Subsystem Loschmidt Echo

• Subsystem Loschmidt Echo is a set of diagnostics that use local or partial overlaps to measure return probabilities and probe reversibility in quantum systems.
• It encompasses various constructions—such as finite-block, cut-and-glue, local spin, and purified-copy echoes—that reveal distinct aspects of nonequilibrium dynamics.
• These methods provide clearer universal signatures during local quenches and serve as effective probes for decoherence, dynamical phase transitions, and information scrambling.

Subsystem Loschmidt echo denotes a family of return-probability diagnostics in which the full many-body Loschmidt echo is replaced by a local, reduced, or partial object. In the literature considered here, the term is not attached to a single universally fixed observable: it can refer to the return probability of a finite contiguous block, the overlap with an initial cut product state after a local quench, the recovery of a local polarization under imperfect time reversal, or the echo generated by acting on only one factor of a purified doubled state. Across these variants, the common objective is to retain the reversibility and sensitivity content of the Loschmidt echo while avoiding, or reinterpreting, the exponentially small full-system overlap (Karch et al., 28 Jan 2025, Stéphan et al., 2011, Zangara et al., 2011, Chenu et al., 2017).

1. Definitions and conceptual variants

The standard Loschmidt echo is the return probability

$\mathcal{L}(t)=|\langle \psi_0|\psi(t)\rangle|^2.$

Subsystem versions replace the full-state overlap by an observable tied to a restricted set of degrees of freedom, or by an overlap in an enlarged construction where only one factor is acted upon (Karch et al., 28 Jan 2025).

Construction	Defining idea	Representative source
Finite-block SLE	Probability that a contiguous block of length $N$ reproduces its initial occupations	(Karch et al., 28 Jan 2025)
Cut-and-glue LLE	Overlap with the initial product state $|A\rangle\otimes|B\rangle$ after gluing two subsystems	(Stéphan et al., 2011)
Local spin echo	Recovery of a local polarization after an imperfect forward-backward protocol	(Zangara et al., 2015, Zangara et al., 2011)
Purified-copy echo	Loschmidt amplitude of a purified mixed state when only one copy is driven	(Chenu et al., 2017)
Effective complement echo	Echo on $S_B$ induced after averaging and tracing over a small subsystem $S_A$	(Yan et al., 2019)

A central conceptual point is that these objects are not equivalent. Some are local observables, some are overlaps with special initial states, and some are amplitudes in doubled Hilbert spaces. A common misconception is to identify “subsystem Loschmidt echo” with reduced-density-matrix fidelity in all contexts; the literature summarized here does not support such a uniform identification. Instead, “subsystem” can denote a finite spatial block, a controlled subsystem in an open many-body setting, one member of a purified pair, or the complement of a traced-out region.

2. Local quenches and logarithmic subsystem echoes in critical one-dimensional systems

A canonical usage appears in the study of local “cut-and-glue” quenches in critical one-dimensional systems. The initial state is

$$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$

the ground state of $H_A+H_B$; at $t=0$, the coupling across the cut is switched on,

$H_A+H_B \longrightarrow H=H_A+H_B+H_{AB}^{\rm I}.$

The Loschmidt echo is

$$\mathcal{L}(t)=\left|\langle A\otimes B|e^{-iHt}|A\otimes B\rangle\right|^2,$$

and the quantity emphasized is the logarithmic Loschmidt echo,

$N$0

which is the time-dependent analog of the logarithmic bipartite fidelity (Stéphan et al., 2011).

