Correlated Hopping in Lattice Systems

• Correlated hopping refers to an occupancy-dependent tunneling process where the effective hopping amplitude is modified by local particle configurations, as seen in extended Hubbard models.
• It underlies phenomena such as asymmetric charge transport, renormalized exchange interactions, and emergent topological behavior in both fermionic and bosonic systems.
• Experimental and theoretical methods like Floquet engineering, DMFT, and cavity QED enable precise exploration and control of correlated hopping in platforms such as ultracold atoms and quantum dots.

Searching arXiv for recent and foundational papers on correlated hopping across lattice, transport, and quantum-simulation contexts. Correlated hopping is an interaction-induced modification of kinetic motion in which a hopping matrix element depends explicitly on site or bond occupancies. In the literature it also appears as assisted hopping, density-dependent hopping, or charge-bond interaction. In second-quantized lattice models, the defining feature is that a term which would be a one-body tunneling process in the Hubbard or Bose–Hubbard limit is multiplied by occupation operators such as $n_i$, $n_j$, or $n_{i\bar\sigma}$, thereby coupling transport directly to the instantaneous many-body configuration. This structure occurs in fermionic and bosonic models, in real-space and synthetic lattices, and in mesoscopic transport setups, where it breaks particle-hole symmetry, renormalizes superexchange, induces clustering, and can even generate interaction-driven topology (Liberto et al., 2013, Matysiak et al., 2021, Eckern et al., 2024, Padhan et al., 1 Mar 2025).

1. Definition and canonical forms

Correlated hopping is introduced by replacing a bare amplitude $t$ with an occupation-dependent amplitude. A standard nearest-neighbour fermionic form is

$$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$

with $X$ the two-body correlated-hopping amplitude and $Y$ a three-body amplitude (Liberto et al., 2013). In the periodically modulated Feshbach realization, the effective kinetic term is written as

$$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$

with $Y=-2X$ to leading order (Liberto et al., 2013).

In quantum-dot transport, the same idea appears in tunneling to leads. The extended Anderson model uses the composite operator

$D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$

so that tunneling onto or off the dot has amplitude $n_j$0 when the opposite spin is absent and $n_j$1 when it is present (Eckern et al., 2024, Eckern et al., 2021). Equivalently, the correlated-hopping term is

$n_j$2

This form makes explicit that a nominally one-electron transfer is conditioned by local occupancy (Eckern et al., 2024).

In bosonic systems, correlated hopping is often extended or bridge-mediated. In the one-dimensional zig-zag Bose–Hubbard geometry,

$n_j$3

so that hopping from $n_j$4 to $n_j$5 depends on the occupation of the intervening site $n_j$6 (Stasińska et al., 2020). In the bosonic scar model, density-assisted hopping appears as terms such as

$n_j$7

which provide a “soft” kinetic constraint (Hudomal et al., 2019).

Multi-orbital versions generalize the dependence to orbital occupations. In molecular nanomagnets the hopping sector is written as

$n_j$8

where $n_j$9 is the correlated-hopping parameter (Matysiak et al., 2021). In such formulations, the modulation is not a phenomenological correction to $n_{i\bar\sigma}$0 but part of the microscopic Hamiltonian.

2. Microscopic origin and formal derivations

The common microscopic origin is nonlocal Coulomb structure in a localized basis. In the Anderson-dot language, the dimensionless parameter $n_{i\bar\sigma}$1 is related to the Coulomb matrix element $n_{i\bar\sigma}$2 via $n_{i\bar\sigma}$3 (Eckern et al., 2024). In semiconductor quantum dots or molecular junctions, the data state that $n_{i\bar\sigma}$4 is often tens of meV, $n_{i\bar\sigma}$5 can be a few meV, and since the bare hybridization $n_{i\bar\sigma}$6 is also of order a meV, one expects $n_{i\bar\sigma}$7 to be of order $n_{i\bar\sigma}$8 (Eckern et al., 2024).

