betaFeedback

WorksPhysics and Astronomy

Light-Induced Odd-Parity h-Wave Magnetism in Penrose Quasicrystals

Abstract

Noncrystallographic rotational symmetry offers a route to unconventional spin splitting beyond the constraints of periodic crystals. We show that circularly polarized light induces odd-parity hh-wave magnetism in a collinear antiferromagnetic Penrose quasicrystal without spin-orbit coupling. Light-induced chiral hopping, generated by virtual two-step processes, combines with staggered exchange to lift the spin degeneracy of the undriven system. Quasicrystalline spin-group symmetry enforces odd-parity spin splitting and selects hh-wave as its leading angular harmonic. An analytical effective quasi-band theory relates this splitting to the geometry of local hopping paths and predicts its reversal upon switching the light helicity. Momentum-resolved spectral calculations reveal the characteristic alternating spin contrast, for which we propose spin-resolved time- and angle-resolved photoemission spectroscopy as an experimental probe. Our results establish a mechanism for optically controlling higher-harmonic odd-parity magnetism and demonstrate how quasicrystalline order extends Floquet engineering beyond crystallographic symmetry constraints.

Introduction

Unconventional collinear magnetism has emerged as a new class of magnetic order characterized by momentum-dependent spin splitting without spin-orbit coupling (SOC) [1, 2, 3, 4]. A prominent example is altermagnetism, which combines compensated magnetic moments with even-parity spin splitting exhibiting characteristic dd-, gg-, or higher-wave angular harmonics [5, 6, 7, 8, 9, 10, 11, 12, 13, 14]. More recently, odd-parity magnetism has attracted growing attention, with spin splitting satisfying Δλ(−k)=−Δλ(k)\Delta_{\lambda}(-\boldsymbol{k})=-\Delta_{\lambda}(\boldsymbol{k}), where Δλ\Delta_{\lambda} denotes the energy difference between opposite-spin states in a matched branch λ\lambda [15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26]. Such unconventional spin splitting can exhibit pp-, ff-, and higher-wave angular harmonics, providing additional possibilities for spin-dependent electronic responses. Floquet engineering offers a promising route to dynamically control these magnetic states by breaking the symmetries that protect spin degeneracy [27, 28]. Recent studies have proposed light-induced pp- and ff-wave spin splitting in collinear antiferromagnets [29, 30, 31, 32, 33, 34, 35, 36]. Nevertheless, realizing higher-harmonic odd-parity magnetism through optical driving remains challenging, motivating the search for systems with unconventional rotational symmetries.

Quasicrystals provide a distinctive platform for exploring unconventional magnetism beyond crystallographic symmetry constraints [37, 38, 39, 40, 41, 42, 43, 44]. Unlike periodic crystals, they possess long-range quasiperiodic order and noncrystallographic rotational symmetries, such as five-, eight-, and tenfold rotations. Long-range antiferromagnetic order has recently been observed in quasicrystalline materials [45, 46], while theoretical studies have predicted ferromagnetic, ferrimagnetic, and altermagnetic phases in quasiperiodic lattices [47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57]. In particular, quasicrystalline altermagnetism demonstrates that noncrystallographic rotational symmetry can support unconventional higher-harmonic spin splitting. The fivefold rotational symmetry of the Penrose quasicrystal is especially intriguing: when preserved within a spin sector, it excludes the lower odd-parity pp- and ff-wave harmonics, making hh-wave the leading symmetry-allowed angular structure. This raises two fundamental questions: Can optical driving generate such higher-harmonic odd-parity magnetism in a quasicrystal? If so, how can its characteristic spin splitting be identified and experimentally detected in the absence of conventional Bloch bands?

In this Letter, we demonstrate that circularly polarized light (CPL) induces odd-parity hh-wave spin splitting in a collinear antiferromagnetic Penrose quasicrystal without SOC. The underlying mechanism originates from light-induced chiral hopping generated by virtual two-step processes, which, together with staggered magnetic exchange, lifts the spin degeneracy of the undriven system. Momentum-resolved spectral calculations reveal a characteristic fifth-harmonic spin contrast, whose odd parity and angular structure are established by quasicrystalline spin-group symmetry. We further develop an analytical effective quasi-band theory that connects the spin splitting to the geometry of local hopping paths and predicts its reversal upon switching the light helicity. Finally, we propose spin-resolved time- and angle-resolved photoemission spectroscopy (trARPES) to detect the characteristic spin splitting. Our findings establish a microscopic mechanism for optically controlling higher-harmonic odd-parity magnetism and reveal new opportunities for Floquet engineering enabled by noncrystallographic rotational symmetry.

Model and light-induced virtual hopping.— The Penrose tiling can be constructed by projecting selected sites of a five-dimensional (5D) hypercubic lattice onto the 2D physical plane [58]. Each parent-lattice site is labeled by an integer vector ni=∑ν=15niνeν\mathbf{n}_i = \sum_{\nu=1}^{5} n_{i\nu}\mathbf{e}_{\nu}, where eν\mathbf{e}_{\nu} denotes the ν\nu-th hypercubic basis vector and niν∈Zn_{i\nu} \in\mathbb{Z}. Its physical-space position is ri=Sni\mathbf{r}_i = S\mathbf{n}_i, where SS is the projection matrix (see Supplemental Material (SM) Sec. I [59]). The parity χi=(−1)∑ν=15niν\chi_i = (-1)^{\sum_{\nu=1}^{5} n_{i\nu}} defines magnetic sublattices AA and BB, connected by edges of length aa in 2D physical plane. The acceptance window, a bounded region in perpendicular space, selects the retained parent sites and determines the frequencies of local vertex environments (Figure 1; SM Sec. I [59]). In the absence of light, electrons hop along edges with amplitude −t1-t_1 and experience a staggered exchange field χiN\chi_i\mathbf{N}, where N=Nz^\mathbf{N}=N\hat{\mathbf{z}}. Without SOC, spin along zz is conserved, with s=±1s=\pm1 denoting spin up and down. The undriven antiferromagnet serves as the reference state, whose bulk spin degeneracy is protected by the symmetries discussed below.

Figure 1. Light-induced chiral virtual hopping in a Penrose quasicrystal. (a) Bipartite Penrose tiling, with red and blue arrows indicating effective virtual hoppings within the AA and BB magnetic sublattices, respectively. (b) A virtual two-step path i−j−ki-j-k connects sites on the same sublattice. Contributions through the two intermediate vertices of a rhombus cancel along its diagonal. (c) Distinct local vertex environments determine the multiplicities of surviving chiral hopping paths [58, 60, 61].

We apply CPL propagating normal to the plane, with vector potential A(t)=A0(cos⁡ωt,ηsin⁡ωt)\mathbf{A}(t)=A_0(\cos\omega t,\eta\sin\omega t), where A0A_0, ω\omega, and η=±1\eta=\pm1 denote the amplitude, angular frequency, and helicity, respectively. Incorporating the orbital coupling through the Peierls substitution gives

H(t)=N∑iχici†σzci−t1∑⟨ij⟩eieℏA(t)⋅(ri−rj)ci†cj,(1)H(t)=N\sum_i\chi_i c_i^\dagger\sigma_z c_i-t_1\sum_{\langle ij\rangle}e^{\frac{ie}{\hbar}\mathbf{A}(t)\cdot(\mathbf{r}_i-\mathbf{r}_j)}c_i^\dagger c_j, \tag*{(1)}

where ci=(ci↑,ci↓)Tc_i=(c_{i\uparrow},c_{i\downarrow})^{T}, σz\sigma_z is a spin Pauli matrix, and e>0e>0 is the elementary charge. The bond sum includes both orientations, ensuring Hermiticity. We assume a spatially uniform optical field and a fixed Néel background.

In the off-resonant high-frequency regime, where ℏω\hbar\omega exceeds the relevant electronic energy scales, the driven system is described by an effective Floquet Hamiltonian. Expanding H(t)=∑nHne−inωtH(t)=\sum_n H_n e^{-in\omega t}, with Hn=(ω/2π)∫02π/ωdt H(t)einωtH_n=(\omega/2\pi)\int_0^{2\pi/\omega}dt\,H(t)e^{in\omega t}, the leading van Vleck expansion yields [62]

HF=H0+∑n>0[H−n,Hn]nℏω+O(ω−2).(2)H_F=H_0+\sum_{n>0}\frac{[H_{-n},H_n]}{n\hbar\omega}+O(\omega^{-2}). \tag*{(2)}

The cycle-averaged term H0H_0 renormalizes hopping as t1→t1J0(α0)t_1\to t_1J_0(\alpha_0), where α0=eA0a/ℏ\alpha_0=eA_0a/\hbar and JnJ_n is the Bessel function of the first kind. For n≠0n\ne0, the Fourier components of the hopping are proportional to inJn(α0)einηθiji^nJ_n(\alpha_0)e^{in\eta\theta_{ij}}, where θij\theta_{ij} is the polar angle of ri−rj\mathbf{r}_i-\mathbf{r}_j. Details of the Fourier derivation are provided in SM Sec. II [59].

