---
title: Flux-Controlled Anomalous Floquet Walk
url: https://www.emergentmind.com/topics/flux-controlled-anomalous-floquet-quantum-walk
type: topic
---

# Flux-Controlled Anomalous Floquet Walk

Searching arXiv for the core paper and closely related work on anomalous Floquet quantum walks and flux control.
arXiv search: "flux-controlled anomalous Floquet quantum walk anomalous Floquet quantum walk magnetic quantum walks"
Flux-controlled anomalous Floquet quantum walks are periodically driven quantum-walk systems in which a controllable flux variable—implemented as a Peierls phase, a drive-phase-controlled effective hopping sign, or a synthetic gauge field—enters the one-period unitary evolution and organizes the topology of the quasienergy gaps at \(0\) and \(\pi/T\). Their defining feature is anomalous Floquet topology: protected boundary transport is determined by gap invariants of the full time-periodic evolution rather than by ordinary static band invariants alone, so edge or corner states can persist even when Chern numbers are insufficient or when net Floquet-band winding cancels [1808.08923][2605.25792][2603.01760].

## 1. Foundational constructions and the meaning of flux control

The two-dimensional prototype is a discrete-time quantum walk of spin-\(1/2\) particles on a square lattice, with position states \(\lvert r\rangle\), \(r=(x,y)\in\mathbb Z^2\), and internal states \(\lvert\uparrow\rangle,\lvert\downarrow\rangle\). One Floquet cycle is generated by
\[
\hat W=\hat F\,\hat S_y\,\hat C\,\hat S_x\,\hat C,
\]
with coin
\[
\hat C=\exp(-i\hat\sigma_y\pi/4),
\]
spin-dependent shifts
\[
\hat S_d=\sum_r \lvert r+\mathbf e_d\rangle\langle r\rvert\otimes\lvert\uparrow\rangle\langle\uparrow\rvert +\lvert r-\mathbf e_d\rangle\langle r\rvert\otimes\lvert\downarrow\rangle\langle\downarrow\rvert
=\exp(-i\hat\sigma_z \hat k_d),
\]
and a magnetic-field operator \(\hat F\) that imprints Peierls phases. For a homogeneous field in Landau gauge \(\mathbf A=(0,Bx,0)\),
\[
\hat F=\exp(i\hat\sigma_z B\hat x)
=
\begin{pmatrix}
e^{iB\hat x} & 0\\
0 & e^{-iB\hat x}
\end{pmatrix},
\qquad B=2\pi\phi.
\]
The Aharonov–Bohm phase around one plaquette is therefore \(2\pi\phi\), and more generally the appendix shows that arbitrary \(\mathbf A(\mathbf r)\) can be realized through
\[
\hat W=\hat F_y \hat S_y \hat C \hat F_x \hat S_x \hat C,
\]
with \(\hat F_d\) defined by line integrals of \(\mathbf A\) [1808.08923].

A one-dimensional formulation appears in a driven bipartite lattice, where a single-period evolution is assembled from a coin-dependent drift and a momentum-dependent coin-mixing step. In momentum space,
\[
U(k)=e^{-i\theta_2(k)\sigma_x}\,e^{-i\theta_1(k)\sigma_z},
\]
with
\[
\theta_1(k)=Tt_1\sin(k-\phi),\qquad
\theta_2(k)=\frac{T}{2}\bigl(M+2t_2\cos k\bigr).
\]
Here the flux phase \(\phi\) shifts the drift by \(k\to k-\phi\), so the phase diagram is controlled directly in the \((M,\phi)\) plane [2605.25792].

A distinct one-dimensional realization uses orbital degrees of freedom in a spin-independent optical lattice. The lowest two orbital bands \((s,p)\) serve as the internal coin space, and multi-frequency lattice-depth modulation engineers a driven \(s\)-\(p\) ladder with tunable effective hopping phases. The crucial ingredient is the nearest-neighbor \(s\)-\(p\) overlap \(\nu_1^{ps}\), whose sign reverses under bond reversal because the \(p\)-orbital Wannier function has odd parity, producing a staggered antisymmetric structure in the hopping matrix. In this setting, the synthetic “flux” is not a geometric lattice flux threaded through plaquettes but a drive-controlled sign structure of the effective transverse couplings [2603.01760].

These constructions show that the phrase “flux-controlled” is used in more than one precise sense. In two dimensions it refers to a literal synthetic magnetic flux per plaquette. In one dimension it refers to a phase parameter or relative drive phase that controls the winding structure of the Floquet walk. The common element is that the control parameter enters the unitary step operator itself rather than an autonomous static Hamiltonian.

