Create and explore imaginative 3D worlds with our new product Dreamfield (opens in a new tab)
Papers
Topics
Authors
Recent
Search
2000 character limit reached

Subsystem Operator Rényi Entropy

Updated 5 July 2026
  • Subsystem operator Rényi entropy is a family of measures applied to subsystem-restricted operators or states, quantifying uncertainty, entanglement, and distinguishability.
  • It incorporates multiple formulations including operator-algebraic, reduced operator, and basis-dependent approaches to capture various quantum phenomena.
  • This framework bridges global phase-space analyses, local operator dynamics, and quantum reference frames, offering actionable insights into entanglement, transport, and thermalization.

Subsystem operator Rényi entropy denotes a family of Rényi-type quantities attached to a subsystem, an operator restricted to a subsystem, or a subsystem state created by an operator. In the literature considered here, the notion appears in several forms: an operator-algebraic entropy of states relative to a chosen reference system on a C∗C^*-algebra, a second Rényi entropy of a reduced Heisenberg operator, a basis-dependent Rényi entropy of subsystem measurement probabilities, and relative or Schatten-type Rényi quantities for reduced density matrices after local operator insertions (Mukhamedov et al., 2019, Roy et al., 25 Feb 2026, Tarighi et al., 2022, Zhao et al., 2024).

1. Conceptual scope

A recurrent structure is a bipartition into a subsystem AA and its complement, followed by a Rényi functional applied either to a reduced density matrix, to a reduced operator, or to a probability distribution induced by a chosen observable or basis. The basic density-matrix form is

Sα(ρA)=11−αlog⁡Tr⁡ρAα,S_\alpha(\rho_A)=\frac{1}{1-\alpha}\log\operatorname{Tr}\rho_A^\alpha,

while basis-dependent formulations replace ρA\rho_A by subsystem probabilities pIAp_{I_A}, and operator-growth formulations replace ρA\rho_A by a normalized reduced operator obtained from O(t)\mathcal O(t) (Kim et al., 2023, Tarighi et al., 2022, Roy et al., 25 Feb 2026).

Formulation Core object Representative source
Operator-algebraic SαS(φ)S_\alpha^{\mathcal S}(\varphi) for a state φ\varphi and reference system SS (Mukhamedov et al., 2019)
Reduced operator SOA(2)(t)S_{\mathcal O A}^{(2)}(t) from ρOA(t)\rho_{\mathcal O A}(t) (Roy et al., 25 Feb 2026)
Basis-dependent subsystem entropy Reα(L)Re_\alpha(L) from pIAp_{I_A} (Tarighi et al., 2022)
Observable-based subsystem entropy HαH_\alpha from subsystem-energy probabilities p(Ej)p(E_j) (Najafi et al., 2018)
Relative/distinguishability form Sk(ρA∥σA)S_k(\rho_A\Vert\sigma_A), DA(n)D_A^{(n)} (Zhao et al., 2024, Zhang et al., 2019)
Global aggregate of all subsystems SW(2)S_W^{(2)} from the Husimi function (Zhang et al., 19 Sep 2025)

A common source of confusion is the assumption that subsystem operator Rényi entropy is necessarily basis independent or necessarily a function of a reduced density matrix alone. The basis-dependent subsystem Rényi-Shannon entropy in a configuration basis is explicitly basis dependent, the operator-growth entropy of a Heisenberg operator is state independent, and the operator-algebraic construction depends on a chosen compact convex reference system S⊆GS\subseteq\mathcal G (Tarighi et al., 2022, Roy et al., 25 Feb 2026, Mukhamedov et al., 2019).