In the extended Falicov–Kimball model, correlated hopping follows from nonlocal Coulomb matrix elements $n_{i\bar\sigma}$9, yielding

$t$0

which modifies $t$1-electron motion according to the $t$2-occupation pattern (Farkašovský et al., 2015). In generalized Hubbard models with orbital degeneracy, the non-diagonal Coulomb matrix elements $t$3 with $t$4 produce occupation-dependent hoppings $t$5 and $t$6, breaking the usual electron-hole symmetry (Skorenkyy et al., 2021).

Floquet engineering provides a controlled derivation. Starting from

$t$7

the high-frequency effective Hamiltonian becomes

$t$8

with $t$9. Expanding the Bessel function gives the correlated-hopping structure and defines $$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$0 (Liberto et al., 2013). This realization is notable because it directly links a drive parameter to an occupation-dependent kinetic coefficient.

Cavity QED yields a different derivation. In a synthetic ladder of ground-state Zeeman levels, far-detuned cavity-mediated interactions and two weak drives produce an atom-only effective Hamiltonian whose leading nontrivial process is

$$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$1

with $$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$2. In Schwinger-boson form this gives

$$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$3

The mechanism relies on collective cavity-mediated dressing of excited levels and interference between two drives to suppress undesired single-particle shifts while retaining correlated hopping (Chu et al., 2022).

3. Effective-Hamiltonian consequences

A central effect of correlated hopping is the renormalization of low-energy exchange. In the multi-orbital Hubbard description of $$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$4, second-order Schrieffer–Wolff projection yields

$$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$5

with $$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$6 and

$$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$7

Correlated hopping reduces only the triplet–triplet virtual processes $$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$8 by $$H = -\sum_{\langle ij\rangle,\sigma} [\,t + X\,(n_{i,\bar\sigma} + n_{j,\bar\sigma}) + Y\,n_{i,\bar\sigma}n_{j,\bar\sigma}\,]\,c^\dagger_{i\sigma}c_{j\sigma} + U \sum_i n_{i\uparrow}n_{i\downarrow} + V \sum_{\langle ij\rangle} n_i n_j + \dots,$$9, leaving $X$0 unchanged (Matysiak et al., 2021). For $X$1, using $X$2 eV, $X$3 eV, and corrected hopping $X$4 eV, the reported values are $X$5 meV and $X$6 meV at $X$7, while fitting the experimental $X$8 meV gives $X$9 from perturbation theory and $Y$0 from exact diagonalization (Matysiak et al., 2021). The concrete implication is that modest correlated hopping can materially reduce antiferromagnetic exchange.

In mean-field treatments of the extended Falicov–Kimball model, the simplest Hartree–Fock decoupling makes correlated hopping a bandwidth-renormalization channel: $Y$1 This feedback alters both the effective $Y$2-band parity and bandwidth and thereby changes the stability of ferroelectric, antiferroelectric, and charge-density-wave phases (Farkašovský et al., 2015). Positive $Y$3 enlarges the excitonic ferroelectric region at negative $Y$4, whereas sufficiently negative $Y$5 suppresses it and can stabilize an antiferroelectric phase with $Y$6, $Y$7 (Farkašovský et al., 2015).

In narrow-band flat-band settings, three distinct correlated-hopping amplitudes $Y$8 govern different physical processes: double–single, single–empty, and double–empty or opposite-spin single–single motion. The effective hopping at uniform filling $Y$9 is

$$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$0

so the flat-band condition is $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$1, reducing at half filling to

$$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$2

These amplitudes have qualitatively different roles in the formation of half-metallic ferromagnetic states, with $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$3 favoring ferromagnetism via band narrowing and $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$4 favoring antiferromagnetic tendencies via kinetic superexchange $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$5 (Westerhout et al., 2022).