## 2. Floquet operator, quasienergy bands, and flux-dependent spectra

Because the walk is repeated stroboscopically, the basic spectral object is a unitary one-period operator rather than a static Bloch Hamiltonian. In the magnetic two-dimensional walk, the effective Floquet Hamiltonian is defined by
\[
\hat H_{\mathrm{eff}}=i\log(\hat W), \qquad
\lvert\Psi_n\rangle=\hat W^n\lvert\Psi_0\rangle=e^{-in\hat H_{\mathrm{eff}}}\lvert\Psi_0\rangle,
\]
so the eigenvalues are quasienergies defined modulo \(2\pi\). For rational flux \(\phi=p/q\), the magnetic unit cell becomes \(q\times 1\), the magnetic Brillouin zone shrinks in \(k_x\), and the original two-band spectrum splits into \(2q\) Floquet bands. At zero field the spectrum has two bands touching at four Dirac points,
\[
\mathbf K_{\pm,\mp}=\left(\pm\frac{\pi}{2},\mp\frac{\pi}{2}\right),\quad E=0,
\qquad
\mathbf K_{\pm,\pm}=\left(\pm\frac{\pi}{2},\pm\frac{\pi}{2}\right),\quad E=\pi.
\]
In a strong field, exemplified by \(\phi=1/3\), the bands become nearly flat and well separated by gaps, producing a Floquet–Hofstadter butterfly. The paper emphasizes that this is a regime where the magnetic length becomes comparable to the lattice spacing and the spectrum resembles a Hofstadter fractal rather than weak-field Landau levels [1808.08923].

In the one-dimensional bipartite walk, the quasienergy dispersion follows from
\[
\cos\!\bigl[\varepsilon(k)T\bigr]=\cos\theta_1(k)\cos\theta_2(k),
\]
hence
\[
\varepsilon_\pm(k)=\pm\frac{1}{T}\arccos\!\Bigl[\cos\theta_1(k)\cos\theta_2(k)\Bigr].
\]
Gap closings occur when
\[
\theta_1(k_*)=m\pi,\qquad \theta_2(k_*)=n\pi,
\]
equivalently
\[
Tt_1\sin(k_*-\phi)=m\pi,\qquad
\frac{T}{2}\bigl(M+2t_2\cos k_*\bigr)=n\pi.
\]
If \(m+n\) is even, the closing is at quasienergy \(0\); if \(m+n\) is odd, it is at quasienergy \(\pi/T\). This gives the \(0\) and \(\pi/T\) gaps distinct topological roles [2605.25792].

In the orbital optical-lattice realization, single-tone driving
\[
\delta V(t)=\delta V_0\cos(\omega_0 t+\phi_0)
\]
opens one resonant BIS pair and yields a minimal nontrivial Floquet topology. Two-tone driving,
\[
\delta V(t)=\delta V_0^{(1)}\cos(\omega_0 t+\phi_0^{(1)})+\delta V_0^{(2)}\cos(2\omega_0 t+\phi_0^{(2)}),
\]
addresses the \(0\)-gap and \(\pi\)-gap resonances simultaneously. The two tones interfere coherently, and the relative phase \(\phi_0^{(2)}\) changes the sign configuration of the effective transverse couplings, so the same drive architecture can make winding contributions add or cancel [2603.01760].

The spectral consequence of flux control is therefore not limited to shifting band energies. It alters gap structure, band splitting, resonance geometry, and the distribution of Berry curvature or winding data across the quasienergy zone.

## 3. Topological characterization beyond ordinary band invariants

A central result of the magnetic two-dimensional walk is that Chern numbers do not fully classify the system. In the strong-field regime, the nearly flat quasienergy bands can carry nonzero Chern numbers, assigned numerically by standard Berry-curvature methods and a Fukui-type algorithm. However, because the system is an anomalous Floquet topological insulator, a second topological number is required: the RLBL invariant. This is a gap invariant defined as a three-dimensional winding number of the periodized Floquet evolution over \((k_x,k_y,t)\). For the chosen branch cut of the logarithm it is naturally associated with the \(E=\pi\) gap, although shifting the branch cut associates it with any chosen gap. Rather than evaluating the full three-dimensional winding directly, the paper uses the spectral-flow method of Asbóth et al. by introducing a fictitious field \(\beta\) and defining
\[
\hat F_1(\beta,B)=\exp\!\big[i\hat\sigma_z(\beta\lfloor \hat x/q\rfloor + B\hat x)\big],
\]
\[
\hat W_1(\beta,B,\tilde E)=e^{i\tilde E}\,\hat F_1(\beta,B)\hat S_y\hat C\hat S_x\hat C,
\]
followed by
\[
R=\frac{1}{2\pi}\left(\sum_{j=1}^{2sq}E_j(1/s,B,\tilde E)-\sum_{j=1}^{2sq}E_j(0,B,\tilde E)\right).
\]
Differences of these gap invariants equal sums of Chern numbers of the bands between the gaps, so the RLBL data determines the Chern data, but not conversely [1808.08923].