2. Operator-algebraic formulation and reference systems

The most general operator-algebraic construction in the sources is the C∗C^*-algebraic SS-mixing Rényi entropy. Let A\mathcal A be a C∗C^*-algebra, SS a compact convex set of states, and φ∈S\varphi\in S. For a countable extremal decomposition

φ=∑kλkφk,φk∈ex S,\varphi=\sum_k \lambda_k \varphi_k,\qquad \varphi_k\in \mathrm{ex}\,S,

the Rényi entropy relative to the reference system SS is

SαS(φ)=inf⁡μ∈Dφ(S)11−αlog⁡∑kλkα,α∈[0,∞)∖{1},S_\alpha^{\mathcal S}(\varphi)= \inf_{\mu\in\mathcal D_\varphi(S)}\frac{1}{1-\alpha}\log\sum_k \lambda_k^\alpha, \qquad \alpha\in[0,\infty)\setminus\{1\},

with value +∞+\infty if no countable extremal decomposition exists (Mukhamedov et al., 2019). This extends Ohya’s S\mathcal S-mixing entropy, is monotonically decreasing in α\alpha, and satisfies

lim⁡α→1SαS(φ)=SS(φ)\lim_{\alpha\to1}S_\alpha^{\mathcal S}(\varphi)=S^{\mathcal S}(\varphi)

(Mukhamedov et al., 2019).

This framework makes subsystem dependence explicit through the choice of reference system. A spatial subsystem can be modeled by a subalgebra B⊆A\mathcal B\subseteq\mathcal A and the restricted state φ∣B\varphi|_{\mathcal B}, leading to

SαG(B)(φ∣B).S_\alpha^{\mathcal G(\mathcal B)}(\varphi|_{\mathcal B}).

A dynamical or thermodynamic subsystem can instead be encoded by a distinguished convex subset of states, such as the invariant states I(Θ)I(\Theta) or KMS states Kβ(Θ)K_\beta(\Theta) of a C∗C^*-dynamical system (A,G,Θ(G))(\mathcal A,G,\Theta(G)) (Mukhamedov et al., 2019).

The same construction recovers familiar entropies in special cases. For a finite probability space with probabilities {pk}\{p_k\}, one obtains the classical Rényi entropy

SαS(φ)=11−αlog⁡∑kpkα.S_\alpha^{\mathcal S}(\varphi)=\frac{1}{1-\alpha}\log\sum_k p_k^\alpha.

For density operators ρ\rho on A=C(H)+CI\mathcal A=\mathcal C(H)+\mathbb CI, the state φρ(A)=Tr⁡(ρA)\varphi_\rho(A)=\operatorname{Tr}(\rho A) satisfies

SαG(φρ)=Sα(ρ)=11−αlog⁡Tr⁡(ρα),α>1,S_\alpha^{\mathcal G}(\varphi_\rho)=S_\alpha(\rho)=\frac{1}{1-\alpha}\log\operatorname{Tr}(\rho^\alpha),\qquad \alpha>1,

and for 0<α<10<\alpha<1,

SαG(φρ)≤Sα(ρ)S_\alpha^{\mathcal G}(\varphi_\rho)\le S_\alpha(\rho)

(Mukhamedov et al., 2019).

Reference-system inclusion generates entropy inequalities. For a KMS state φ∈Kβ(Θ)\varphi\in K_\beta(\Theta),

SαKβ(Θ)(φ)≤SαI(Θ)(φ),Sα(φ)≥SαKβ(Θ)(φ),S_\alpha^{K_\beta(\Theta)}(\varphi)\le S_\alpha^{I(\Theta)}(\varphi),\qquad S_\alpha(\varphi)\ge S_\alpha^{K_\beta(\Theta)}(\varphi),

and under GG-commutativity,

Sα(φ)≥SαI(Θ)(φ)≥SαKβ(Θ)(φ).S_\alpha(\varphi)\ge S_\alpha^{I(\Theta)}(\varphi)\ge S_\alpha^{K_\beta(\Theta)}(\varphi).

If the KMS state is unique, then

SαKβ(Θ)(φ)=0S_\alpha^{K_\beta(\Theta)}(\varphi)=0

(Mukhamedov et al., 2019). This shows that subsystem operator Rényi entropy can quantify uncertainty relative not only to a spatial restriction but also to an admissible decomposition structure.