The commutator in Eq. (2) describes virtual two-step hopping through an intermediate vertex jj, accompanied by the absorption and emission of nn photons. Interference between the two operator orderings generates an imaginary, helicity-dependent hopping. Combining a path ii-jj-kk with its Hermitian conjugate gives

HFi−j−k=iηΛ(δ)(ci†ck−ck†ci),H_F^{i-j-k}=i\eta\Lambda(\delta)(c_i^\dagger c_k-c_k^\dagger c_i),
Λ(δ)=2t12ℏω∑n>0Jn2(α0)nsin⁡(nδ),(3)\Lambda(\delta)=\frac{2t_1^2}{\hbar\omega}\sum_{n>0}\frac{J_n^2(\alpha_0)}{n}\sin(n\delta), \tag*{(3)}

where δ=θkj−θij\delta=\theta_{kj}-\theta_{ij} is the oriented angle between the two bonds directed outward from jj. Since each elementary hop switches sublattice, the effective hopping connects sites with parallel local magnetic moments. Its imaginary amplitude changes sign upon reversing either the path orientation or the light helicity.

Crucially, contributions from different intermediate vertices interfere coherently. For endpoints across a rhombus diagonal, the two virtual paths yield opposite amplitudes and cancel Figure 1(b). The surviving chiral virtual hoppings therefore depend on the local vertex environments, as illustrated in Figure 1(c), rather than solely on the endpoint pairs. Their enumeration and weights of distinct types of local vertex environments and path counting are detailed in SM Secs. I and II [59].

In the weak-drive limit, Λ(δ)=t12α02sin⁡δ/(2ℏω)+O(α04)\Lambda(\delta) = t_1^2\alpha_0^2\sin\delta/(2\hbar\omega) + O(\alpha_0^4). Although the induced hopping is spin-independent, it connects sites within the same magnetic sublattice. Combined with staggered exchange, which couples spin to sublattice, the difference between sublattice-dependent hopping contributions generates momentum-dependent spin splitting. This mechanism is the quasiperiodic counterpart of Haldane-type hopping in crystalline Floquet magnets and related sublattice-current mechanisms [22, 29, 34, 35, 36, 63, 64]. Here, the chiral hoppings are generated optically, with their path interference governed by Penrose geometry, without requiring spontaneous current order. In contrast, the cycle-averaged renormalization of real hoppings alone cannot lift the spin degeneracy.

Momentum-resolved spectral signature.— We first calculate the spectrum of the finite real-space Hamiltonian HFH_{\mathrm{F}}, fully accounting for local environments and interference between hopping paths. Although momentum is not a conserved quantum number in quasicrystals, projection onto plane-wave states provides a momentum-resolved spectral function without assuming translational symmetry:

∣k,s⟩=1Nsite∑jeik⋅rj∣j,s⟩,\lvert\mathbf{k},s\rangle= \frac{1}{\sqrt{N_{\mathrm{site}}}}\sum_j e^{i\mathbf{k}\cdot\mathbf{r}_j}\lvert j,s\rangle,
As(k,ε)=−1πIm⁡⟨k,s∣(ε+iΓ−HF)−1∣k,s⟩.(4)A_s(\mathbf{k},\varepsilon) = -\frac{1}{\pi}\operatorname{Im}\langle\mathbf{k},s\lvert(\varepsilon+i\Gamma-H_{\mathrm{F}})^{-1}\rvert\mathbf{k},s\rangle. \tag*{(4)}

Here NsiteN_{\mathrm{site}} is the number of vertices, and Γ>0\Gamma> 0 denotes the Lorentzian energy broadening. The resulting spectral function resolves the eigenenergies weighted by their spin-polarized plane-wave overlaps, revealing dispersive features whose intensities encode the diffraction and local-environment structure of the quasicrystal.

Figure Figure 2(a) shows dispersive spin-difference spectra along adjacent radial momentum paths from Γ\Gamma, which exhibit opposite spin characters. At fixed energy, ΔA=A↑−A↓\Delta\mathcal{A}=\mathcal{A}_{\uparrow}-\mathcal{A}_{\downarrow} displays ten alternating angular sectors in the 2D momentum plane Figure 2(b), consistent with odd-parity hh-wave magnetism. This spin contrast vanishes in the undriven bulk limit, confirming its light-induced origin.

Figure 2. Spin-resolved spectral functions of CPL-induced odd-parity hh-wave magnetism in the Penrose quasicrystal. (a) Spin-difference spectral cuts along the indicated momentum paths. Inset: decagonal pseudo-Brillouin zone used as a plotting domain, not a primitive reciprocal unit cell. (b) ΔA=A↑−A↓\Delta A = A_{\uparrow} - A_{\downarrow} in the momentum plane at ε=−1.20 eV\varepsilon= -1.20\,\mathrm{eV}. (c) Angular weights PℓP_{\ell} versus radius kk for the map in (b). The direct spectral calculation uses Nsite=1387N_{\mathrm{site}} = 1387, t1=N=1 eVt_1 = N = 1\,\mathrm{eV}, ω=20 rad/fs\omega= 20\,\mathrm{rad/fs}, α0=1.5\alpha_0 = 1.5, and Γ=0.01 eV\Gamma= 0.01\,\mathrm{eV}.

To quantify the angular pattern, we write k=k(cos⁡θ,sin⁡θ)\mathbf{k}=k(\cos\theta,\sin\theta), where θ\theta is the momentum polar angle, distinct from the bond angles in Eq. (3), and define

aℓ(k,ε)=12π∫02πdθ ΔA(k,θ,ε)e−iℓθ,Pℓ=∣aℓ∣2,(5)a_{\ell}(k,\varepsilon)=\frac{1}{2\pi}\int_{0}^{2\pi}d\theta\,\Delta\mathcal{A}(k,\theta,\varepsilon)e^{-i\ell\theta},\qquad P_{\ell}=\lvert a_{\ell}\rvert^{2}, \tag*{(5)}

where the integer ℓ\ell labels the angular harmonic. An hh-wave response corresponds to ∣ℓ∣=5\lvert\ell\rvert=5, with ten alternating sign sectors around a momentum circle. As shown in Figure 2(c), P5P_{5} dominates the displayed angular harmonics over the radial range with appreciable spin contrast, confirming the predominant hh-wave character. Reversing the light helicity reverses the sign of the spin spectral contrast while leaving PℓP_{\ell} unchanged, demonstrating helicity-controlled reversal of the spin splitting without altering its hh-wave symmetry (see SM Sec. III [59] for details).

Symmetry origin of light-induced hh-wave magnetism. — We first establish the undriven bulk spin degeneracy and then identify the symmetry constraints on the light-induced splitting. In quasicrystals, symmetry is defined through indistinguishability: symmetry-related configurations have identical bulk frequencies of finite local patterns, without requiring pointwise coincidence [65, 66, 67, 68, 69, 70, 71]. For magnetic quasicrystals, these patterns include the magnetic labels and hopping amplitudes, rather than the tiling geometry alone. We denote independent spin and spatial operations by [gs∥gr][g_s\Vert g_r] and use A‾s\overline{\mathcal{A}}_s for the thermodynamic bulk (or hull-averaged) spectral function. The resulting symmetry relations need not hold exactly for an arbitrary finite patch.

In the undriven system, the combined spin-flip–inversion operation [C2⊥s∥P][C^{s}_{2\perp}\Vert\mathcal{P}] relates indistinguishable magnetic configurations. Here C2⊥sC^{s}_{2\perp} denotes a π\pi spin rotation about an axis perpendicular to the Néel vector, while P\mathcal{P} is an affine hyperspace inversion, n↦eν−n\mathbf{n}\mapsto\mathbf{e}_{\nu}-\mathbf{n}, which acts in physical space as r↦aν−r\mathbf{r}\mapsto\mathbf{a}_{\nu}-\mathbf{r}, with aν=Seν\mathbf{a}_{\nu}=S\mathbf{e}_{\nu}. This operation exchanges the magnetic sublattices, while its perpendicular-space action preserves the bulk frequencies of magnetically labeled local patterns (see SM Sec. IV [59]). The combined operation therefore relates opposite spins at opposite probe momenta. Moreover, each undriven spin block is real in the site basis, implying As(k,ε)=As(−k,ε)\mathcal{A}_s(\mathbf{k},\varepsilon)=\mathcal{A}_s(-\mathbf{k},\varepsilon) through complex conjugation. Combining these relations gives

A‾s(k,ε)=A‾−s(−k,ε)=A‾−s(k,ε).(6)\overline{A}_s(k,\varepsilon)=\overline{A}_{-s}(-k,\varepsilon)=\overline{A}_{-s}(k,\varepsilon). \tag*{(6)}

Thus, the opposite-spin bulk spectra coincide at the same probe momentum, without requiring an exact time-spatial inversion symmetry or Kramers pairing of individual eigenstates in a finite Penrose patch.