In one-dimensional chiral walks, the anomalous structure is encoded in two symmetric time frames. For the driven bipartite model,
\[
U_1(k)=e^{-i\frac{\theta_1(k)}{2}\sigma_z}e^{-i\theta_2(k)\sigma_x}e^{-i\frac{\theta_1(k)}{2}\sigma_z},
\]
\[
U_2(k)=e^{-i\frac{\theta_2(k)}{2}\sigma_x}e^{-i\theta_1(k)\sigma_z}e^{-i\frac{\theta_2(k)}{2}\sigma_x},
\]
with winding numbers
\[
W_\ell=\frac{1}{2\pi}\int_{-\pi}^{\pi} dk\,\partial_k\arg\!\bigl[n_{x,\ell}(k)+i n_{z,\ell}(k)\bigr], \qquad \ell=1,2.
\]
The physically relevant gap invariants are then
\[
\nu_0=\frac{W_1+W_2}{2},\qquad \nu_\pi=\frac{W_1-W_2}{2}.
\]
This separates the topology of the \(0\) and \(\pi/T\) quasienergy gaps and yields trivial, \(0\)-only, \(\pi\)-only, and coexistence sectors in the \((M,\phi)\) plane [2605.25792].

The BIS-based optical-lattice formulation resolves topology gap by gap through local charges
\[
\nu_j=\frac12\Big\{\operatorname{sgn}\!\big[h_{F,y}(k_{j,R})\big]-\operatorname{sgn}\!\big[h_{F,y}(k_{j,L})\big]\Big\},
\]
defined on BIS momenta satisfying \(h_{F,z}^{(j)}(k_{j,L/R})=0\). The gap-resolved invariants are
\[
W_0=\sum_{\alpha_j=0}\nu_j,\qquad
W_\pi=\sum_{\alpha_j=\pi}\nu_j,
\]
and the total winding is
\[
W=\sum_j(-1)^{\alpha_j/\pi}\nu_j.
\]
For \(\phi_0^{(2)}=0\), the paper reports \((W_0,W_\pi)=(1,1)\) and \(W=2\). For \(\phi_0^{(2)}=\pi\), it reports \((W_0,W_\pi)=(1,-1)\) and \(W=0\). The latter is a canonical anomalous situation: the net Floquet-band winding cancels while the gap invariants remain nontrivial [2603.01760].

A common misconception is that anomalous Floquet quantum walks can be classified by band topology alone. The literature considered here shows the opposite. In two dimensions, Chern numbers are incomplete without gap invariants. In one-dimensional chiral settings, the topology is intrinsically bipartite between \(0\) and \(\pi/T\) gaps. Gap topology, not a single static-band invariant, is the organizing principle.

## 4. Bulk–boundary correspondence, edge transport, and dynamical readout

For magnetic quantum walks, the Floquet bulk–boundary correspondence is formulated directly in terms of gap invariants. At an interface between two magnetic domains \(A\) and \(B\), the minimal number of protected edge modes crossing a chosen quasienergy gap is
\[
R_A-R_B.
\]
The main example takes opposite flux domains, with \(\phi=1/3\) outside and \(\phi=-1/3\) inside a stripe or island. Under \(\phi\to-\phi\), both Chern numbers and RLBL invariants change sign while the bulk gap positions remain fixed, so every gap carries a nonzero invariant mismatch. The computed stripe spectrum shows chiral midgap branches in every quasienergy gap, and real-space evolution of a walker initialized near a curved boundary remains confined to the interface, splits into counterpropagating components, and propagates without backscattering even around sharp corners [1808.08923].