3. Global phase-space formulations and subsystem-independent aggregation

A different construction appears in the Wehrl–Rényi-2 entropy of the Husimi function for NN distinguishable qubits. For a density matrix ρ^\hat\rho, the Husimi function is

PH(ρ^,n)=1(2π)N⟨n∣ρ^∣n⟩,P_H(\hat\rho,\mathbf n)=\frac{1}{(2\pi)^N}\langle \mathbf n|\hat\rho|\mathbf n\rangle,

and the Wehrl–Rényi entropy of order $2$ is

SW(2)(ρ^)=−ln⁡ ⁣[∫dn (PH(ρ^,n))2]S_W^{(2)}(\hat\rho)=-\ln\!\left[\int d\mathbf n\,\big(P_H(\hat\rho,\mathbf n)\big)^2\right]

(Zhang et al., 19 Sep 2025).

The central exact identity is

e−SW(2)(ρ^)=1(6π)N∑ATr⁡(ρ^A2),e^{-S_W^{(2)}(\hat\rho)} = \frac{1}{(6\pi)^N}\sum_A \operatorname{Tr}(\hat\rho_A^2),

where the sum runs over all 2N2^N subsystems A⊆{1,…,N}A\subseteq\{1,\dots,N\} and ρ^A\hat\rho_A is the reduced density matrix on AA (Zhang et al., 19 Sep 2025). The same quantity can therefore be read as a subsystem-independent scalar built from the entire system, or as a weighted aggregate of all subsystem Rényi-2 data via

Tr⁡(ρ^A2)=e−SA(2).\operatorname{Tr}(\hat\rho_A^2)=e^{-S_A^{(2)}}.

This exact relation is notable because it replaces the choice of a privileged subsystem by a sum over all subsystem purities. In the terminology of the source, SW(2)S_W^{(2)} is subsystem independent in its definition but encodes contributions from all possible subsystems (Zhang et al., 19 Sep 2025). The bounds

Nln⁡(3π)≤SW(2)(ρ^)≤Nln⁡(4π)N\ln(3\pi)\le S_W^{(2)}(\hat\rho)\le N\ln(4\pi)

follow from purity bounds, and explicit examples show that Haar-random states approach the upper bound, while GHZ and W states approach the lower bound in the large-NN limit (Zhang et al., 19 Sep 2025).

The derivation uses local Haar averaging, SWAP operators, and the identity

Tr⁡[(ρ^⊗ρ^)W^A]=Tr⁡(ρ^A2),\operatorname{Tr}\big[(\hat\rho\otimes\hat\rho)\hat W_A\big]=\operatorname{Tr}(\hat\rho_A^2),

so the same algebraic backbone appears in operator-entanglement and randomized-measurement settings (Zhang et al., 19 Sep 2025). This suggests a bridge between global phase-space Rényi entropies and aggregated subsystem operator Rényi data.

4. Reduced operators, operator growth, and Schwinger–Keldysh formalisms

In non-interacting fermionic systems, subsystem operator Rényi entropy is defined directly from a time-evolved operator rather than from an operator-state mapping. For a Hermitian operator O(t)=eiHtOe−iHt\mathcal O(t)=e^{i\mathcal H t}\mathcal O e^{-i\mathcal H t}, a bipartition into AA and BB, and normalization ZO=Tr⁡[O]Z_{\mathcal O}=\operatorname{Tr}[\mathcal O], the reduced operator is

ρOA(t)=Tr⁡B[O(t)]ZO,\rho_{\mathcal O A}(t)=\frac{\operatorname{Tr}_B[\mathcal O(t)]}{Z_{\mathcal O}},

and the second Rényi subsystem operator entropy is

e−SOA(2)(t)=Tr⁡A[ρOA2(t)]e^{-S_{\mathcal O A}^{(2)}(t)}=\operatorname{Tr}_A\big[\rho_{\mathcal O A}^2(t)\big]

(Roy et al., 25 Feb 2026). For the local density operator O=n^l=cl†cl\mathcal O=\hat n_l=c_l^\dagger c_l, the dynamical growth

ΔSOA(2)(t)=SOA(2)(t)−SOA(2)(0)\Delta S_{\mathcal O A}^{(2)}(t)=S_{\mathcal O A}^{(2)}(t)-S_{\mathcal O A}^{(2)}(0)

is bounded by

ΔSOA(2)(t)≤ln⁡2\Delta S_{\mathcal O A}^{(2)}(t)\le \ln 2

(Roy et al., 25 Feb 2026).