Under CPL, the induced imaginary hopping removes the real-Hamiltonian constraint that enforces equal spectral weights at k\mathbf{k} and −k-\mathbf{k} within each spin sector. Nevertheless, [C2⊥s∥P][C^{s}_{2\perp}\parallel\mathcal{P}] remains a bulk symmetry relation: P\mathcal{P} preserves the oriented angles and coherent path interference of the induced chiral hopping, while the accompanying spin rotation restores the staggered exchange field (see SM Sec. IV [59]). Consequently,

A‾s(k,ε)=A‾−s(−k,ε),Δλ(−k)=−Δλ(k),(7)\begin{aligned} \overline{A}_s(k,\varepsilon)&=\overline{A}_{-s}(-k,\varepsilon),\\ \Delta_{\lambda}(-k)&=-\Delta_{\lambda}(k), \tag*{(7)} \end{aligned}

where Δλ(k)=Eλ,↑(k)−Eλ,↓(k)\Delta_{\lambda}(k)=E_{\lambda,\uparrow}(k)-E_{\lambda,\downarrow}(k) denotes the splitting between matched bulk spectral or quasi-band branches. Thus, the combined spin-flip–inversion symmetry enforces odd parity without requiring opposite-spin degeneracy at generic momentum.

The spin-preserving fivefold operation [Es∥C5z][E_s\parallel C_{5z}] further constrains the angular dependence:

Δλ(k,θ+2π/5)=Δλ(k,θ).(8)\Delta_{\lambda}(k,\theta+2\pi/5)=\Delta_{\lambda}(k,\theta). \tag*{(8)}

Expanding Δλ(k,θ)\Delta_{\lambda}(k,\theta) in angular harmonics eiℓθe^{i\ell\theta}, fivefold symmetry requires ℓ\ell to be a multiple of five. Odd parity under θ→θ+π\theta\to\theta+\pi further restricts ℓ\ell to odd integers, leaving ∣ℓ∣=5,15,25,…|\ell|=5,15,25,\ldots. Thus, pp- and ff-wave components are forbidden, and hh-wave is the lowest symmetry-allowed harmonic. For an analytic quasi-band branch near Γ\Gamma, the lowest-order allowed contribution is Δλ(k,θ)=k5[cλcos⁡(5θ)+dλsin⁡(5θ)]+O(k7)\Delta_{\lambda}(k,\theta)=k^5[c_{\lambda}\cos(5\theta)+d_{\lambda}\sin(5\theta)]+O(k^7), with coefficients determined by the microscopic hopping geometry.

Combining [Es∥C5z][E_s\parallel C_{5z}] with [C2⊥s∥P][C^{s}_{2\perp}\parallel\mathcal{P}] yields

A‾s(Rπ/5k,ε)=A‾−s(k,ε),(9)\overline{A}_s(R_{\pi/5}k,\varepsilon)=\overline{A}_{-s}(k,\varepsilon), \tag*{(9)}

where Rπ/5R_{\pi/5} rotates momentum by π/5\pi/5. Accordingly, the spin-summed spectrum is invariant under a π/5\pi/5 rotation, whereas the spin-difference spectrum changes sign, producing ten alternating spin sectors. This relation corresponds to the unitary spin-group operation [C2⊥s∥C10z][C^{s}_{2\perp}\Vert C_{10z}], rather than physical time-reversal symmetry [72]. Thus, optical driving permits spin splitting, while [C2⊥s∥P][C^{s}_{2\perp}\Vert P] enforces odd parity and the fivefold rotational symmetry selects hh-wave as the lowest allowed angular harmonic. The microscopic theory determines whether this splitting is nonzero, and its fifth-harmonic dominance is tested through Eq. (5).

Long-wavelength quasi-band theory.— The direct spectral calculation reveals the angular spin contrast without relying on an effective momentum-space model. To derive its microscopic origin and magnitude analytically, we construct a quasi-band theory by projecting the Hamiltonian onto states associated with groups of sites sharing specified local environments [73]. In the cut-and-project construction, these groups correspond to subregions of the acceptance window [74]. The effective hopping matrix elements are determined by overlaps of translated acceptance-window regions, which encode the frequencies of allowed hopping paths. For Floquet-induced two-step processes, these overlaps also enforce the existence of the intermediate vertex. Refining the window partition systematically incorporates more local environments. This construction provides an effective momentum-resolved description without assuming independent exact Bloch sectors, which are absent in a quasicrystal. The overlap weights and coherent spectral weights are derived in SM Sec. V [59].

For a group WbW_b containing NbN_b sites, we introduce the normalized state

∣k,Wb,s⟩=1Nb∑i∈Wbeik⋅ri∣i,s⟩(10)\lvert k,W_b,s\rangle=\frac{1}{\sqrt{N_b}}\sum_{i\in W_b}e^{i k\cdot r_i}\lvert i,s\rangle \tag*{(10)}

and construct the effective Hamiltonian from the matrix elements ⟨k,Wa,s∣HF∣k,Wb,s⟩\langle k,W_a,s\lvert H_F\lvert k,W_b,s\rangle. States associated with disjoint groups are orthogonal at fixed kk, but generally not at different momenta. The resulting matrix elements retain coherent interference among allowed virtual-hopping paths.

At the coarsest level, retaining one state per magnetic sublattice yields

H(0)(k)=Nσzτz−B5(k)τx−ηD5(k)τz(11)\mathcal{H}^{(0)}(k)=N\sigma_z\tau_z-B_5(k)\tau_x-\eta D_5(k)\tau_z \tag*{(11)}

where τ\tau are Pauli matrices in the A/BA/B sublattice space and aj=a(cos⁡2πj/5,sin⁡2πj/5)a_j=a(\cos2\pi j/5,\sin2\pi j/5) denote the five edge directions. The even function B5=4t1J0(α0)∑j=04cos⁡(k⋅aj)/5B_5=4t_1J_0(\alpha_0)\sum_{j=0}^{4}\cos(k\cdot a_j)/5 describes real inter-sublattice hybridization, whereas the chiral intra-sublattice hopping generates opposite diagonal shifts on the two sublattices, characterized by the odd function

D5(k)=ξ5∑δ=2π/5, 4π/5Λ(δ)∑j=04sin⁡(Lδk⋅uj)(12)D_5(k)=\frac{\xi}{5}\sum_{\delta=2\pi/5,\,4\pi/5}\Lambda(\delta)\sum_{j=0}^{4}\sin(L_\delta k\cdot u_j) \tag*{(12)}

Here the geometric factor ξ=55−11\xi= 5\sqrt{5}-11 arises from the local-environment-weighted path counts. The two values of δ\delta correspond to the surviving virtual-hopping geometries, with endpoint separations Lδ=2asin⁡(δ/2)L_{\delta} = 2a\sin(\delta/2) and displacement directions uj=(cos⁡[(4j+1)π/10],sin⁡[(4j+1)π/10])\mathbf{u}_j = (\cos[(4j+1)\pi/10],\sin[(4j+1)\pi/10]). Thus, D5D_5 follows directly from the Floquet-induced virtual hopping in Eq. (3), rather than being introduced phenomenologically.

Diagonalizing the effective Hamiltonian gives

Eλs(k)=λB52(k)+[sN−ηD5(k)]2(13)E_{\lambda s}(\mathbf{k}) = \lambda\sqrt{B_5^2(\mathbf{k}) + [sN-\eta D_5(\mathbf{k})]^2} \tag*{(13)}

where λ=±1\lambda= \pm1 labels the upper and lower quasi-bands for each spin sector. The cross term −2sηND5-2s\eta N D_5 reveals the cooperative origin of the splitting: both staggered exchange and chiral hopping are required. For ∣D5∣≪E0=N2+B52|D_5| \ll E_0 = \sqrt{N^2+B_5^2}, the spin splitting becomes Δλ=−2ληND5/E0+O(D53/E02)\Delta_\lambda= -2\lambda\eta N D_5/E_0 + O(D_5^3/E_0^2). Reversing the light helicity exchanges the opposite-spin quasi-band energies while leaving their spin average unchanged.

