---
title: Livshitz-Type Theorem in Dynamical Systems
url: https://www.emergentmind.com/topics/livshitz-type-theorem
type: topic
---

# Livshitz-Type Theorem in Dynamical Systems

Searching arXiv for relevant Livšic/Livshitz-type theorem papers to support the article.
A Livshitz-type theorem, in the sense represented by the papers considered here, is a rigidity statement in which periodic orbit data imply a cohomological or geometric conclusion. In its classical additive form, for a transitive Anosov diffeomorphism or flow on a closed connected Riemannian manifold, a Hölder observable is a coboundary when all of its periods over closed orbits vanish. More recent formulations retain the same periodic-orbit philosophy while changing the amount of periodic data, the target of the cocycle, or the dynamical category: vanishing on a positive weighted proportion of periodic orbits already forces coboundary triviality for Hölder observables on transitive Anosov systems; a periodic Jacobian identity forces transitivity and a unique \(C^1\) invariant volume form for \(C^2\) Anosov endomorphisms; and the periodic orbit obstruction is sufficient for Hölder cocycles with values in the group of germs at the origin of analytic diffeomorphisms of \(\mathbb C^d\) [2304.01372] [2606.15542] [1110.1911].

## 1. Classical cohomological framework

For a transitive Anosov diffeomorphism \(f:M\to M\) or a transitive Anosov flow \(f_t:M\to M\) on a closed connected Riemannian manifold \(M\), the basic object is a Hölder function \(\varphi:M\to\mathbb R\). Two Hölder functions \(\varphi_1,\varphi_2\) are cohomologous if there exists a Hölder \(u:M\to\mathbb R\) such that
\[
\varphi_1-\varphi_2=u\circ f-u
\]
in discrete time, or
\[
\varphi_1-\varphi_2=\left.\frac{d}{dt}\right|_{t=0}u\circ f_t
\]
for flows. A coboundary is a function cohomologous to \(0\) [2304.01372].

The periodic data are defined on the set \(\mathcal P\) of closed orbits. If \(\gamma\in\mathcal P\), its length is \(\ell(\gamma)\). For a diffeomorphism, \(\ell(\gamma)\) is the least period \(n\), and for \(p\in\gamma\),
\[
\ell_\varphi(\gamma)=S_n\varphi(p):=\sum_{k=0}^{n-1}\varphi(f^k p).
\]
For a flow,
\[
\ell_\varphi(\gamma)=\int_\gamma \varphi\,dt=\int_0^{\ell(\gamma)}\varphi(f_s x)\,ds.
\]
The classical Livšic theorem states that if \(\ell_\varphi(\gamma)=0\) for every closed orbit \(\gamma\), then \(\varphi\) is a coboundary [2304.01372].

A parallel multiplicative formulation appears for group-valued cocycles. If \(T:X\to X\) is a topologically transitive homeomorphism of a compact metric space satisfying the closing property, and \(F:X\to G\) is a Hölder cocycle with values in a noncommutative group, the periodic orbit obstruction requires that the cocycle product along every periodic orbit be the identity. For cocycles with values in the group \(\mathcal G\!erm_d\) of germs of analytic diffeomorphisms of \(\mathbb C^d\) fixing the origin, this means
\[
F(T^{n-1}p)\circ\cdots\circ F(p)=\mathrm{id}_{\mathbb C^d}
\quad\text{whenever }T^n p=p.
\]
The problem is again whether periodic orbit obstructions are sufficient for solvability of a cohomological equation [1110.1911].

## 2. Positive-proportion rigidity for Anosov systems

The principal strengthening in "A positive proportion Livshits theorem" replaces the hypothesis that all periods vanish by a weighted density condition on the zero-period set
\[
Q=\{\gamma\in\mathcal P:\ell_\varphi(\gamma)=0\}.
\]
For a subset \(A\subset\mathcal P\), the paper writes \(A(n)=\{\gamma\in A:\ell(\gamma)=n\}\) in discrete time and \(A(T,\Delta)=\{\gamma\in A:\ell(\gamma)\in (T,T+\Delta]\}\) for flows. Given a Hölder weight \(\psi:M\to\mathbb R\), each orbit is counted with weight \(e^{\ell_\psi(\gamma)}\), and positive asymptotic upper density means that the corresponding weighted proportion has positive limsup [2304.01372].