This formulation is explicitly state independent. The entropy depends on the Heisenberg operator, the spatial bipartition, and the Hamiltonian, rather than on an initial many-body state. The source emphasizes that it encodes both spatial and temporal information and therefore directly connects to transport for a local operator related to a conserved quantity (Roy et al., 25 Feb 2026).

The same work constructs a unified Schwinger–Keldysh field-theory formalism for SOA(2)(t)S_{\mathcal O A}^{(2)}(t) and for state Rényi and von Neumann entanglement entropies. For non-interacting systems, the resulting correlation-matrix formulas have the same structure, but the operator entropy is written in terms of infinite-temperature Keldysh Green’s functions, whereas state entanglement entropies are written in terms of vacuum Green’s functions (Roy et al., 25 Feb 2026). This produces explicit formulas for Aubry–André and Anderson models and shows that subsystem operator Rényi entropy can capture ballistic, diffusive, anomalous diffusive, and localized behavior through finite-size scaling of saturation times and through temporal growth profiles (Roy et al., 25 Feb 2026).

A related distinction is important: operator-growth entropies of this type are not relative entropies and are not defined by choosing a basis of subsystem measurement outcomes. They are reduced-operator Rényi entropies in real time.

5. Local operator quenches, relative Rényi entropy, and subsystem distances

In two-dimensional CFT, local operator quenches lead to subsystem Rényi quantities that measure either entanglement or distinguishability of operator-generated states. For two reduced density matrices ρA\rho_A and σA\sigma_A on an interval AA, the relative Rényi entropy used in rational and holographic CFTs is

Sk(ρA∥σA)=1k−1log⁡tr⁡ρAktr⁡(ρAσAk−1),S_k(\rho_A\Vert\sigma_A)=\frac{1}{k-1}\log\frac{\operatorname{tr}\rho_A^k}{\operatorname{tr}(\rho_A\sigma_A^{k-1})},

and is interpreted there as quantifying how distinguishable two local operator excitations are when restricted to AA (Zhao et al., 2024). In rational CFTs it is zero before the light cone reaches the entangling point and becomes a constant at late times for several operator families, with the late-time value controlled by finite-dimensional matrices of two-point coefficients. In holographic CFTs, the collision relative entropy reconstructs the entanglement wedge and induces a metric proportional to the Bures metric on the corresponding bulk region (Zhao et al., 2024).

A complementary set of quantities is given by subsystem trace and Schatten distances after local operator quenches. For a nonchiral primary field O\mathcal O, the Rényi entropy increase obeys

ΔSA(n)(t)={0,0<t<ℓ, t>2ℓ,log⁡dO,ℓ<t<2ℓ,\Delta S_A^{(n)}(t)= \begin{cases} 0, & 0<t<\ell,\ t>2\ell,\\ \log d_{\mathcal O}, & \ell<t<2\ell, \end{cases}

where dOd_{\mathcal O} is the quantum dimension (Zhang et al., 2019). The same analysis shows that the reduced density matrix of an interval hosting a quasiparticle is orthogonal to the reduced density matrix of the interval without quasiparticles, and that reduced density matrices hosting quasiparticles at different positions are also orthogonal to each other (Zhang et al., 2019). Consequently, the Schatten distances are piecewise constant and, in the orthogonal regimes, the trace distance reaches its maximal value in the source’s normalization (Zhang et al., 2019).

For the free non-compact boson and its harmonic-chain discretization, exact excited-state Rényi entropies and subsystem Schatten distances were obtained for several low-lying multi-particle states, together with short-interval expansions for general excited states (Zhang et al., 2020). In the CFT regime, the leading correction to FA,X(n)\mathcal F^{(n)}_{A,\mathcal X} is

FA,X(n)=1−π2(n2−1)ΔX3n x2+o(x2),x=ℓL,\mathcal F^{(n)}_{A,\mathcal X} =1-\frac{\pi^2(n^2-1)\Delta_{\mathcal X}}{3n}\,x^2+o(x^2), \qquad x=\frac{\ell}{L},

so the leading short-interval behavior depends only on the scaling dimension ΔX\Delta_{\mathcal X} (Zhang et al., 2020). In the extremely gapped limit of the harmonic chain, by contrast, the leading behavior depends only on the total number of excited quasiparticles RR, both for Rényi entropies and for subsystem Schatten distances (Zhang et al., 2020). This contrast isolates two regimes: conformal-energy control in the gapless theory and purely combinatorial quasiparticle counting in the deeply gapped theory.