In the long-wavelength limit ka≪1ka \ll1, the fivefold sum in Eq. (12) cancels the linear and cubic terms, leaving a leading fifth-order contribution (see SM Sec. VI [59]). The resulting spin splitting is

Δλ(k)=−2ληNg5N2+16t12J02(α0)k5sin⁡(5θ)+O(k7)(14)\Delta_\lambda(\mathbf{k}) = -\frac{2\lambda\eta N g_5}{\sqrt{N^2+16t_1^2J_0^2(\alpha_0)}}k^5\sin(5\theta) + O(k^7) \tag*{(14)}

where g5=(55−11)∑δ=2π/5, 4π/5Λ(δ)Lδ5/1920g_5 = (5\sqrt{5}-11)\sum_{\delta=2\pi/5,\,4\pi/5}\Lambda(\delta)L_\delta^5/1920. The sin⁡(5θ)\sin(5\theta) dependence fixes the nodal orientation through the virtual-hopping geometry and divides the momentum plane into ten alternating spin sectors. Equation (14) thus provides a microscopic derivation of the symmetry-allowed hh-wave splitting, accounting for the angular pattern in Figure 2(b) and its reversal under opposite light helicities.

Spin-resolved trARPES and optical control. — To probe the light-induced hh-wave spin splitting, we consider spin-resolved trARPES in a pump–probe setup [Fig. 3(a)] [75, 76, 77, 78, 79]. To account for the finite pulse duration and probe resolution, we employ a circularly polarized pump Apump(t)=A0e−(t/Tpump)2(cos⁡ωt,ηsin⁡ωt)\mathbf{A}_{\mathrm{pump}}(t) = A_0 e^{-(t/T_{\mathrm{pump}})^2}(\cos\omega t,\eta\sin\omega t) and a Gaussian probe envelope spr(t,td)=e−[(t−td)/Tprobe]2s_{\mathrm{pr}}(t,t_d) = e^{-[(t-t_d)/T_{\mathrm{probe}}]^2}, where TpumpT_{\mathrm{pump}} and TprobeT_{\mathrm{probe}} are the 1/e1/e half-widths of the respective amplitude envelopes, and tdt_d is the probe delay relative to the pump maximum. Neglecting momentum-dependent photoemission matrix elements and final-state effects, the spin-resolved photocurrent is given by [75, 76]

Is(k,ε,td)∝Im⁡∫dt1dt2 spr(t1,td)spr(t2,td)eiε(t1−t2)/ℏGss<(k;t1,t2),(15)I_s(\mathbf{k},\varepsilon,t_d) \propto\operatorname{Im}\int dt_1dt_2\,s_{\mathrm{pr}}(t_1,t_d)s_{\mathrm{pr}}(t_2,t_d)e^{i\varepsilon(t_1-t_2)/\hbar}G^{<}_{ss}(\mathbf{k};t_1,t_2), \tag*{(15)}

where Gss<(k;t1,t2)=i⟨cks†(t2)cks(t1)⟩G^{<}_{ss}(\mathbf{k};t_1,t_2)=i\langle c^\dagger_{\mathbf{k}s}(t_2)c_{\mathbf{k}s}(t_1)\rangle is the lesser Green function describing occupied electronic correlations. Here cksc_{\mathbf{k}s} is the normalized plane-wave transform of the site operators, as defined in Eq. (4). The probe envelope selects a time window centered at tdt_d, while the Fourier phase resolves the photoemission energy ε\varepsilon.

For independent electrons initially described by HinitH_{\mathrm{init}}, the real-space lesser Green function evolves as

G<(t1,t2)=iU(t1,t0)f(Hinit)U†(t2,t0),(16)G^{<}(t_1,t_2)=iU(t_1,t_0)f(H_{\mathrm{init}})U^\dagger(t_2,t_0), \tag*{(16)}

where U(t,t0)U(t,t_0) is the single-particle evolution operator under the pump and f(Hinit)f(H_{\mathrm{init}}) is the initial Fermi–Dirac occupation matrix, with chemical potential μ\mu and temperature TT. Projecting this correlator onto plane-wave states yields Gss<(k;t1,t2)G^{<}_{ss}(\mathbf{k};t_1,t_2) in Eq. (15). Details of the derivation and numerical implementation are provided in SM Sec. VII [59].

The calculated spin-difference photocurrent I↑−I↓I_{\uparrow}-I_{\downarrow} Figure 3(b) exhibits ten alternating angular sectors, superimposed on diffraction-related intensity modulations. Reversing the CPL helicity interchanges the opposite-spin photocurrents, providing an additional means of optical control over the spectroscopic spin contrast. Its sign indicates the dominant spin contribution to the occupied photoemission weight at a given energy and probe delay. The resulting pattern provides a spin-resolved spectroscopic signature of the predicted hh-wave electronic structure. Unlike the Floquet spectral function, however, the photocurrent also depends on nonequilibrium occupations, pulse envelopes, and probe resolution. These effects, together with the relevant optical scales and experimental limitations, are discussed in SM Sec. VIII [59].

Figure 3. Spin-resolved pump–probe photoemission. (a) Schematic of the detection protocol: a circularly polarized pump drives the sample, followed by a delayed probe that resolves the spin and in-plane momentum of photoemitted electrons. (b) Calculated spin contrast I↑−I↓I_{\uparrow} - I_{\downarrow} at ε=−1.20 eV\varepsilon= -1.20\,\mathrm{eV} and delay td=0t_d = 0. Parameters are t1=N=1 eVt_1 = N = 1\,\mathrm{eV}, α0=1.5\alpha_0 = 1.5, ω=20 rad/fs\omega= 20\,\mathrm{rad/fs}, Tpump=150 fsT_{\mathrm{pump}} = 150\,\mathrm{fs}, Tprobe=10 fsT_{\mathrm{probe}} = 10\,\mathrm{fs}, T=0T = 0, and μ=0\mu= 0.

Discussion and conclusions.— We have demonstrated that CPL can induce odd-parity hh-wave spin splitting in an antiferromagnetic Penrose quasicrystal without SOC. The central conceptual advance lies in the complementary roles of optical driving and noncrystallographic rotational symmetry: the former generates the orbital hopping responsible for lifting spin degeneracy, while the latter, together with spin-spatial symmetry, selects the unconventional hh-wave angular structure. This mechanism extends symmetry-guided design of momentum-dependent spin splitting beyond periodic crystals, where crystallographic restrictions exclude fivefold rotational symmetry.

Our predictions concern the driven electronic structure on a fixed antiferromagnetic background, rather than the formation of a new magnetic order. Their experimental realization will depend on material-specific optical coupling, nonequilibrium occupations, heating, and relaxation, while self-consistent magnetic dynamics remain an important direction for future study. More broadly, the interplay between optical driving and noncrystallographic symmetry offers a route to engineering unconventional spin-split electronic structures in quasiperiodic and other aperiodic magnetic systems.

Acknowledgments

This work is supported by the National Natural Science Foundation of China (Grants No. 12474056 and No. 12674195), the National Key R&D Program of China (Grant No. 2021YFA1401600), and the 2022 basic discipline top-notch students training program 2.0 research project (Grant No. 20222005). The work was carried out at the National Supercomputer Center in Tianjin, and the calculations were performed on Tianhe new generation supercomputer. The high-performance computing platform of Peking University supported the computational resources.

[cond-mat.mes-hall].

Bipartite Penrose tiling: cut-and-project construction

Projection from five dimensions

We construct the Penrose tiling from the integer lattice n=(n1,…,n5)T∈Z5\mathbf{n} = (n_1,\ldots,n_5)^{\mathsf{T}} \in\mathbb{Z}^5, with standard hypercubic basis eν\mathbf{e}_{\nu}, ν=1,…,5\nu= 1,\ldots,5. The five-dimensional space is decomposed into a two-dimensional physical space V∥V_{\parallel} and a three-dimensional perpendicular space V⊥V_{\perp}. A convenient choice of projection matrices is

S=12(2−σ−τ−τ−σ0βτβ−β−βτ),(17)S = \frac{1}{2} \begin{pmatrix} 2 & -\sigma& -\tau& -\tau& -\sigma\\ 0 & \beta\tau& \beta& -\beta& -\beta\tau \end{pmatrix}, \tag*{(17)}
S⊥=12(2−τ−σ−σ−τ0β−βτβτ−β22222),(18)S_{\perp} = \frac{1}{2} \begin{pmatrix} 2 & -\tau& -\sigma& -\sigma& -\tau\\ 0 & \beta& -\beta\tau& \beta\tau& -\beta\\ \sqrt{2} & \sqrt{2} & \sqrt{2} & \sqrt{2} & \sqrt{2} \end{pmatrix}, \tag*{(18)}

where τ=(1+5)/2\tau= (1+\sqrt{5})/2, σ=(1−5)/2=−1/τ\sigma= (1-\sqrt{5})/2 = -1/\tau, and β=3−τ\beta= \sqrt{3-\tau}. Each hyperlattice site has physical and perpendicular coordinates

r(n)=Sn,r⊥(n)=S⊥n.\mathbf{r}(\mathbf{n}) = S\mathbf{n}, \qquad\mathbf{r}_{\perp}(\mathbf{n}) = S_{\perp}\mathbf{n}.