For transitive Anosov diffeomorphisms, Theorem 1.1 states that if there exists a Hölder \(\psi:M\to\mathbb R\) such that
\[
\limsup_{n\to\infty}
\frac{\sum_{\gamma\in Q(n)} e^{\ell_\psi(\gamma)}}
{\sum_{\gamma\in\mathcal P(n)} e^{\ell_\psi(\gamma)}}>0,
\]
then \(\varphi\) is a coboundary. In the unweighted case \(\psi=0\), this becomes ordinary counting density,
\[
\limsup_{n\to\infty}\frac{\#Q(n)}{\#\mathcal P(n)}>0.
\]
Thus vanishing on a positive asymptotic upper density subset of period-\(n\) orbits already forces cohomological triviality [2304.01372].

For transitive Anosov flows, Theorem 1.2 assumes additionally that the stable and unstable distributions are not jointly integrable. If \(\Delta>0\) and there exists a Hölder \(\psi\) such that
\[
\limsup_{T\to\infty}
\frac{\sum_{\gamma\in Q(T,\Delta)} e^{\ell_\psi(\gamma)}}
{\sum_{\gamma\in\mathcal P(T,\Delta)} e^{\ell_\psi(\gamma)}}>0,
\]
then \(\varphi\) is a coboundary. If also \(P(\psi)>0\), the same conclusion holds with cumulative counting sets \(\mathcal P(0,T)\) and \(Q(0,T)\) in place of windowed sets \(\mathcal P(T,\Delta)\) and \(Q(T,\Delta)\) [2304.01372].

The significance of this result lies in the replacement of a universal periodic verification condition by a positive-proportion hypothesis in an asymptotic weighted counting sense. The paper explicitly presents this as a genuine strengthening of the classical theorem [2304.01372].

## 3. The strengthened nonpositive theorem

The same paper also treats the nonpositive Livšic theorem. The classical form states that if
\[
\ell_\varphi(\gamma)\le 0 \quad \forall\gamma\in\mathcal P,
\]
then \(\varphi\) is cohomologous to a nonpositive function. Concretely, there exists a Hölder \(u\) such that
\[
\varphi-(u\circ f-u)\le 0
\]
for diffeomorphisms, or
\[
\varphi-\left.\frac{d}{dt}\right|_{t=0}u\circ f_t\le 0
\]
for flows [2304.01372].

The strengthened version allows positive periods, provided they are sufficiently sparse. With
\[
Q'=\{\gamma\in\mathcal P:\ell_\varphi(\gamma)>0\},
\]
Theorem 1.3 states that for a transitive Anosov diffeomorphism, if
\[
\lim_{n\to\infty}\frac1n\log |Q'(n)|=0,
\]
then \(\varphi\) is cohomologous to a nonpositive function. Theorem 1.4 gives the flow analogue: if
\[
\lim_{T\to\infty}\frac1T\log |Q'(T,\Delta)|=0,
\]
then \(\varphi\) is cohomologous to a nonpositive function, and the same holds with \(Q'(0,T)\) in place of \(Q'(T,\Delta)\) [2304.01372].

The exceptional set therefore need not be empty; it may even be infinite. What matters is subexponential growth. This is strictly stronger than the classical theorem because it admits positive periods on infinitely many periodic orbits while preserving the same cohomological conclusion [2304.01372].

The paper also gives a limitation showing that zero weighted density is not sufficient for the nonpositive conclusion. On the full one-sided shift on two symbols, with
\[
\varphi(x)=(-1)^{x_1},
\]
and Bernoulli-type weight
\[
\psi(x)=
\begin{cases}
\log p,& x_1=1,\\
\log(1-p),& x_1=0,
\end{cases}
\qquad p\in(0,1/2),
\]
the law of large numbers implies that positive-period orbits form a zero-density set with respect to \(\psi\), but \(\varphi\) is not cohomologous to a nonpositive function. The paper uses this example to justify the stronger hypothesis of subexponential growth rather than mere zero weighted density [2304.01372].