6. Basis-dependent and observable-based many-body formulations

A distinct usage of subsystem operator Rényi entropy appears in basis-dependent Rényi-Shannon entropies. For a ground state ∣g⟩=∑IaI∣I⟩|g\rangle=\sum_I a_I|I\rangle written in a local product basis, the subsystem probabilities are

pIA=∑IAˉPIAIAˉ,PI=∣aI∣2,p_{I_A}=\sum_{I_{\bar A}} P_{I_A I_{\bar A}},\qquad P_I=|a_I|^2,

and the subsystem operator Rényi entropy is

Reα(L)=11−αln⁡∑IApIAαRe_\alpha(L)=\frac{1}{1-\alpha}\ln\sum_{I_A} p_{I_A}^\alpha

(Tarighi et al., 2022). At a critical point, this obeys

Reα(L)=aαL+xαln⁡L+O(1),Re_\alpha(L)=a_\alpha L + x_\alpha \ln L + \mathcal O(1),

with universal logarithmic coefficient xαx_\alpha (Tarighi et al., 2022). For critical quadratic fermions with U(1)U(1) symmetry,

xα={c8,α≤4,αα−1c8,α>4,x_\alpha= \begin{cases} \frac{c}{8}, & \alpha\le 4,\\[0.5ex] \frac{\alpha}{\alpha-1}\frac{c}{8}, & \alpha>4, \end{cases}

where cc is the central charge. Without U(1)U(1) symmetry,

xα={b(α)8,α≤1,αα−1c8,α>1.x_\alpha= \begin{cases} \frac{b(\alpha)}{8}, & \alpha\le 1,\\[0.5ex] \frac{\alpha}{\alpha-1}\frac{c}{8}, & \alpha>1. \end{cases}

These formulas exhibit non-analytic changes at α=4\alpha=4 or α=1\alpha=1, respectively (Tarighi et al., 2022).

An observable-based formulation is the Rényi entropy of subsystem energy. For a truncated subsystem Hamiltonian HlH_l with eigenstates ∣ej⟩|e_j\rangle, subsystem-energy probabilities are

pj=⟨ej∣ρl∣ej⟩,p_j=\langle e_j|\rho_l|e_j\rangle,

and

Hα=11−αln⁡∑jpjαH_\alpha=\frac{1}{1-\alpha}\ln\sum_j p_j^\alpha

(Najafi et al., 2018). This quantity obeys an area law in gapped phases and, at criticality,

Hα(l)=ϵ(α)ln⁡l+βαH_\alpha(l)=\epsilon(\alpha)\ln l+\beta_\alpha

with universal coefficient ϵ(α)\epsilon(\alpha) that scales with the central charge in the Ising and XX universality classes (Najafi et al., 2018). The same work shows that the largest subsystem-energy probabilities closely mimic the largest Schmidt coefficients and that truncated Shannon and truncated von Neumann entropies are almost indistinguishable (Najafi et al., 2018). Because the relevant observable is the truncated Hamiltonian itself, the source emphasizes that this entropy is associated with a natural observable and can be connected to a Loschmidt-echo protocol (Najafi et al., 2018).

Taken together, these formulations show that subsystem operator Rényi entropy need not refer to the spectrum of ρA\rho_A alone. It may instead quantify the Rényi complexity of subsystem measurement outcomes in a chosen basis or of a distinguished subsystem observable.