The displayed coordinates use the edge length as the unit, a=1a=1; physical lengths are restored by multiplying its entries by aa. The columns SeνS\mathbf{e}_{\nu}, ν=1,…,5\nu=1,\ldots,5, give the same five edge directions as the main-text star aj\mathbf{a}_j, with j=ν−1j=\nu-1. Noting that the geometric constants σ,τ\sigma,\tau in these matrices are scalars, distinct from the spin and sublattice Pauli matrices σ,τ\boldsymbol{\sigma},\boldsymbol{\tau}.

A site is retained when its perpendicular coordinate lies inside the acceptance window W⊥W_{\perp}. The window is the projection of the five-dimensional unit hypercube, up to the fixed internal-space shift that specifies the layer origin below, and is a three-dimensional polytope with decagonal cross sections. One representation of its vertices in cylindrical coordinates consists of

(0,0,3/2),(1,θ1,2/2),(τ,θ2,1/2),(0,0,3/\sqrt{2}),\quad(1,\theta_1,2/\sqrt{2}),\quad(\tau,\theta_2,1/\sqrt{2}),
(τ,θ1,0),(1,θ2,−1/2),(0,0,−2/2),(\tau,\theta_1,0),\quad(1,\theta_2,-1/\sqrt{2}),\quad(0,0,-2/\sqrt{2}),

with

θ1∈{±π/5,±3π/5,π},θ2∈{0,±2π/5,±4π/5}.\theta_1 \in\{\pm\pi/5,\pm3\pi/5,\pi\},\qquad\theta_2 \in\{0,\pm2\pi/5,\pm4\pi/5\}.

The finite samples used in numerical calculations are obtained by restricting nn to a sufficiently large hypercubic box, applying the window test, and then selecting a nearly circular patch in physical space. Boundary sites can either be retained with open boundaries or discarded from local observables using an interior mask.

Magnetic sublattices and inversion

The parent hypercubic lattice is bipartite. We assign

χ(n)=(−1)n1+n2+n3+n4+n5,\chi(n)=(-1)^{n_1+n_2+n_3+n_4+n_5},

and define sublattice AA by χ=+1\chi=+1 and BB by χ=−1\chi=-1. A step along any hypercubic basis vector changes χ\chi, hence every projected nearest-neighbor bond connects opposite sublattices. The staggered exchange term is therefore

HN=N∑iχici†σzci=N∑ici†σzτzci.H_N=N\sum_i\chi_i c_i^\dagger\sigma_z c_i=N\sum_i c_i^\dagger\sigma_z\tau_z c_i.

Here cic_i is a two-component spinor. The sublattice Pauli matrix τz\tau_z is introduced only after projection onto the explicit A/BA/B basis, with τz=+1\tau_z=+1 on AA and −1-1 on BB. Pure parent-lattice inversion n→−n\mathbf n\to-\mathbf n preserves χ\chi and does not exchange AA and BB. The sublattice-exchanging operation denoted by P\mathcal P is instead realized by the affine map n↦eν−n\mathbf n\mapsto\mathbf e_\nu-\mathbf n, where eν\mathbf e_\nu is a parent-lattice basis vector. It acts in physical space as inversion about Seν/2\mathcal S\mathbf e_\nu/2 and maps the acceptance window to a common transverse internal-space translate. Section Section D proves that this preserves the bulk frequencies of magnetically labeled hopping patterns for the regular window specified here. It does not require an arbitrary finite patch to coincide with its image.

Local environments

The first-neighbor coordination number ranges from z=3z = 3 to z=7z = 7. Accounting also for bond orientation yields eight vertex classes, traditionally denoted DD, QQ, KK, SS, S5, JJ, TT, VV [58, 60, 61]. Each class corresponds to a subregion of W⊥W_{\perp}, and its thermodynamic frequency is the induced measure of that subregion divided by the total occupied-window measure. For the four-layer Penrose construction, this is a sum of polygonal areas over the occupied layers, not an unrestricted integral through the three-dimensional convex hull. The weights used in the two-step enumeration are

ρQ=−στ3,ρD=−στ,ρK=σ2τ3,ρS=σ25 τ3,ρS5=σ45 τ3,ρJ=σ2τ,ρT=σ4τ3,ρV=−σ3τ3.(19)\begin{aligned} \rho_Q &= -\frac{\sigma}{\tau^3}, & \rho_D &= -\frac{\sigma}{\tau}, & \rho_K &= \frac{\sigma^2}{\tau^3}, \\ \rho_S &= \frac{\sigma^2}{\sqrt{5}\,\tau^3}, & \rho_{S5} &= \frac{\sigma^4}{\sqrt{5}\,\tau^3}, & \rho_J &= \frac{\sigma^2}{\tau}, \\ \rho_T &= \frac{\sigma^4}{\tau^3}, & \rho_V &= -\frac{\sigma^3}{\tau^3}. \tag*{(19)} \end{aligned}

The construction and perpendicular-space subwindows are shown in Figure 4. All rotated versions of a local environment belong to the same class.

Figure 4. Cut-and-project construction and local classification of the antiferromagnetic Penrose tiling. The illustrative patch has 301 vertices. The physical and perpendicular projections and the eight local environments illustrate the construction.

High-frequency Floquet expansion

Fourier components of the Peierls hopping

For a directed nearest-neighbor bond j→ij \to i, let rij=a(cos⁡θij,sin⁡θij)\mathbf{r}_{ij} = a(\cos\theta_{ij}, \sin\theta_{ij}). Here θij\theta_{ij} is measured counterclockwise from the positive xx axis to ri−rj\mathbf{r}_i - \mathbf{r}_j. The orbital coupling to a spatially uniform light field is included by the Peierls phase exp⁡[i(e/ℏ)A(t)⋅rij]\exp[i(e/\hbar)\mathbf{A}(t)\cdot\mathbf{r}_{ij}], with e>0e > 0. For CPL, A(t)=A0(cos⁡ωt,ηsin⁡ωt)\mathbf{A}(t) = A_0(\cos\omega t, \eta\sin\omega t), with light helicity η=±1\eta= \pm1 for the fixed propagation direction used in the main text. Then

eℏA(t)⋅rij=α0[cos⁡ωtcos⁡θij+ηsin⁡ωtsin⁡θij]=α0cos⁡(ωt−ηθij),(20)\frac{e}{\hbar}\mathbf{A}(t)\cdot\mathbf{r}_{ij} = \alpha_0[\cos\omega t\cos\theta_{ij}+\eta\sin\omega t\sin\theta_{ij}] = \alpha_0\cos(\omega t-\eta\theta_{ij}), \tag*{(20)}

where α0=eA0a/ℏ\alpha_0 = eA_0a/\hbar. The Jacobi–Anger identity gives

eiα0cos⁡(ωt−ηθ)=∑n=−∞∞inJn(α0)e−inωteinηθ.e^{i\alpha_0\cos(\omega t-\eta\theta)} = \sum_{n=-\infty}^{\infty} i^n J_n(\alpha_0)e^{-in\omega t}e^{in\eta\theta}.

Here JnJ_n is the Bessel function of the first kind of integer order nn, including the zeroth-order function J0J_0. With H(t)=∑nHne−inωtH(t) = \sum_n H_n e^{-in\omega t}, the Fourier components are

Hn=1TL∫0TLdt H(t)einωt,H_n = \frac{1}{T_{\mathrm{L}}}\int_0^{T_{\mathrm{L}}} \mathrm{d}t\, H(t)e^{in\omega t},

where TL=2π/ωT_{\mathrm{L}} = 2\pi/\omega. For n≠0n \ne0, only hopping contributes, and therefore

Hn=−t1inJn(α0)∑⟨ij⟩einηθijci†cj.(21)H_n = -t_1 i^n J_n(\alpha_0) \sum_{\langle ij\rangle} e^{in\eta\theta_{ij}} c_i^\dagger c_j. \tag*{(21)}

The time average H0H_0 renormalizes the real hopping by J0(α0)J_0(\alpha_0) and leaves the staggered exchange unchanged. The sum contains both bond orientations, so H−n=Hn†H_{-n} = H_n^\dagger. Using the opposite Fourier convention exchanges the order of the commutator in the effective Hamiltonian. Keeping these conventions together is essential for a consistent helicity sign.