## 4. Statistical and thermodynamic mechanism

The positive-proportion theorem is driven by a finer statistical description of periodic orbit sums. For a Hölder weight \(\psi\), the paper introduces weighted periodic-orbit measures
\[
\mu_{n,\psi}=
\frac{\sum_{\gamma\in\mathcal P(n)} e^{\ell_\psi(\gamma)}\delta_\gamma}
{\sum_{\gamma\in\mathcal P(n)} e^{\ell_\psi(\gamma)}}
\]
in discrete time and
\[
\mu_{T,\Delta,\psi}=
\frac{\sum_{\gamma\in\mathcal P(T,\Delta)} e^{\ell_\psi(\gamma)}\delta_\gamma}
{\sum_{\gamma\in\mathcal P(T,\Delta)} e^{\ell_\psi(\gamma)}}
\]
for flows. Under these measures, \(\ell_\varphi(\gamma)\) behaves asymptotically like a normal random variable after scaling by \(\sqrt n\) or \(\sqrt T\), provided the dynamical variance is positive [2304.01372].

In discrete time, the paper cites Coelho–Parry: if \(\sigma^2_{\varphi,\psi}>0\) and the equilibrium state \(\mu_\psi\) satisfies \(\mu_\psi(\varphi)=0\), then
\[
\frac{\ell_\varphi(\gamma)}{\sqrt n}
\]
on \(\mathcal P(n)\), weighted by \(e^{\ell_\psi(\gamma)}\), converges in distribution to a centered normal law with standard deviation \(\sigma_{\varphi,\psi}\). For flows, the paper proves a weighted version of Cantrell–Sharp’s periodic-orbit CLT: if \(P(\psi)>0\), \(\sigma^2_{\varphi,\psi}>0\), and \(\mu_\psi(\varphi)=0\), then
\[
\frac{\ell_\varphi(\gamma)}{\sqrt T}
\]
on \(\mathcal P(T,\Delta)\), weighted by \(e^{\ell_\psi(\gamma)}\), converges to a centered normal law with standard deviation \(\sigma_{\varphi,\psi}\) [2304.01372].

The contradiction mechanism is elementary in form but strong in consequence. If \(Q=\{\gamma:\ell_\varphi(\gamma)=0\}\) has positive asymptotic weighted upper density, then a positive proportion of the weighted orbit distribution is concentrated at a single value. A nondegenerate Gaussian limit is non-atomic, so this cannot occur when the variance is positive. If \(\mu_\psi(\varphi)\neq 0\), the zero-period condition drifts under centering and scaling; if \(\mu_\psi(\varphi)=0\), the Gaussian limit still rules out positive mass at a point. The paper concludes that the variance must vanish, and a standard thermodynamic fact then implies that \(\varphi\) is cohomologous to a constant; since there is at least one zero period, that constant is \(0\), hence \(\varphi\) is a coboundary [2304.01372].

Thermodynamic formalism enters through pressure, equilibrium states, prime periodic orbit reductions, and weighted orbit-counting asymptotics. For flows, a key technical device is the weighted dynamical \(L\)-function
\[
L(s,t)=\prod_{\gamma\in\mathcal P^*}
\left(1-e^{\ell_\psi(\gamma)-s\ell(\gamma)+it\ell_\varphi(\gamma)}\right)^{-1},
\]
whose singularity at
\[
s(t)=P(\psi+it\varphi)
\]
encodes the mean and variance. Contour integration and convergence of characteristic functions then yield the weighted periodic-orbit CLT [2304.01372].

## 5. Livšic–Sinai rigidity for Anosov endomorphisms

A different Livshitz-type direction concerns Jacobian cocycles and invariant volume forms. The classical Livšic–Sinai theorem, as stated in the endomorphism paper, says that if \(f:M\to M\) is a transitive \(C^2\) Anosov diffeomorphism, then
\[
Jf^n(p)=1 \quad \text{for all } p\in M \text{ with } f^n(p)=p
\]
if and only if \(f\) admits an invariant measure absolutely continuous with respect to the Riemannian volume. The same paper also cites a more recent result of Micena: the same periodic Jacobian condition already implies transitivity and the existence of an invariant \(C^1\) volume form [2606.15542].

The non-invertible analogue replaces \(1\) by the degree. If \(f:M\to M\) is a \(C^2\) Anosov endomorphism with \(\Omega(f)\) completely invariant and
\[
Jf^n(p)=\deg(f)^n=d^n
\quad\text{for all periodic }p,
\]
then Theorem A states that \(f\) is transitive and preserves a unique \(C^1\) normalized volume form. The complete invariance hypothesis is
\[
f(\Omega(f))=\Omega(f)=f^{-1}(\Omega(f)),
\]
and the paper notes that \(f(\Omega)=\Omega\) always holds for Anosov endomorphisms, while \(f^{-1}(\Omega)=\Omega\) is not known in general [2606.15542].