In the one-dimensional chiral bipartite walk, the coexistence sector contains a \(0\) mode and a \(\pi\) mode on the same edge,
\[
\mathcal H_{\mathrm{edge}}^{(L)}=\mathrm{span}\{\ket{L,0},\ket{L,\pi}\}.
\]
On this subspace the Floquet operator acts as
\[
U\big|_{\mathcal H_{\mathrm{edge}}^{(L)}}=
\begin{pmatrix}
1&0\\
0&-1
\end{pmatrix}
\equiv\tau_z,
\]
so after \(m\) periods a superposition acquires a relative factor \((-1)^m\). Local boundary observables therefore take the form
\[
\expval{\hat O_{\mathrm{edge}}(m)}=O_{\mathrm{dc}}+(-1)^m O_{\mathrm{ac}},
\]
which gives a clear \(2T\) response. The same paper also introduces frame-resolved mean chiral displacements \(C_\ell(m)\), whose running averages satisfy
\[
-2C_1(m)\to W_1,\qquad -2C_2(m)\to W_2,
\]
in the clean pre-reflection window, and compares representative \(0\)-gap and \(\pi\)-gap critical points, finding that the \(\pi\)-gap benchmark shows much stronger odd-even alternation [2605.25792].

The optical-lattice experiment detects its gap invariants with a mixed-modulation Ramsey interferometer. A lattice-position shaking pulse prepares a coherent superposition, a dark interval allows phase accumulation, and a lattice-depth modulation pulse performs the readout. The measured spin imbalance is
\[
\langle \sigma_z(k)\rangle = \frac{n_s(k)-n_p(k)}{n_s(k)+n_p(k)} \equiv n_{f,z}(k),
\]
with a robust BIS signature given by a \(\pi\)-phase contrast between left and right BIS momenta. Controlled quenches then probe the effect of the phase knob away from exact resonance: for \(\phi_0^{(2)}=0\), the node \(k_c\) defined by \(h_{F,y}(k_c)=0\) lies between the BISs and produces nearly frozen dynamics, whereas for \(\phi_0^{(2)}=\pi\) there is no such node and the dynamics remain oscillatory and damped [2603.01760].

Across these models, flux control does not merely select whether boundary modes exist. It also determines whether anomalous sectors coexist on the same boundary, which gaps carry chiral transport, and what temporal signatures appear in local observables.

## 5. Experimental realizations and implementation strategies

The two-dimensional magnetic walk was proposed for neutral \(^{133}\)Cs atoms in overlapping spin-dependent square optical lattices with lattice constant \(a=\lambda_L/\sqrt2\). The two internal states are \(\lvert\uparrow\rangle=\lvert F=4,m_F=3\rangle\) and \(\lvert\downarrow\rangle=\lvert F=3,m_F=3\rangle\). The shifts \(\hat S_x\) and \(\hat S_y\) are implemented by rotating the polarization of one lattice beam so that the two spin-dependent lattices move in opposite directions, the coin is a microwave \(\pi/2\) pulse, and the magnetic phase operator is realized by short light pulses that generate spin-dependent AC Stark shifts rather than a real magnetic-field gradient. To avoid the large intensities required for a linear phase ramp, the proposal uses a sawtooth intensity profile that folds the phase into the first \(2\pi\) interval. Because the phase is \(2\pi\)-periodic, the folded sawtooth is equivalent to a linear ramp modulo \(2\pi\). The paper also introduces a superlattice version with “super-shift” operators that move atoms by multiple sites, effectively increasing the optical resolution by a factor \(m\) and improving robustness against misalignment and smoothing of the sawtooth edges [1808.08923].

The one-dimensional orbital realization is an experiment in a spin-independent optical lattice with
\[
H(t)=\frac{\hat{p}^2}{2m}+\frac{1}{2}V_0\cos(2k_L x)+\frac{1}{2}\delta V(t)\cos(2k_L x).
\]
Restricting to the lowest two orbital bands yields a tight-binding model in the \((p,s)\) basis with onsite matrix \(M(t)\) and hopping matrix \(J(t)\), where the modulation-induced nearest-neighbor \(s\)-\(p\) overlap \(\nu_1^{ps}\) is sizable in the relevant lattice depths. The experimental contribution is a lattice-depth modulation scheme that induces staggered nearest-neighbor \(s\)-\(p\) orbital couplings, together with multi-frequency control whose tunable relative phase directly sets whether the windings in the \(0\) and \(\pi\) gaps align or oppose [2603.01760].

