic infrastructure of the subject [2303.00312], [2507.06792].

## 1. Formal setting and conjectural statement

The equivariant framework is built from the following data: a smooth manifold \(M\), a unimodular locally compact group \(G\) acting properly on \(M\) with compact quotient \(M/G\), a \(G\)-equivariant flow \(\varphi\) generated by a smooth nowhere-vanishing vector field, a flat \(G\)-equivariant vector bundle \(F\to M\) with \(G\)-invariant flat connection \(\nabla^F\), and a fixed element \(g\in G\) whose centralizer \(Z=Z_G(g)\) is such that \(G/Z\) carries a nonzero \(G\)-invariant Borel measure. In this setting, the basic dynamical objects are \(g\)-periodic trajectories rather than closed orbits in the ordinary sense [2303.00312].

A central hypothesis is \(g\)-nondegeneracy. The flow is called \(g\)-nondegenerate if for all \(l\in\mathbb R\) and \(m\in M\) with \(\varphi_l(m)=gm\),
\[
\det\bigl( (T_{m}(\varphi_{l}\circ g^{-1}) - 1)|_{T_{m}M/\mathbb{R} u(m)} \bigr)\not=0.
\]
Equivalently, the induced map \(P_\gamma^g-1\) on \(T_{\gamma(0)}M/\mathbb{R}u(\gamma(0))\) is invertible for every \(g\)-periodic flow line. This is the equivariant analogue of the ordinary nondegeneracy condition for periodic orbits [2303.00312].

The conjectural Fried-type relation asks whether, under odd-dimensionality and vanishing of the relevant Laplacian kernel, the equivariant Ruelle zeta function extends meromorphically to a neighborhood of \(0\), is regular there, and satisfies an identity at \(0\) with equivariant analytic torsion. In the formulation of Hochs–Saratchandran recorded in the equivariant-zeta paper, the expected identity is
\[
R^g_{\varphi, \nabla^F}(0) = T_g(\nabla^F).
\]
In the suspension-flow normalization later used by Hochs and Pirie, the expected identity is
\[
R^g_{\phi,\nabla^E}(0)=T_g(\nabla^E)^2.
\]
The difference is a normalization issue rather than a contradiction; in the compact unitary case with \(G\) compact and \(g=e\), the suspension-flow paper states that \(R^e_{\phi,\nabla^E}\) is the absolute value squared of the classical Ruelle zeta in the relevant normalization [2303.00312], [2507.06792].

## 2. Equivariant dynamical zeta functions and equivariant torsion

For \(x\in G\), the \(x\)-delocalized length spectrum is
\[
L_{x}(\varphi) := \{l \in \mathbb{R}; \exists m \in M: \varphi_l(m) = x m\}.
\]
If \(l\in L_x(\varphi)\), an \((x,l)\)-periodic flow curve is a flow curve \(\gamma\) with \(\gamma(l)=x\gamma(0)\), and \(\Gamma_l^x(\varphi)\) denotes the set of such curves modulo time shift. For \(\gamma\in\Gamma_l^x(\varphi)\), one has the equivariant Poincaré map \(P_\gamma^x\), the holonomy endomorphism \(\rho_x(\gamma)\), and a cutoff-weighted primitive period \(T_\gamma\) defined from a cutoff function \(\chi\) adapted to the proper action [2303.00312].

With these ingredients, the equivariant Ruelle dynamical zeta is defined by
\[
2\log R_{\varphi, \nabla^F}^g(\sigma) =
\int_{G/Z} \sum_{l \in L_{g}(\varphi)}
\frac{ e^{-|l|\sigma}}{|l|}
\sum_{\gamma \in \Gamma^{hgh^{-1}}_l(\varphi)}
\sgn\bigl(\det(1-P_{\gamma}^{hgh^{-1}}) \bigr)
\tr(\rho_{hgh^{-1}}(\gamma)) T_{\gamma}\, d(hZ),
\]
whenever the expression converges for \(\Re(\sigma)\) sufficiently large. The construction permits negative lengths, a feature emphasized as essential already in elementary examples such as the translation flow on \(\mathbb R\) [2303.00312].