The correct cocycle in the endomorphism setting is
\[
\psi(x)=\log\!\left(\frac{Jf(x)}{d}\right).
\]
The periodic condition becomes
\[
\sum_{i=0}^{n-1}\psi(f^i(p))
=\log\frac{Jf^n(p)}{d^n}=0,
\]
so the degree-normalized Jacobian is precisely the quantity whose periodic sums vanish. This normalization is essential because a local diffeomorphism of degree \(d\) carries an intrinsic \(d^n\) multiplicity contribution to \(n\)-step volume transport [2606.15542].

The cohomological engine is Theorem B: if \(f:M\to M\) is a transitive Anosov endomorphism and \(\psi:M\to\mathbb R\) is \(C^1\) with vanishing periodic sums, then there exists a \(C^1\) function \(\varphi\) such that
\[
\psi=\varphi\circ f-\varphi,
\]
and \(\varphi\) is unique up to an additive constant. The proof passes to the universal cover, where a lift of an Anosov endomorphism is an Anosov diffeomorphism, establishes \(C^1\) regularity of the transfer function along stable and unstable leaves, and then invokes Katok–Hasselblatt’s regularity lemma to obtain global \(C^1\) regularity [2606.15542].

The proof of Theorem A combines this cohomological step with SRB and inverse SRB theory. It constructs an inverse SRB measure on a repeller basic set, uses the periodic Jacobian condition and the Anosov Closing Lemma to show
\[
\Lambda_f^u(x)+\Lambda_f^s(x)=\log d
\]
for recurrent Lyapunov-regular points, deduces that the inverse SRB measure is also an SRB measure, and then derives nonempty interior for the repeller basic set. Complete invariance and connectedness then force the repeller to be all of \(M\), yielding transitivity. Finally, solving the cohomological equation for \(\log(Jf/d)\) produces a \(C^1\) invariant density \(e^{-\varphi(x)}\,dm\), hence a \(C^1\) invariant volume form [2606.15542].

The paper also establishes Theorem C: if \(\Lambda_f^u\) and \(\Lambda_f^s\) are constant on \(Per(f)\), then
\[
\Lambda_f^u+\Lambda_f^s=\log d
\]
and \(f\) is transitive. It further stresses that the converse direction fails in the non-invertible setting: preserving volume does not force the periodic Jacobian condition. The example is a transitive Anosov endomorphism \(f=h\times A:T^3\to T^3\) with \(\deg(f)=2\), preserving a product volume, but having a fixed point \(0\) with \(Jf(0)\neq 2\) [2606.15542].

## 6. Nonlinear cocycles with values in analytic germs

The paper by Navas and Ponce treats a genuinely nonlinear target group. The base dynamics is a topologically transitive homeomorphism \(T:X\to X\) of a compact metric space satisfying the closing property. The cocycle takes values in \(\mathcal G\!erm_d\), the group of germs at the origin of analytic diffeomorphisms of \(\mathbb C^d\), represented by convergent vector-valued power series with invertible linear part. The main theorem states that if a Hölder-continuous cocycle \(F:X\to\mathcal G\!erm_d\) satisfies the periodic orbit obstruction, then there exists a Hölder-continuous map \(H:X\to\mathcal G\!erm_d\) such that
\[
F(x)=H(Tx)\circ H(x)^{-1},
\]
equivalently
\[
F(x)\circ H(x)=H(Tx)
\]
[1110.1911].

The proof has three layers. First, the derivative cocycle \(A_1(x)\in GL(d,\mathbb C)\) inherits the periodic orbit obstruction. By Kalinin’s theorem, there exists a Hölder map \(H_1:X\to GL(d,\mathbb C)\) with
\[
A_1(x)=H_1(Tx)H_1(x)^{-1},
\]
so the cocycle can be reduced to the tangent-to-identity case [1110.1911].