A related but distinct cold-atom experiment realizes an anomalous Floquet topological system with bosonic \(^{39}\mathrm{K}\) atoms in a periodically driven honeycomb lattice. The intensities of the three lattice beams are modulated as
\[
I_i(t)=I_0(1- m + m\, \cos(\omega t + \phi_i)),\qquad
\phi_i=\frac{2\pi}{3}\times(i-1).
\]
This platform is not formulated as a discrete-time quantum walk in the narrow sense, but it is described as a continuous-time Floquet analog of a topological quantum walk. Its key advance is the experimental reconstruction of the complete set of Floquet invariants \((W^0,W^\pi)\) from quasienergy-gap spectroscopy and local Hall-deflection measurements, including an anomalous phase with \((W^0,W^\pi)=(1,1)\) even though \(\mathcal C^\pm=0\) [2002.09840].

These implementations establish that flux-controlled anomalous Floquet walk physics is not tied to a single microscopic architecture. State-dependent transport, orbital ladders, and continuously modulated tunnel networks can all realize the same topological logic: periodic driving creates distinct \(0\) and \(\pi\) sectors, and a flux-like control parameter tunes their winding structure.

## 6. Generalizations, limitations, and related boundary topologies

A recent generalization embeds anomalous Floquet topology in a non-Hermitian photonic lattice built from a physical resonator coordinate \(x\) and a synthetic frequency coordinate \(w\). The one-period evolution is
\[
U_T = \exp\left(-i H_2T/2\right) \exp\left(-i H_1T/2\right),
\]
with
\[
H_1 = \frac{t_w}{2i} \left( e^{\gamma_w+i\phi_w}T_w - e^{-\gamma_w-i\phi_w}T_w^{-1} \right) \otimes\sigma_z,
\]
\[
H_2 = \left[ M + t_x \left( e^{\gamma_x}T_x(\Phi) + e^{-\gamma_x}T_x^{-1}(\Phi) \right) \right] \otimes\sigma_x,
\qquad
T_x(\Phi)|x,w\rangle = e^{i\Phi w}|x+1,w\rangle .
\]
This model hosts anomalous corner pairs at quasienergies \(0\) and \(\pi/T\). The non-Bloch higher-order invariant
\[
N_\epsilon^{corner}(\Phi) = \widetilde\nu_\epsilon^x(\Phi)\, \widetilde\mu_\epsilon^w(\Phi)
\]
predicts topological existence, whereas the imaginary gauge fields \(\gamma_x,\gamma_w\) determine skin accumulation and the real flux \(\Phi\) controls the local interference matrix element
\[
S^{RR}_\mathcal{C}(\Phi) = \sum_a S_a e^{i m_a\Phi}.
\]
As a result, the same topological sector can be bright, skin-dark, or flux-dark in local optical measurements. The same complex gauge can also tune an exceptional point in the projected two-period corner propagator, where the doubled-period sign alternation survives but the response acquires an algebraic envelope because of a Jordan block [2606.07038].

A separate line of work shows that anomalous Floquet boundary states in quantum walks can also arise from extrinsic boundary topology. In that framework, a general quantum walk is written as
\[
U_{\rm QW}(\mathbf{k})=\prod_j U_j(\mathbf{k}),
\]
and boundary-local unitary operators can create or remove gapless boundary states even when the bulk effective Hamiltonian is topologically trivial. The classification is organized by invariants of the boundary unitary operator \(U_{\rm BDQW}(\mathbf{k}_\parallel)\), and the paper derives
\[
\sum_{\epsilon_\alpha=0}\nu_\alpha^0 = (-1)^d\sum_{\epsilon_\alpha=\pi}\nu_\alpha^\pi = n,
\]
with \(n\) the topological number of the boundary unitary operator. That work does include a flux-insertion formula,
\[
w_1[\hat U(\Phi)] = \int_0^{2\pi}\frac{d\Phi}{2\pi} \,\mathrm{tr}\!\left[\hat U^\dagger(\Phi)\,i\partial_\Phi \hat U(\Phi)\right],
\]
but it treats flux as a diagnostic for a winding number rather than the central physical knob of a flux-controlled anomalous phase [2112.03167].

These extensions sharpen two distinctions. First, anomalous Floquet boundary phenomena need not reduce to band topology, even in first-order quantum walks. Second, flux control and boundary topology are logically distinct: flux can tune the full Floquet winding structure, while boundary-local unitary structure can independently add or remove anomalous states. A plausible implication is that future classifications of flux-controlled anomalous Floquet quantum walks will continue to separate bulk gap topology, boundary unitary topology, and measurement visibility rather than compressing them into a single invariant.

Source: https://www.emergentmind.com/topics/flux-controlled-anomalous-floquet-quantum-walk