Commutator and induced same-sublattice hopping

To first order in 1/ω1/\omega,

HF=H0+∑n>0[H−n,Hn]nℏω+O(ω−2).H_{\mathrm{F}}=H_0+\sum_{n>0}\frac{[H_{-n},H_n]}{n\hbar\omega}+O(\omega^{-2}).

This is the van Vleck Hamiltonian [62]. To the retained order the evolution has the form U(t,t0)=e−iK(t)e−iHF(t−t0)/ℏeiK(t0)U(t,t_0)=e^{-iK(t)}e^{-iH_{\mathrm{F}}(t-t_0)/\hbar}e^{iK(t_0)}, with a periodic kick K(t)K(t). A stroboscopic Hamiltonian at a specified time origin is unitarily related to this operator; the kick matters for spectral weights. Terms in the commutator sharing an intermediate site generate i→j→ki\to j\to k. Combining a path with its Hermitian conjugate yields

HFi−j−k=iηΛ(δ)(ci†ck−ck†ci),(22)H_{\mathrm{F}}^{i-j-k}=i\eta\Lambda(\delta)(c_i^\dagger c_k-c_k^\dagger c_i), \tag*{(22)}
Λ(δ)=2t12ℏω∑n>0Jn2(α0)nsin⁡(nδ).(23)\Lambda(\delta)=\frac{2t_1^2}{\hbar\omega}\sum_{n>0}\frac{J_n^2(\alpha_0)}{n}\sin(n\delta). \tag*{(23)}

Equation (B3) is the effective hopping between endpoints ii and kk mediated by the intermediate site jj. Its angle δ=θkj−θij\delta=\theta_{kj}-\theta_{ij} is the oriented difference between the two bonds pointing outward from jj. Each nearest-neighbor hop changes magnetic sublattice, so the two endpoints belong to the same sublattice. Equation (B3) has the following properties:

  1. it is imaginary and antisymmetric under i↔ki\leftrightarrow k;

  1. it changes sign when the helicity η\eta is reversed;

  1. it vanishes for a collinear two-step path;

  1. it is odd under reversal of the oriented turning angle;

  1. its leading weak-field scale is t12α02/(ℏω)t_1^2\alpha_0^2/(\hbar\omega).

The same basic cooperation between a spin-independent orbital term and staggered exchange appears in crystalline Floquet antiferromagnets [29, 30]. Lin and Vila formulate a related mechanism in terms of opposite sublattice currents: the orbital sector produces a momentum-dependent sublattice imbalance that exchange converts into spin splitting [36]. Our calculation fixes the exchange field and derives the imaginary hopping from the external drive. A complex hopping coefficient is not itself an expectation value of a bond-current operator; neither spontaneous loop-current order nor a steady current distribution is calculated here. The specifically quasicrystalline input is the coherent sum of geometrically allowed paths.

Figure 5. Enumeration of Floquet-induced hopping. (a) Same-sublattice hopping in the illustrative 301-site patch. (b) Orientation convention and rhombus cancellation. (c) Remaining paths around each vertex class, with BB–BB hopping shown as an example.

If ii and kk are opposite vertices of a Penrose rhombus, there are two intermediate vertices j1j_1, j2j_2. The two paths enclose opposite oriented areas and have opposite turning angles. Their amplitudes in Eq. (22) cancel. Removing these diagonal pairs and enumerating the remaining paths around the eight vertex classes gives 0, 0, 2, 5, 5, 5, 8, and 12 directed processes around DD, QQ, KK, SS, S5S5, JJ, TT, VV, respectively [60, 61].

Finite-tiling spectra and numerical validation

We construct the finite real-space effective Hamiltonian before applying the local-environment projection. Plane-wave overlaps resolve its spectral weight at continuous probe momentum, while a circular angular decomposition characterizes the spin contrast. For Fig 2 of the main text, we use parameters: Nsite=1387N_{\mathrm{site}} = 1387, t1=N=1t_1 = N = 1 eV, ω=20\omega= 20 rad/fs, α0=1.5\alpha_0 = 1.5, and Γ=0.01\Gamma= 0.01 eV. The fixed-energy map and its angular analysis are evaluated at ε=−1.20\varepsilon= -1.20 eV. The Hamiltonian assembled below retains terms through order 1/ω1/\omega.

Effective Hamiltonian and path bookkeeping

For each spin block, assemble

HF,s=sN∑iχi∣i⟩⟨i∣−t1J0(α0)∑⟨ij⟩∣i⟩⟨j∣+∑j∑{i,k}⊂N(j)iηΛ(θkj−θij)(∣i⟩⟨k∣−∣k⟩⟨i∣).(24)H_{\mathrm{F},s}=sN\sum_i\chi_i|i\rangle\langle i|-t_1J_0(\alpha_0)\sum_{\langle ij\rangle}|i\rangle\langle j|+\sum_j\sum_{\{i,k\}\subset\mathcal{N}(j)}i\eta\Lambda(\theta_{kj}-\theta_{ij})(|i\rangle\langle k|-|k\rangle\langle i|). \tag*{(24)}

Here HF,sH_{\mathrm{F},s} is the spin-ss block of the real-space effective Hamiltonian HFH_{\mathrm{F}} in the main text. N(j)\mathcal{N}(j) is the nearest-neighbor list of jj. The edge sum contains both orientations, whereas the last sum contains each unordered pair of distinct neighbors only once. Choose either endpoint ordering for that pair and use it consistently in both the angle and the operator. Exchanging i,ki,k reverses both signs and leaves the contribution unchanged. This convention avoids an extra factor of two. Contributions from different intermediate sites are accumulated on the same endpoint matrix element; rhombus cancellations then occur algebraically, not by removing arbitrary endpoint pairs.

The Bessel sum can be truncated at nmax⁡n_{\max}. Its accuracy is characterized by the change in matrix elements and spectra upon increasing nmax⁡n_{\max}. Direct path assembly and the Fourier-commutator construction give equivalent matrices when the same bond orientations and cutoffs are used. Hermiticity is measured by ∥H−H†∥F/∥H∥F\lVert H-H^\dagger\rVert_{\mathrm{F}}/\lVert H\rVert_{\mathrm{F}}, with ∥⋅∥F\lVert\cdot\rVert_{\mathrm{F}} the Frobenius norm. Retaining only n=1n=1 gives the leading weak-drive approximation. For the amplitude of α0=1.5\alpha_0=1.5 in the main text, this truncation is set as nmax⁡=8n_{\max}=8.

Plane-wave projection, normalization, and spectral cuts

With real radial weights wiw_i, define a normalized probe state

∣k,s;w⟩=1∑iwi2∑iwieik⋅ri∣i,s⟩.(25)|k,s;w\rangle=\frac{1}{\sqrt{\sum_i w_i^2}}\sum_iw_i e^{i k\cdot r_i}|i,s\rangle. \tag*{(25)}

The main text uses wi=1w_i=1. A smooth mask can suppress the outer boundary. The mask affects boundary sensitivity and angular resolution; a symmetry-breaking mask can produce forbidden angular components.

The spectral function is

As(k,ϵ)=−1πIm⁡⟨k,s;w∣(ϵ+iΓ−HF,s)−1∣k,s;w⟩(26)A_s(k,\epsilon)=-\frac{1}{\pi}\operatorname{Im}\langle k,s;w|(\epsilon+i\Gamma-H_{\mathrm{F},s})^{-1}|k,s;w\rangle \tag*{(26)}
=∑m∣⟨k,s;w∣m,s⟩∣2Γ/π(ϵ−Ems)2+Γ2.(27)=\sum_m|\langle k,s;w|m,s\rangle|^2\frac{\Gamma/\pi}{(\epsilon-E_{ms})^2+\Gamma^2}. \tag*{(27)}

For a complete eigenbasis, As≥0\mathcal{A}_s \ge0 and ∫−∞∞As dε=1\int_{-\infty}^{\infty} \mathcal{A}_s\,\mathrm{d}\varepsilon= 1 at each k\boldsymbol{k}. An integral on a finite numerical energy grid misses the Lorentzian tails; its truncation error must not be mistaken for loss of state normalization. The real-space eigenvalues themselves have no momentum label. Dispersive peaks arise from their momentum-dependent overlaps.

Using the same patch, mask, and broadening for both spins isolates the spin difference ΔA=A↑−A↓\Delta A = \mathcal{A}_{\uparrow} - \mathcal{A}_{\downarrow}. The spin sum identifies regions of appreciable spectral intensity. A large normalized polarization where this sum vanishes has little spectroscopic weight. The undriven spectrum provides a baseline for finite-boundary effects.