Second, one writes
\[
F(x)(Z)=Z+\Big(\sum_{|\mathbf j|>1} a^i_{\mathbf j}(x)Z^{\mathbf j}\Big)_{1\le i\le d}
\]
and seeks
\[
H(x)(Z)=Z+\Big(\sum_{|\mathbf j|>1} h^i_{\mathbf j}(x)Z^{\mathbf j}\Big)_{1\le i\le d}.
\]
Using a multivariate Faa di Bruno formula, the cohomological equation yields recursive scalar equations
\[
h^i_{\mathbf j_*}(Tx)-h^i_{\mathbf j_*}(x)
=
\sum_{1<|\mathbf j|,\ \mathbf j\le \mathbf j_*}
a^i_{\mathbf j}(x)\,P(\mathbf j,\mathbf j_*)\{H\}(x).
\]
The right-hand side depends only on lower-order coefficients, so one can solve inductively in total degree. The crucial point is that the periodic obstruction propagates to each coefficient equation; Lemma 10 proves that for every periodic point \(p\),
\[
\sum_{v=0}^{n-1}R^i_{\mathbf j_*}(T^v p)=0.
\]
Classical scalar Livšic theory then gives Hölder solutions \(h^i_{\mathbf j_*}\) coefficient by coefficient [1110.1911].

Third, a majorant series argument converts the formal solution into an analytic one. The paper introduces a model inverse germ \(G_S\) whose coefficients satisfy a recurrence with positive coefficients and proves an estimate
\[
[h^i_{\mathbf j}]_\alpha \le g^i_{S,\mathbf j}.
\]
Since the majorant coefficients grow at most exponentially, the formal series defining \(H(x)\) converges on a uniform neighborhood of the origin. The result is therefore a genuine Livšic theorem for a nonlinear, noncommutative target group [1110.1911].

The scope is deliberately narrow. The proof uses analyticity, the closing property, and Hölder regularity in an essential way. The paper explicitly does not prove the corresponding statement for arbitrary diffeomorphism groups, non-analytic smooth germ groups, weaker non-hyperbolic base systems, or cocycles with lower regularity than Hölder in the base variable [1110.1911].

## 7. Comparative structure, scope, and misconceptions

The literature represented here uses the spellings Livšic, Livshits, and Livsic. Across these variants, the invariant theme is that periodic orbit data determine a cohomological or rigidity conclusion, but the precise hypothesis and target category vary substantially [2304.01372] [2606.15542] [1110.1911].

| Variant | Periodic hypothesis | Conclusion |
|---|---|---|
| Classical additive Anosov case | \(\ell_\varphi(\gamma)=0\) for every closed orbit | \(\varphi\) is a coboundary |
| Positive-proportion theorem | Zero periods on a positive asymptotic upper density subset | \(\varphi\) is a coboundary |
| Strengthened nonpositive theorem | Positive-period set has subexponential growth | \(\varphi\) is cohomologous to a nonpositive function |
| Endomorphism Livšic–Sinai | \(Jf^n(p)=\deg(f)^n\) on all periodic points | Transitivity and a unique \(C^1\) normalized volume form |
| Analytic germ cocycles | Periodic orbit obstruction in \(\mathcal G\!erm_d\) | The cocycle is a Hölder coboundary |

Several misconceptions are corrected by these results. One is that a Livšic-type theorem must require checking every periodic orbit; the positive-proportion theorem shows that this is false for vanishing of Hölder periods on transitive Anosov systems. Another is that any asymptotically negligible exceptional set should suffice for nonpositive rigidity; the shift example shows that zero weighted density is too weak, and subexponential growth is the relevant condition in the theorem actually proved. A third is that invariant volume and periodic Jacobian normalization are equivalent in the non-invertible hyperbolic setting; the endomorphism example \(h\times A:T^3\to T^3\) shows that volume preservation does not force \(Jf^n(p)=\deg(f)^n\) on periodic points. Finally, the analytic-germ theorem should not be read as a theorem for arbitrary smooth diffeomorphism-valued cocycles; its proof depends on the triangular power-series structure and the majorant-series convergence mechanism [2304.01372] [2606.15542] [1110.1911].

Taken together, these theorems show that the periodic-orbit obstruction principle has a wider range than the classical scalar theorem might suggest. It supports weighted positive-density rigidity, sparse-exception nonpositive rigidity, Jacobian-based volume rigidity for non-invertible hyperbolic maps, and nonlinear cohomology for analytic-germ-valued cocycles. The common content is not a single theorem but a family of periodic-data rigidity principles whose exact form depends on the dynamical category and the target of the cocycle [2304.01372] [2606.15542] [1110.1911].

Source: https://www.emergentmind.com/topics/livshitz-type-theorem