Correlated hopping can also generate exact symmetries. In the one-dimensional Floquet-engineered fermionic model, the Hamiltonian retains spin-SU(2) symmetry for all $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$6, and at the integrable point $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$7 it acquires charge-SU(2) symmetry generated by $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$8 and $$H_J = -t \sum_{\langle ij\rangle,\sigma} (1 - X[n_{i,\bar\sigma}+n_{j,\bar\sigma} -2n_{i,\bar\sigma}n_{j,\bar\sigma}])\,c^\dagger_{i\sigma}c_{j\sigma},$$9, with $Y=-2X$0 and $Y=-2X$1 (Liberto et al., 2013). This opens a direct route to $Y=-2X$2-pairing physics in a correlated-hopping setting.

4. Phases and dynamical phenomena induced by correlated hopping

In bosonic lattice systems, correlated hopping reshapes the phase diagram beyond the standard Mott-insulator/superfluid dichotomy. In the one-dimensional extended correlated-hopping Bose–Hubbard model, the effective bridge-dependent amplitude is

$Y=-2X$3

so that the presence of a particle on the bridge site can suppress or enhance tunneling (Stasińska et al., 2020). The reported zero-temperature phases include Mott insulator, conventional superfluid, paired-hole superfluid, paired-particle superfluid, droplet superfluid, and phase-separated superfluid (Stasińska et al., 2020). At $Y=-2X$4, a boson is completely blocked from hopping over an occupied site, and correlated hopping induces effective attraction among particles despite $Y=-2X$5 (Stasińska et al., 2020). A closely related two-species Bose–Hubbard study found a rich phase diagram including a pair superfluid and a superfluid quantum droplet phase, with the former arising from the interplay between single-particle and correlated hopping and the latter from large correlated hopping (Stasińska et al., 2019).

Correlated hopping can stabilize hidden nonlocal order. In the one-dimensional hole-superconductor model with next-nearest-neighbour correlated hopping,

$Y=-2X$6

the integrable point $Y=-2X$7 yields exact and asymmetric spin-charge separation, a finite spin gap, gapless charge excitations, and a non-vanishing den Nijs–Rommelse type string correlator (Chhajlany et al., 2015). The spin gap is nonzero for all $Y=-2X$8, and numerical studies away from the integrable point show persistence of both long-range string order and spin gap for all $Y=-2X$9 at $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$0 (Chhajlany et al., 2015).

In constrained bosonic dynamics, density-assisted hopping can produce many-body scars. The bosonic model $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$1 exhibits a bipartite Fock-space structure with an exponentially large number of exact zero-energy eigenstates at filling $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$2, weakly entangled scarred eigenstates at high energy densities, and periodic revivals after quenches from special product states such as $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$3 (Hudomal et al., 2019). The data report a dominant revival period $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$4 for $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$5, while a more constrained related model $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$6 gives exact revivals with $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$7 and period $D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$8 (Hudomal et al., 2019). These results establish that a “soft” density-dependent constraint is sufficient to generate anomalously slow thermalization.

Correlated hopping also supports topological mechanisms with no single-particle analogue. In the two-component fermionic mixture

$D_\sigma \equiv d_\sigma(1-x\,n_{\bar\sigma}),$9

the $n_j$00 species experiences a staggered potential $n_j$01, and for $n_j$02 forms a density wave on the even sublattice. Through $n_j$03, the $n_j$04 species then sees an effective dimerization equivalent to an SSH model; for $n_j$05, the dimerization is trivial (Padhan et al., 1 Mar 2025). The topological transition occurs at $n_j$06, and MPS simulations show that $n_j$07 closes while $n_j$08 remains finite, with edge-state polarization, string order, and entanglement-spectrum degeneracy characterizing the topological regime (Padhan et al., 1 Mar 2025). In a Thouless pump protocol, the pumped charge of the $n_j$09 species is quantized, with $n_j$10 when the loop encloses the topological point and $n_j$11 otherwise (Padhan et al., 1 Mar 2025). This suggests an interaction-driven route to topological order mediated entirely by density-dependent kinetics.