Complex angular harmonics and fifth-component weight

For a circle of fixed radius kk and fixed energy ε\varepsilon, sample θj=2πj/Nθ\theta_j = 2\pi j/N_\theta, j=0,…,Nθ−1j = 0,\ldots,N_\theta-1, and use

aℓ(k,ε)=1Nθ∑jΔA(k,θj,ε)e−iℓθj,(28)a_\ell(k,\varepsilon) = \frac{1}{N_\theta}\sum_j \Delta A(k,\theta_j,\varepsilon)e^{-i\ell\theta_j}, \tag*{(28)}
Pℓ(k,ε)=∣aℓ(k,ε)∣2.(29)P_\ell(k,\varepsilon) = \lvert a_\ell(k,\varepsilon)\rvert^2. \tag*{(29)}

This is the discrete form of the main-text definition. For real ΔA\Delta A, a−ℓ=aℓ∗a_{-\ell} = a_\ell^*. If real coefficients are desired, define cℓ=2Re⁡aℓc_\ell= 2\operatorname{Re}a_\ell and dℓ=−2Im⁡aℓd_\ell= -2\operatorname{Im}a_\ell for ℓ>0\ell> 0, so that ΔA=a0+∑ℓ>0(cℓcos⁡ℓθ+dℓsin⁡ℓθ)\Delta A = a_0 + \sum_{\ell>0}(c_\ell\cos\ell\theta+ d_\ell\sin\ell\theta). The amplitude Wℓ=cℓ2+dℓ2=2∣aℓ∣W_\ell= \sqrt{c_\ell^2+d_\ell^2} = 2\lvert a_\ell\rvert therefore obeys Pℓ=Wℓ2/4P_\ell= W_\ell^2/4. This fixes the factor-of-two convention that differs from some real sine/cosine decompositions. For a pure Fsin⁡5θF\sin5\theta, a5=−iF/2a_5 = -iF/2 and P5=F2/4P_5 = F^2/4.

Angular sampling with Nθ>2ℓmax⁡N_\theta> 2\ell_{\max} resolves harmonics through ℓmax⁡\ell_{\max}; refinement controls aliasing from higher components. A normalized fifth-harmonic fraction is

f5(k,ε)=P5∑ℓ=1ℓmax⁡Pℓ,(30)f_5(k,\varepsilon) = \frac{P_5}{\sum_{\ell=1}^{\ell_{\max}}P_\ell}, \tag*{(30)}

defined when the denominator is nonzero. The absolute weight P5P_5 and the retained cutoff distinguish a strong signal from a large fraction of a weak signal. Even and symmetry-forbidden components diagnose boundary or discretization effects. Figure 2(c) in the main text compares the displayed angular components; the bulk symmetry also allows ℓ=15,25,…\ell= 15,25,\ldots, whose weights are not fixed by symmetry.

The complex coefficient aℓa_\ell, unlike PℓP_\ell, retains the sign and angular phase of the spin contrast. To state the helicity relation explicitly, write HF,s(η)H_{F,s}^{(\eta)} and As(η)\mathcal{A}_s^{(\eta)} for the effective Hamiltonian and spectrum at helicity η\eta. Complex conjugation in the site basis gives

KHF,s(η)K−1=HF,s(−η),As(−η)(k,ε)=As(η)(−k,ε).(31)K H_{\mathrm{F},s}^{(\eta)} K^{-1}=H_{\mathrm{F},s}^{(-\eta)},\qquad\mathcal{A}_{s}^{(-\eta)}(\mathbf{k},\varepsilon)=\mathcal{A}_{s}^{(\eta)}(-\mathbf{k},\varepsilon). \tag*{(31)}

The latter identity holds for a finite patch with the same real mask and broadening. Combining it with the bulk combined spin-flip–inversion relation in Sec. D gives

A‾s(−η)(k,ε)=A‾−s(η)(k,ε).(32)\overline{\mathcal{A}}_{s}^{(-\eta)}(\mathbf{k},\varepsilon)=\overline{\mathcal{A}}_{-s}^{(\eta)}(\mathbf{k},\varepsilon). \tag*{(32)}

Thus the bulk spin contrast and its complex angular coefficients reverse sign, while their squared weights are unchanged: aℓ(−η)=−aℓ(η)a_{\ell}^{(-\eta)}=-a_{\ell}^{(\eta)} and Pℓ(−η)=Pℓ(η)P_{\ell}^{(-\eta)}=P_{\ell}^{(\eta)}. These statements concern the Floquet spectrum; finite boundaries can modify the bulk relations.

Extracting an energy splitting and testing the radial law

A fixed-energy ΔA\Delta\mathcal{A} is not an energy difference. For example, for two identical line shapes displaced by Δ/2\Delta/2 in opposite directions, ΔA≃−Δ ∂εAref\Delta\mathcal{A}\simeq-\Delta\,\partial_{\varepsilon}\mathcal{A}_{\mathrm{ref}}, where Aref\mathcal{A}_{\mathrm{ref}} is the common unsplit line shape, to first order. Changes of spectral weight add further terms. Thus a spectral harmonic can change sign with energy even when the underlying splitting does not.

An energy splitting is extracted by tracking matched, resolvable opposite-spin peaks continuously in kk and θ\theta. Denoting the difference by Δλpeak\Delta_{\lambda}^{\mathrm{peak}}, its fifth harmonic is

aλ,5peak(k)=12π∫02πdθ Δλpeak(k,θ)e−5iθ.(33)a_{\lambda,5}^{\mathrm{peak}}(k)=\frac{1}{2\pi}\int_{0}^{2\pi}d\theta\,\Delta_{\lambda}^{\mathrm{peak}}(k,\theta)e^{-5i\theta}. \tag*{(33)}

In the coordinate convention of Sec. F, the analytic prediction is −iKλk5/2-iK_{\lambda}k^{5/2}, where Kλ=−2ληNg5/N2+16t12J02(α0)K_{\lambda}=-2\lambda\eta N g_{5}/\sqrt{N^{2}+16t_{1}^{2}J_{0}^{2}(\alpha_{0})}. The phase distinguishes orientation and sign, while the modulus measures the splitting amplitude. A leading radial correction has the form Ck5(1+bk2)Ck^{5}(1+bk^{2}). Overlapping or unresolved peaks do not define an unambiguous branch splitting; this energy-based diagnostic is separate from the fixed-energy intensity weights shown in main-text Figure 2(c), and those weights alone do not establish a numerical radial k5k^{5} law.

Controls and finite-size convergence

The microscopic mechanism has three null limits: no illumination, no staggered exchange, and omission of the chiral commutator. In each, the bulk spin splitting vanishes. Reversing circular helicity changes the sign of the chiral hopping. For a monochromatic linearly polarized drive, the leading chiral commutator vanishes because there is no oriented optical area.

This control should not be generalized to arbitrary combinations of linearly polarized fields. Yu considers orthogonal components at frequencies ω\omega and 2ω2\omega, with their relative phase chosen to generate a distinct orbital symmetry breaking [80]. For the time-even waveform used there, Hn=H−nH_n = H_{-n}, so the order-1/ω1/\omega commutators vanish; an odd-parity contribution nevertheless arises from the cycle-averaged Hamiltonian in an initially even-parity altermagnet. The resulting splitting has mixed parity. This differs both in starting state and in Floquet order from our CPL-induced splitting of a spin-degenerate antiferromagnet.

Finite-sample departures from the bulk symmetry relations can be measured on a common momentum and energy grid by

rP=∥ΔA(k)+ΔA(−k)∥22∥ΔA∥2,(34)r_P = \frac{\lVert\Delta A(\mathbf{k}) + \Delta A(-\mathbf{k}) \rVert_2}{2\lVert\Delta A \rVert_2}, \tag*{(34)}
r5=∥ΔA(R2π/5k)−ΔA(k)∥22∥ΔA∥2.(35)r_5 = \frac{\lVert\Delta A(R_{2\pi/5}\mathbf{k}) - \Delta A(\mathbf{k}) \rVert_2}{2\lVert\Delta A \rVert_2}. \tag*{(35)}

Here ∥⋅∥2\lVert\cdot\rVert_2 is the discrete norm on that grid; absolute residuals are used when the denominator vanishes. An exact finite-patch relation requires a symmetry of the full labeled Hamiltonian and of the boundary mask, not only of the uncolored vertex set. For the bulk construction proved in Sec. D, residuals can be monitored as the patch size or phason ensemble is increased at fixed broadening. Such convergence is not necessarily monotonic for individual cuts. These diagnostics test the numerical realization of the symmetry. In addition, sample radius, boundary mask, broadening, harmonic cutoff, and momentum/energy resolution also control the finite-system result.