The analytic side uses a \(g\)-trace. For a bounded \(G\)-equivariant operator \(T\),
\[
\Tr_g(T) := \int_{G/Z} \Tr(hgh^{-1} T\chi)\, d(hZ),
\]
whenever \(T\chi\) is trace class and the integral converges. If \(\Delta_F^p\) is the Laplacian on \(L^2\Omega^p(M;F)\) and \(P_p\) is the orthogonal projection onto \(\ker_{L^2}(\Delta_F^p)\), then
\[
T_g(t) := \sum_{p=1}^{\dim(M)} (-1)^p p \Tr_g(e^{-t\Delta_F^p} - P_p),
\]
and the equivariant analytic torsion is
\[
T_g(\nabla^F) := \exp\left( -\frac{1}{2} \left. \frac{d}{ds}\right|_{s=0} \frac{1}{\Gamma(s)} \int_0^1 t^{s-1} T_g(t) \, dt -\frac{1}{2} \int_1^\infty t^{-1}  T_g(t) \, dt \right).
\]
This is a \(g\)-localized or delocalized torsion invariant, and for \(g=e\) it reduces to the von Neumann-trace version [2303.00312].

In the suspension-flow setting, the geometric structure becomes more rigid. If \(M=(Y\times \mathbb R)/\mathbb Z\) is the mapping torus of a \(G\)-equivariant isometry \(T:Y\to Y\), then the \(g\)-periodic flow problem reduces to fixed points of the maps \(g^{-1}T^n\) on \(Y\), and the equivariant torsion admits a fibration formula. The main identity proved there is
\[
R^g_{\phi,\nabla^E}(\sigma) = \exp\big(\sigma\,\chi_{(g,0)}(\nabla^{E_1})\big)\, T_g(\nabla^E,\sigma^2)^2,
\]
which yields \(R^g_{\phi,\nabla^E}(0)=T_g(\nabla^E)^2\) whenever \(T_g(\nabla^E)\) is well-defined [2507.06792].

## 3. Meanings of “equivariant” and adjacent twisted formulations

In the strict sense relevant to proper cocompact group actions, “equivariant” refers to dependence on a symmetry element \(g\in G\), \(g\)-periodic orbits, \(g\)-traces, and delocalized orbital integrals. This is the sense formalized in the 2023 equivariant-zeta framework and realized in the 2025 suspension-flow theorem [2303.00312], [2507.06792].

A broader literature uses “equivariant” in a looser, representation-theoretic sense. Shen’s work on admissible twists of the Fried conjecture is explicit that it is not a paper on equivariant analytic torsion in the classical sense of torsion depending on an external compact group action, nor on a representation-valued Ruelle zeta function. The objects remain scalar-valued after taking traces. Nevertheless, the twisting bundle is obtained by restricting a finite-dimensional representation \(\rho:G\to \mathrm{GL}(E)\), the Ruelle zeta weights closed orbits by \(\mathrm{Tr}[\rho(\gamma)]\), and the proof uses orbital integrals, trace formulas, and spectral decompositions by irreducible unitary \(G\)-representations. In that precise sense, the paper gives a strong representation-theoretic analogue of an equivariant Fried picture [2009.04897].

The same terminological broadening appears in the dynamical literature on twisted Fried conjectures for singular hyperbolic systems. For smooth pseudo-Anosov flows on closed \(3\)-manifolds, the zeta function \(\zeta_{\varphi,\rho}(s)\) is twisted by a representation \(\rho\) of \(\pi_1(M)\), and the main theorem identifies \(|\zeta_{\varphi,\rho}(0)|^{-1}\) with twisted Reidemeister torsion under acyclicity and monodromy assumptions. The paper presents this as one of the standard manifestations of the “equivariant Fried conjecture,” even though it does not introduce an external symmetry group action or an equivariant trace formalism [2409.17014].

This terminological distinction matters. Some works prove a genuinely equivariant statement in the sense of \(g\)-delocalized traces and proper actions; others prove twisted or character-weighted analogues in which the output is still scalar after taking traces. The literature uses both viewpoints, but it does not identify them.

## 4. Established cases and principal results

The subject presently consists of a general conjectural framework, one proved equivariant case in substantial generality, several verified model examples, and a larger body of twisted or representation-theoretic analogues.

| Setting | Main identity | Status |
|---|---|---|
| Proper cocompact \(G\)-actions, \(g\)-nondegenerate equivariant flows | \(R^g_{\varphi,\nabla^F}(0)=T_g(\nabla^F)\) | Formulated as a question, verified in explicit examples [2303.00312] |
| Suspension flow of a \(G\)-equivariant isometry | \(R^g_{\phi,\nabla^E}(0)=T_g(\nabla^E)^2\) | Proved in several cases [2507.06792] |
| Closed odd-dimensional locally symmetric spaces, admissible \(G\)-induced flat bundles | \(|R_\rho(0)|=T(F)^2\), and \(R_\rho(0)=T(F)^2\) if \(\rho\simeq \rho^\theta\) | Proved [2009.04897] |
| Smooth pseudo-Anosov flows on closed \(3\)-manifolds | \(|\zeta_{\varphi,\rho}(0)|^{-1}=\tau_\rho(M)\) | Proved under acyclicity and monodromy assumptions [2409.17014] |
| Smooth Anosov flows in dimension \(3\) | \(|\zeta_{X,\rho}(0)|^{-1}=\tau_\rho(M)\) in the stated regime | Proved, with local constancy under a spectral condition in all dimensions [1807.01189] |