In the CFT scaling limit, the short-time behavior is universal: $N$1 whereas the entanglement entropy grows as

$N$2

This difference in prefactors is one of the defining results of the local-quench formulation. For finite systems with a symmetric cut $N$3, the logarithmic Loschmidt echo follows the chord-length form

$N$4

The emitted quasiparticle-like excitations generate light-cone and bounce effects. The paper’s central comparison is that the entanglement entropy becomes non-universal once reflected excitations evolve inside the same subsystem, whereas the logarithmic Loschmidt echo remains much closer to the CFT prediction. In asymmetric geometries it develops a plateau structure that is nearly symmetric under $N$5. The same work also introduces a “detector” echo,

$N$6

specifically tailored to detect the excitations emitted just after the quench. In this setting, subsystem Loschmidt echo is therefore an overlap-based probe of local quench dynamics whose universal content is cleaner than that of the entanglement entropy in the post-bounce regime.

3. Local polarization echoes, many-body fidelity, and partial time reversal

In many-spin systems, subsystem Loschmidt echo often means a local recovery observable rather than a full overlap. For a high-temperature local-excitation state, the local echo $N$7 is the autocorrelation of the initially polarized spin after an imperfect time-reversal protocol. The corresponding global or many-body echo $N$8 is the average probability that the full many-spin configuration is recovered. Their relation is quantified by the extensive scaling law

$N$9

called the extensive decay hypothesis (Zangara et al., 2015).

The short-time expansions make the distinction explicit: $|A\rangle\otimes|B\rangle$0

$|A\rangle\otimes|B\rangle$1

$|A\rangle\otimes|B\rangle$2

with $|A\rangle\otimes|B\rangle$3. The hierarchy

$|A\rangle\otimes|B\rangle$4

and the empirical relation

$|A\rangle\otimes|B\rangle$5

motivate the paper’s “central hypothesis of irreversibility”: in the thermodynamic limit, the irreversibility scale is tied more closely to the complexity-driven many-body scale $|A\rangle\otimes|B\rangle$6 than to the bare perturbative scale $|A\rangle\otimes|B\rangle$7.

A closely related but distinct formulation treats a controlled spin chain $|A\rangle\otimes|B\rangle$8 coupled weakly to an uncontrolled spin-chain environment $|A\rangle\otimes|B\rangle$9. The time reversal is applied only to $S_B$0, not to the full composite $S_B$1. The local Loschmidt echo is the recovered site-1 polarization after total time $S_B$2,

$S_B$3

Its dynamics exhibits short-time quadratic decay, intermediate Fermi-golden-rule exponential decay, and a long-time saturation plateau. The decoherence rate decomposes into separate XY and Ising contributions,

$S_B$4

with

$S_B$5

Here “subsystem” refers operationally to a controlled component whose internal dynamics is reversed while the environment and coupling remain unreversed, so the echo directly quantifies decoherence (Zangara et al., 2011).

4. Purification, doubled Hilbert spaces, and information scrambling

A different notion of subsystem Loschmidt echo emerges from purification. For a driven isolated quantum system, the work characteristic function can be written as a Loschmidt amplitude of a purified post-measurement state in a doubled Hilbert space. If

$S_B$6

then the paper’s main universal statement is that, for any initial state and any driving protocol, the quantum work characteristic function is the Loschmidt amplitude of the purified state under a local echo on one copy, with

$S_B$7

When the initial state is canonical, the purification becomes the thermofield double state, and under sudden Hamiltonian negation one obtains

$S_B$8

The short-time decay is universally controlled by work fluctuations,

$S_B$9

In this framework, what may appear to be a reduced-state or subsystem echo is reinterpreted as the echo of a purified copy in an enlarged Hilbert space, and for thermal states it becomes a direct probe of scrambling in the thermofield double geometry (Chenu et al., 2017).

A complementary bridge to scrambling is provided by the relation between OTOCs and Loschmidt echoes. After averaging over local unitary operators on a small subsystem $S_A$0 and its complement $S_A$1, the OTOC becomes approximately a thermal average of Loschmidt-echo signals on $S_A$2,

$S_A$3

This is not a reduced-density-matrix fidelity, but it is a subsystem-resolved echo in the sense that tracing out $S_A$4 induces an effective stochastic perturbation on the complementary subsystem. The decay of the averaged OTOC is then interpreted as the decay of effective echo signals generated by scrambling (Yan et al., 2019).