7. Typicality, thermal ensembles, and observer dependence

Typicality results provide useful baselines. For a Haar-random pure state on HA⊗HB\mathcal H_A\otimes\mathcal H_B with dimensions mm and nn, the average subsystem Rényi entropy

Sα(m,n)=E[Sα(ρA)]S_\alpha(m,n)=\mathbb E[S_\alpha(\rho_A)]

admits an exact solution for m=α=2m=\alpha=2 and an analytic approximation for general α\alpha (Kim et al., 2023). In the large-nn limit,

S~α(m,n)∼ln⁡m−α2n(m−m−1),\widetilde S_\alpha(m,n)\sim \ln m-\frac{\alpha}{2n}(m-m^{-1}),

which matches the asymptotic Page behavior at α→1\alpha\to1 (Kim et al., 2023). The source explicitly notes that these results can be repurposed for subsystem operator Rényi entropies whenever an operator is purified or vectorized into a bipartite pure state (Kim et al., 2023).

Thermal subsystem Rényi entropy in all-to-all systems was analyzed for SYK-like models. For a subsystem AA of MM modes inside NN total modes, the second Rényi entropy

SA(2)=11−2log⁡Tr⁡A(ρA2)\mathcal S_A^{(2)}=\frac{1}{1-2}\log\operatorname{Tr}_A(\rho_A^2)

is computed from a replicated large-NN path integral with subsystem-dependent bilocal fields (Zhang et al., 2020). For Majorana SYKq_q with q≥4q\ge4, the small-subsystem limit is maximally mixed,

SA(n)(λ→0)=Mlog⁡22,λ=MN,\mathcal S_A^{(n)}(\lambda\to0)=\frac{M\log2}{2},\qquad \lambda=\frac{M}{N},

and for M≤N/2M\le N/2 the results are well approximated by a thermal entropy with an effective temperature determined by energy matching (Zhang et al., 2020).

For the Hubbard model, a different replica path integral introduces the Rényi entanglement through a local kick term between replicas. Within inhomogeneous DMFT, the second Rényi entropy is extracted as

SA(2)=∫01dλ ⟨Skick⟩ZA(2)(λ),S_A^{(2)}=\int_0^1 d\lambda\, \langle \mathcal S_{\rm kick}\rangle_{Z_A^{(2)}(\lambda)},

and is computed for extended subsystems in one and two dimensions (Bera et al., 2023). In the correlated metallic phase, the subsystem-size scaling is described by the crossover formula interpolating between the volume-law thermal Rényi entropy and the universal boundary-law Rényi entanglement entropy with logarithmic violation (Bera et al., 2023). The same framework yields a Rényi mutual information that shows hysteresis across the first-order Mott transition and non-monotonic temperature dependence near the critical endpoint (Bera et al., 2023).

Observer dependence appears explicitly in quantum reference frames. There, one considers relational subsystems X⊆SRiˉX\subseteq SR_{\bar i}, dephases the reduced relational state ρX(Ri)\rho_X^{(R_i)} in a group-label basis, and defines the diagonal Rényi entropy

Sα(ΔρX(Ri)).S_\alpha(\Delta\rho_X^{(R_i)}).

For ideal frames, these are frame-independent diagonal Rényi invariants: Sα(ΔρX(Ri))=Sα(ΔρY(Rj))S_\alpha(\Delta\rho_X^{(R_i)})=S_\alpha(\Delta\rho_Y^{(R_j)}) for the corresponding subsystem YY in frame RjR_j (Hamette, 24 Mar 2026). The same source proves a coherence–entanglement tradeoff,

Sα(ΔρX(Ri))=Sα(ρX‾(Ri))+Cα(ρX(Ri)),S_\alpha(\Delta\rho_X^{(R_i)})=S_\alpha(\rho_{\overline X}^{(R_i)})+C_\alpha(\rho_X^{(R_i)}),

and for non-ideal frames derives a bound on observer-dependent entropy differences in terms of the effective relational Hilbert-space dimension (Hamette, 24 Mar 2026). This makes explicit that subsystem operator Rényi entropy can be basis dependent and observer dependent even when it is frame-invariant in a diagonalized form.

Across these formulations, the common core is a Rényi functional applied after restriction: to a subalgebra, to a reduced state, to a reduced operator, to a subsystem probability distribution, or to a dephased relational state. What changes from framework to framework is the meaning of the restriction, the role of the operator, and the physical interpretation of the resulting entropy—mixing, entanglement, distinguishability, transport, thermalization, or observer dependence.

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to Subsystem Operator Rényi Entropy.