5. Transport, thermoelectricity, and localization

In quantum dots, correlated hopping modifies both charge and heat transport and is especially visible in the non-linear regime. In the extended Anderson model, the term $n_j$12 breaks the particle-hole symmetry of the standard $n_j$13 model, so conductance $n_j$14 is no longer symmetric in $n_j$15 (Eckern et al., 2024). The lower Hubbard peak moves to more negative $n_j$16, while no compensating shift occurs at $n_j$17 (Eckern et al., 2024). Differential conductance curves at $n_j$18 and $n_j$19 show that the minimum between the two Hubbard peaks is shifted to $n_j$20, the left peak splits more strongly than the right, and the upper band is almost unchanged (Eckern et al., 2024). Differential thermopower similarly loses exact antisymmetry, develops higher, narrower extrema at negative $n_j$21, and can exhibit sign changes for moderate $n_j$22 in a range of $n_j$23 (Eckern et al., 2024). The proposed experimental signature is a systematic shift of the lower-energy peak to $n_j$24 without a corresponding shift of the upper-energy peak (Eckern et al., 2024).

An exact transport symmetry can nevertheless survive. For the transport Green function

$n_j$25

the equation-of-motion treatment shows that the $n_j$26-dependence enters through $n_j$27 or $n_j$28, producing an exact $n_j$29 symmetry of transport observables (Eckern et al., 2021). The average dot occupation also obeys this symmetry, whereas the spectral function calculated within an analogous decoupling scheme does not, because some lifetime-cutoff processes appear only at fourth order in the hybridization (Eckern et al., 2021). This distinction is a formal subtlety rather than a contradiction: transport and spectroscopy are controlled by different Green functions in the assisted-hopping Anderson model.

In the Falicov–Kimball model, dynamical mean-field theory shows that correlated hopping can strongly enhance thermoelectric response by reshaping the two-particle transport function more dramatically than the one-particle DOS. On the Bethe lattice, the transport coefficients follow from

$n_j$30

with $n_j$31, $n_j$32, and $n_j$33 (Dobushovskyi et al., 2016). Correlated hopping generates sharp resonant features in $n_j$34 when a two-particle resonance condition is met, while the one-particle DOS remains comparatively smooth (Dobushovskyi et al., 2016). Bringing the Fermi level near the resonant frequency strongly increases electrical conductivity, thermal conductivity, and thermoelectric power (Dobushovskyi et al., 2016). In a related DMFT study on the hypercubic lattice, the most pronounced thermoelectric effects occur for $n_j$35, and for $n_j$36, $n_j$37, half filling of $n_j$38-electrons, and $n_j$39, the Seebeck coefficient reaches $n_j$40 at $n_j$41, with power factor enhanced by more than an order of magnitude and $n_j$42 at $n_j$43 (Shvaika, 2014). Doping to $n_j$44, $n_j$45 yields $n_j$46 and $n_j$47 (Shvaika, 2014).

When hopping between occupied sites is strongly reduced, correlated hopping can induce localization and an in-gap band. In the microdoped Mott-insulator phase of the correlated-hopping Falicov–Kimball model, taking $n_j$48 with $n_j$49 makes $n_j$50, so hopping is completely blocked on $n_j$51-occupied sites (Dobushovskyi et al., 2020). For $n_j$52, this produces a third narrow band inside the Mott gap whose spectral weight is $n_j$53 (Dobushovskyi et al., 2020). In the limit $n_j$54, the band shrinks to a $n_j$55-peak carrying no transport weight; for slightly positive $n_j$56, it broadens into a narrow resonant mid-gap band, giving a giant low-temperature Seebeck response whose sign can be switched by microdoping (Dobushovskyi et al., 2020).

Localization phenomena also arise in quasiperiodic systems with correlated off-diagonal structure. In a one-dimensional chain with AAH on-site modulation and Fibonacci or Bronze-Mean hopping sequences, the Hamiltonian

$n_j$57

uses $n_j$58 arranged quasiperiodically (Karmakar et al., 3 Jun 2025). The reported diagnostics—IPR, NPR, and fractal dimension—show a reentrant localization-delocalization-localization sequence: a fully localized regime beyond $n_j$59, followed by reentrant delocalization windows $n_j$60 for Fibonacci and $n_j$61 for Bronze Mean, before a final localized phase at large $n_j$62 (Karmakar et al., 3 Jun 2025). The effect is attributed to the interplay of diagonal quasiperiodicity and correlated binary hopping with long-range order (Karmakar et al., 3 Jun 2025).