The 2023 equivariant-zeta paper provides the formal definition, an equivariant generalization of Guillemin’s trace formula, and explicit computations in examples where the classical Ruelle zeta is not defined. It verifies the conjectural equality for the line, the circle, and certain discrete Euclidean-motion examples when the Laplacian kernel vanishes, but it does not claim a general theorem [2303.00312].

The 2025 suspension-flow paper proves the equivariant Fried conjecture for the suspension flow of an equivariant isometry in several cases: the case where the group is compact, the case where the group element has compact centralizer and closed conjugacy class, and the case of the identity element of a non-compact discrete group. Under its hypotheses, including \(\ker(\Delta_E)=0\), fixed-point discreteness assumptions, and either compactness or growth/Novikov–Shubin conditions, it obtains the exact identity \(R^g_{\phi,\nabla^E}(0)=T_g(\nabla^E)^2\) [2507.06792].

The nearby locally symmetric literature gives higher-rank and non-unitary evidence in a different formalism. Shen proves that for a closed odd-dimensional locally symmetric manifold \(Z=\Gamma\backslash G/K\) and a flat bundle induced from a finite-dimensional \(G\)-representation with admissible metric,
\[
R_\rho(\sigma)= C_\rho\, T(F)^2\, \sigma^{r_\rho}+O(\sigma^{r_\rho+1}),
\]
with \(|R_\rho(0)|=T(F)^2\) in the acyclic case and \(R_\rho(0)=T(F)^2\) when \(\rho\simeq \rho^\theta\) [2009.04897]. Earlier, Shen had proved the unitary locally symmetric case, obtaining \(R_\rho(0)=T(F)^2\) for acyclic unitarily flat bundles on closed odd-dimensional locally symmetric reductive manifolds [1602.00664]. Müller extended Fried’s hyperbolic theorem from orthogonal twists to arbitrary finite-dimensional representations by replacing Ray–Singer torsion with Cappell–Miller complex-valued torsion [2005.01450].

Beyond locally symmetric geometry, Dang–Guillarmou–Rivière–Shen proved the Fried conjecture in dimension \(3\) for smooth Anosov flows with unitary acyclic local systems, and more generally showed that \(\zeta_{X,\rho}(0)\) is locally constant under a no-zero-resonance condition. This yields new examples near hyperbolic geodesic flows and is explicitly described as structurally relevant for future equivariant extensions [1807.01189]. The pseudo-Anosov extension to singular flows on closed \(3\)-manifolds pushes the twisted theory beyond the Anosov category while retaining an exact orbit-product formula and a torsion identity at \(s=0\) [2409.17014].

## 5. Analytic and representation-theoretic methods

The defining analytic mechanism of the strict equivariant theory is an equivariant generalization of Guillemin’s trace formula. The 2023 framework introduces a distributional flat \(g\)-trace \(\Tr_g^\flat\) and proves that the equivariant zeta can be recovered from it by
\[
2 \log R_{\varphi, \nabla^F}^g(\sigma) =- \langle \Tr_g^{\flat}((-1)^N N \Phi^*), \psi_{\sigma} \rangle,
\]
with \(\psi_\sigma(t)=e^{-\sigma|t|}/|t|\). This formula is the direct equivariant analogue of the classical relation between periodic-orbit expansions and flat traces, and it explains the sign \(\sgn(\det(1-P_\gamma^g))\) through a linear-algebra identity involving \(\sum_j (-1)^j j\,\tr(\wedge^j A)\) [2303.00312].

In the suspension-flow theorem, tractability comes from the mapping-torus geometry. A \(g\)-periodic orbit of the suspension flow must correspond to an integer time \(n\in\mathbb Z\setminus\{0\}\) and a fixed point of \(g^{-1}T^n\) on the fiber \(Y\). The proof then combines: reduction of orbit sums to fixed-point sums on \(Y\); a fibration formula for equivariant torsion; explicit one-dimensional torsion on the \(\mathbb R/\mathbb Z\)-direction; and a noncompact Atiyah–Bott-type fixed-point formula rewriting the mixed equivariant Euler characteristic \(\chi_{(g,n)}\) in terms of the same fixed-point data that appear in the zeta function. The equality is therefore term-by-term after the substitution \(\sigma\mapsto \sigma^2\) on the torsion side [2507.06792].