5. Quasi-local string echoes, dynamical quantum phase transitions, and thermodynamic information

The most explicit recent formulation defines subsystem Loschmidt echo as a quasi-local string observable measurable with site-resolved snapshots. For a subsystem of length $S_A$5, the spatially averaged quantity is

$S_A$6

where $S_A$7 projects site $S_A$8 onto its initial-state occupation. The subsystem echo becomes the full Loschmidt echo when $S_A$9 (Karch et al., 28 Jan 2025).

In the short-time regime, this observable reveals dynamical quantum phase transitions after a quench from the charge-density-wave state $$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$0 in the 1D Bose-Hubbard model. The measured $$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$1 develops an increasingly sharp kink-like feature with critical time

$$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$2

and the associated rate function is

$$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$3

The crucial structural claim is that the kink originates in genuine higher-order correlations. The expansion begins with

$$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$4

$$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$5

while for larger $$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$6 the observable requires successively higher connected correlators; for $$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$7, the decomposition includes density, two-point, three-point, and four-point connected contributions. The observed nonanalyticity is therefore not captured by low-order local observables alone.

In the long-time regime, the time-averaged subsystem echo becomes an estimator of thermodynamic entropy and accessible Hilbert-space dimension: $$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$8 and at high temperature

$$|\psi(0)\rangle=|A\otimes B\rangle=|A\rangle\otimes|B\rangle,$$9

Once $H_A+H_B$0 exceeds the correlation length, the slope of $H_A+H_B$1 yields

$H_A+H_B$2

This was used to distinguish ergodic and fragmented dynamics. In the ergodic regime, the extracted values are around

$H_A+H_B$3

whereas in the fragmented regime they are approximately $H_A+H_B$4 for the CDW state and $H_A+H_B$5 for the domain wall. In this formulation, subsystem Loschmidt echo is simultaneously a dynamical probe of DQPTs, a correlator hierarchy detector, and a thermodynamic estimator.

6. Scope, terminology, and adjacent constructions

The modern literature makes clear that subsystem Loschmidt echo should not be conflated with every local or partial echo-related quantity. Several influential works explicitly do not formulate a subsystem echo. In strongly disordered non-interacting Fermi systems, the Loschmidt echo is the full many-body overlap

$H_A+H_B$6

and the paper states that it “does not formulate the problem as a reduced-subsystem Loschmidt echo” (Vanhala et al., 2022). Likewise, the transfer-matrix treatment of dynamical quantum phase transitions is built around the boundary partition function

$H_A+H_B$7

and it does not define a subsystem Loschmidt echo, even though its boundary-state and transfer-matrix structure could plausibly be adapted to such a purpose (Andraschko et al., 2013).

A second ambiguity concerns locality. Local perturbations do not automatically define subsystem echoes. In perturbed cat maps, restricting the perturbation to a phase-space window of width $H_A+H_B$8 leads to decay governed by the LDOS width, with $H_A+H_B$9 for sufficiently small windows and $t=0$0 for $t=0$1; this is highly relevant to local sensitivity, but it is not a reduced-state construction (0908.0702). Similarly, in Krylov-subspace error analysis, the approximation error is mapped to a Loschmidt echo on a virtual Lanczos tight-binding chain,

$t=0$2

where the chain is effectively cut after site $t=0$3; this is a partial-chain echo, but again not a subsystem observable in the experimental many-body sense (Ruffinelli et al., 2021).

The most robust general conclusion is therefore terminological. “Subsystem Loschmidt echo” is best understood as a class of constructions rather than a single formula. The literature distinguishes at least four non-equivalent meanings: a quasi-local block return probability, a local overlap following a local quench, a local observable under imperfect reversal, and a one-copy echo in a purified doubled Hilbert space. This plurality is not a defect; it reflects the fact that different communities use the Loschmidt paradigm to interrogate different kinds of partial information—spatial, operational, or purified—about nonequilibrium quantum dynamics.