6. Methods, realizations, and recurring misconceptions

Correlated-hopping models are studied by a diverse set of methods, and the method is usually tightly matched to the physical regime. Strong-coupling and Schrieffer–Wolff projections are used when $n_j$63, as in the multi-orbital nanomagnet problem (Matysiak et al., 2021). Cluster mean-field theory, exact diagonalization, and analytical strong-coupling expansion are combined for bosonic lattice models with extended correlated hopping (Stasińska et al., 2019, Stasińska et al., 2020). Matrix product state simulations diagnose equilibrium topology and quantized pumping in one-dimensional mixtures (Padhan et al., 1 Mar 2025). DMFT gives exact charge and heat transport in infinite-dimensional Falicov–Kimball models with correlated hopping (Shvaika, 2014, Dobushovskyi et al., 2016, Dobushovskyi et al., 2020). Equation-of-motion Green-function methods address linear and nonlinear transport in Anderson-type quantum dots (Eckern et al., 2021, Eckern et al., 2024).

Experimental implementations span several platforms. Ultracold fermions in optical lattices can realize correlated-hopping models by a modulated magnetic field near a Feshbach resonance; for $n_j$64 in a $n_j$65 nm lattice, the data give $n_j$66 Hz, $n_j$67 kHz, and $n_j$68 Hz, allowing $n_j$69 to be tuned from the Hubbard limit to the $n_j$70-symmetric point $n_j$71 (Liberto et al., 2013). In cavity QED, alkaline-earth atoms such as $n_j$72 and $n_j$73 with $n_j$74 can reach $n_j$75 kHz while $n_j$76 kHz, with typical working parameters $n_j$77, $n_j$78, and $n_j$79 (Chu et al., 2022). In atomic mixtures, density-dependent tunneling can be engineered via resonant Floquet modulation of inter-species interactions or Feshbach-induced occupation-dependent lattice shifts (Padhan et al., 1 Mar 2025). In mesoscopic transport, detection protocols emphasize bias- and gate-dependent asymmetries of differential conductance and thermopower (Eckern et al., 2024).

Several recurrent misconceptions are corrected by the literature. First, correlated hopping is not merely a small perturbative correction to $n_j$80. In $n_j$81, $n_j$82 suffices to reduce the calculated antiferromagnetic exchange from $n_j$83 meV to the experimental $n_j$84 meV (Matysiak et al., 2021). In Falicov–Kimball models, even modest $n_j$85 can change the preferred ordered state from checkerboard to stripes or segregation and strongly alter finite-temperature critical behaviour (Cencarikova et al., 2010). Second, the main signatures are not limited to spectral functions. Several studies emphasize that two-particle transport functions or nonlinear response reveal stronger and cleaner correlated-hopping effects than one-particle DOS alone (Dobushovskyi et al., 2016, Eckern et al., 2024). Third, correlated hopping need not only renormalize pre-existing band topology or magnetism. The one-dimensional mixture with staggered $n_j$86 and density-dependent $n_j$87 shows that topological order can be induced through specific interaction couplings, without inheriting from a corresponding single-particle Hamiltonian (Padhan et al., 1 Mar 2025).

Taken together, these results define correlated hopping as a kinetic interaction channel of broad scope rather than a model-specific embellishment. Across cold atoms, nanomagnets, quantum dots, Falicov–Kimball systems, and quasiperiodic lattices, the same formal ingredient—occupation-dependent tunneling—recurrently controls exchange renormalization, symmetry breaking, clustering, hidden order, anomalous transport, localization, and interaction-driven topology (Liberto et al., 2013, Matysiak et al., 2021, Chu et al., 2022, Padhan et al., 1 Mar 2025).