The representation-theoretic branch of the literature uses a different arsenal. In the admissible-twist locally symmetric setting, the Ruelle zeta is decomposed into Selberg zeta factors attached to virtual \(M\)-representations \(\eta_\beta\), with character identities such as
\[
\sum_{\beta\in b^*} e^{\langle \beta,a\rangle}\, \mathrm{Tr}_s[\eta_\beta(k^{-1})] = \frac{\mathrm{Tr}[\rho(e^a k^{-1})]} {\det(1-\mathrm{Ad}(e^a k^{-1})|_{z^\perp(b)})^{1/2}},
\]
and determinant formulas for the Selberg zetas. The proof uses Selberg trace formula techniques, Bismut orbital integral formulas, Dirac cohomology, and cohomological formulas for correction exponents. This is one reason the paper is often viewed as highly relevant to an equivariant Fried picture even though its final theorem is scalar [2009.04897].

The pseudo-Anosov and general Anosov works exhibit yet another route. For smooth pseudo-Anosov flows, the key tools are singular local models, Markov partitions with connected regular and singular rectangles, Baladi–Tsujii anisotropic spaces, symbolic dynamical determinants, and a filtered cellular chain complex computing twisted Reidemeister torsion. The exact correction from symbolic zeta to geometric zeta is expressed through distinguished stable and unstable boundary orbits [2409.17014]. In the smooth Anosov setting, microlocal analysis of Pollicott–Ruelle resonances yields a deformation theorem: if \(0\) is not a resonance, then \(X\mapsto \zeta_{X,\rho}(0)\) is locally constant. This resonance-based rigidity is explicitly highlighted as suggestive for future equivariant refinements [1807.01189].

## 6. Scope, normalization issues, and unresolved problems

The Equivariant Fried Conjecture is not a theorem in full generality. The 2023 framework asks precisely under what additional hypotheses the defining series converges, admits meromorphic continuation, is regular near \(0\), and satisfies the conjectural equality. The same paper explicitly states that the authors do not expect the equality to hold in complete generality. A key warning comes from the universal-cover situation: if \(M\) is the universal cover of a compact manifold and \(G=\pi_1(X)\), the classical Ruelle zeta decomposes into factors indexed by conjugacy classes, but equality with torsion does not hold term-by-term. In particular, the trivial conjugacy class \(g=e\) can yield \(R_{\varphi,\nabla^F}^e=1\) when there are no closed lifted trajectories, whereas the torsion side is the \(L^2\)-analytic torsion, which is generally nontrivial [2303.00312].

A persistent source of confusion is normalization. One strand of the literature states the conjectural equality as \(R^g(0)=T_g\); the suspension-flow theorem proves \(R^g(0)=T_g^2\). The suspension paper explains this by noting that in its normalization \(R^e\) is the absolute value squared of the classical Ruelle zeta in the relevant compact unitary setting. Similar normalization differences already occur in nonequivariant Fried theory: Shen’s locally symmetric papers use a convention in which the natural comparison is with \(T(F)^2\) rather than \(T(F)\) [2507.06792], [1602.00664].

Another common misconception is that every “equivariant Fried” result is equivariant in the strict group-action sense. Shen’s admissible-twist paper, Müller’s non-unitary hyperbolic paper, the Anosov work in small dimensions, and the pseudo-Anosov extension are all deeply representation-theoretic and twisted, but they remain scalar-valued after traces and do not define a group-valued or character-valued torsion theory [2009.04897], [2005.01450], [1807.01189], [2409.17014].

The current proven strict-equivariant regime is narrow but substantive. The suspension-flow theorem is limited to mapping tori of equivariant isometries, with further hypotheses such as \(\ker(\Delta_E)=0\), discreteness or emptiness of the fixed-point sets \(Y^{g^{-1}T^n}\), and in some cases positivity of Novikov–Shubin numbers or growth conditions on conjugacy classes. It does not treat general Anosov or geodesic flows with noncompact group actions [2507.06792].

A plausible implication is that a fully developed general theory would need several ingredients that are currently separate in the literature: equivariant flat traces and twisted trace formulas of the sort developed for proper actions; equivariant or delocalized analytic torsion with robust metric-independence; and a resonance or harmonic-analysis formalism capable of handling noncompact symmetry, non-unitary twisting, and zero-mode corrections simultaneously. The existing papers isolate these components in different settings, but they do not yet assemble them into a single general theorem [2303.00312], [1807.01189], [2009.04897].

Source: https://www.emergentmind.com/topics/equivariant-fried-conjecture