Symmetry origin of odd-parity spin splitting

Spin-group analyses of collinear odd-parity magnets emphasize that the orbital Hamiltonian must be included in the symmetry assessment, not only the ordered spin pattern [29, 34]. In particular, a spin-preserving antiunitary operation can be lost through complex hopping even when the local moments remain collinear. Odd parity of the spin splitting does not assert physical time-reversal invariance of that magnetic state. We apply this principle below to the full effective Hamiltonian, using quasicrystalline indistinguishability rather than importing a periodic spin space group.

Indistinguishability and phase functions

Because a quasicrystal is not invariant under a finite set of primitive translations, its symmetry cannot be defined by exact pointwise coincidence with a periodically repeated unit cell. We write the spin density in the spin-space-group framework [65, 70] as

S(r)=∑k∈LS(k)eik⋅r,(36)S(\mathbf{r}) = \sum_{\mathbf{k}\in\mathcal{L}} S(\mathbf{k})e^{i\mathbf{k}\cdot\mathbf{r}}, \tag*{(36)}

where the Fourier module L\mathcal{L} contains all integral linear combinations of the independent diffraction basis vectors. Two spin configurations are indistinguishable when they contain the same spatial distribution of every bounded local pattern. In Fourier space, each symmetry or antisymmetry operation [γ∥g][\gamma\Vert g] in the spin point group GSG_S is then associated with a gauge phase function Φgγ(k)\Phi_g^\gamma(\mathbf{k}) such that

γS(gk)=e2πiΦgγ(k)S(k),[γ∥g]∈GS.(37)\gamma S(g\mathbf{k}) = e^{2\pi i\Phi_g^\gamma(\mathbf{k})}S(\mathbf{k}), \qquad[\gamma\Vert g]\in G_S. \tag*{(37)}

The phase allows the transformed structure to differ by a phason-space translation while remaining physically indistinguishable. It is the appropriate replacement for the equality condition used in a periodic crystal.

For the bulk spectral constraints, rotations about the Néel axis form the spin-only subgroup [SO(2)∥E][SO(2)\Vert E]. We call C2⊥sC_{2\perp}^{s} a spin flip: it is a π\pi spin rotation about an axis perpendicular to the Néel vector, exchanging the two conserved-spin sectors, rather than inversion of an arbitrary spin vector. The notation P\mathcal{P} denotes the spatial operation constructed below. Their combination [C2⊥s∥P][C_{2\perp}^{s}\Vert\mathcal{P}] is termed the combined spin-flip–inversion relation throughout this section. In contrast, [Es∥C5z][E_s\Vert C_{5z}] preserves each spin sector. The spatial operations relate indistinguishable bulk configurations, not necessarily a finite sample to itself. Independently, the undriven real-hopping Hamiltonian has the spin-preserving antiunitary operation

Θs=C2⊥sT\Theta_s = C_{2\perp}^{s}\mathcal{T}

Taking C2⊥s=−iσxC_{2\perp}^{s}=-i\sigma_x and T=iσyK\mathcal{T}=i\sigma_y K, with KK complex conjugation, gives Θs=iσzK\Theta_s=i\sigma_z K. It preserves the spin label, sends the plane-wave probe momentum to −k-\mathbf{k}, and satisfies Θs2=+1\Theta_s^2=+1, so it does not itself enforce Kramers degeneracy. Combining its spectral relation with the bulk spin-flip–inversion relation gives the same-momentum opposite-spin spectral equality derived below.

Hyperspace action of inversion and fivefold rotation

The sublattice label of a five-dimensional hyperlattice site is fixed by the parity of its integer coordinates,

n=∑ν=15nνeν∈{A,∑νnν∈2Z,B,∑νnν∈2Z+1.(38)\mathbf{n}=\sum_{\nu=1}^{5}n_{\nu}\mathbf{e}_{\nu}\in \begin{cases} A, & \sum_{\nu}n_{\nu}\in2\mathbb{Z},\\ B, & \sum_{\nu}n_{\nu}\in2\mathbb{Z}+1. \end{cases} \tag*{(38)}
An integer-lattice-compatible affine inversion

The four occupied layers L1L_1, L2L_2, L3L_3, L4L_4 have q=∑ini=2,1,0,−1q=\sum_i n_i=2,1,0,-1, respectively. Write S⊥n=(y,q/2)S_{\perp}\mathbf{n}=(\mathbf{y},q/\sqrt{2}), with y\mathbf{y} the two transverse internal coordinates, and define their acceptance polygons by

Wq={y:(y,q/2)∈W⊥},q∈{−1,0,1,2}.(39)W_q=\{\mathbf{y}:(\mathbf{y},q/\sqrt{2})\in W_{\perp}\},\qquad q\in\{-1,0,1,2\}. \tag*{(39)}

The window vertices in Sec. A occur in opposite pairs about c⊥=(0,0,2/4)\mathbf{c}_{\perp}=(0,0,\sqrt{2}/4). Since the window is their convex hull,

2c⊥−W⊥=W⊥,W1−q=−Wq.(40)2\mathbf{c}_{\perp}-W_{\perp}=W_{\perp},\qquad W_{1-q}=-W_q. \tag*{(40)}

Thus the layer pairs L1↔L4L_1\leftrightarrow L_4 and L2↔L3L_2\leftrightarrow L_3 have inverted acceptance polygons, not merely equal total areas. The layers at the two axial tips of the polytope have zero area and do not contribute to bulk frequencies.

Central symmetry of a window alone does not construct an operation on the integer parent lattice. In particular, the linear map n↦−n\mathbf{n}\mapsto-\mathbf{n} preserves parity, whereas an affine map centered at the window center need not preserve Z5\mathbb{Z}^5. An explicit lattice-compatible choice is

Pν:n′=eν−n,r′=aν−r,aν=Seν,r⊥′=S⊥eν−r⊥,q′=1−q.(41)\begin{aligned} \mathcal{P}_{\nu}:&\quad\mathbf{n}'=\mathbf{e}_{\nu}-\mathbf{n},\\ &\quad\mathbf{r}'=\mathbf{a}_{\nu}-\mathbf{r},\qquad\mathbf{a}_{\nu}=S\mathbf{e}_{\nu},\\ &\quad\mathbf{r}'_{\perp}=S_{\perp}\mathbf{e}_{\nu}-\mathbf{r}_{\perp},\qquad q'=1-q. \tag*{(41)} \end{aligned}

It is an involution on Z5\mathbb{Z}^{5} and satisfies χ(n′)=−χ(n)\chi(n^\prime)=-\chi(n). Its physical action is inversion about aν/2\mathbf{a}_{\nu}/2, including the physical translation that accompanies the parent-lattice shift. Different choices of ν\nu give equivalent bulk relations; we suppress this choice in the symbol P\mathcal{P}.

Define bν\mathbf{b}_{\nu} by

(bν,0)=S⊥eν−2c⊥.(42)(\mathbf{b}_{\nu},0)=S_{\perp}\mathbf{e}_{\nu}-2\mathbf{c}_{\perp}. \tag*{(42)}

Its third component vanishes because the last row of S⊥S_{\perp} gives (S⊥eν)3=1/2(S_{\perp}\mathbf{e}_{\nu})_{3}=1/\sqrt{2}. Equations (D5)–(D7) imply

y′=bν−y,W1−q′=bν−Wq=W1−q+bν.(43)\mathbf{y}^{\prime}=\mathbf{b}_{\nu}-\mathbf{y},\qquad W^{\prime}_{1-q}=\mathbf{b}_{\nu}-W_{q}=W_{1-q}+\mathbf{b}_{\nu}. \tag*{(43)}

Consequently, Pν\mathcal{P}_{\nu} maps the selected tiling to one with all four acceptance polygons translated by the same transverse internal vector. It generally does not return the same vertex set pointwise. The remaining step is to establish that this common translation preserves the statistics of all labeled local patterns.

Acceptance-domain proof of bulk indistinguishability

We use the regular cut-and-project construction: the polygon boundaries have zero area, and boundary choices do not affect bulk frequencies. Within each layer, the parent points form a coset of {d∈Z5:∑νdν=0}\{\mathbf{d}\in\mathbb{Z}^{5}:\sum_{\nu}d_{\nu}=0\}. Uniform distribution of their internal coordinates converts patch frequencies into acceptance-domain areas, with a common density factor for these cosets [81].

For a finite set F\mathcal{F} of required occupied parent offsets, including d=0\mathbf{d}=0, put qd=∑νdνq_{d}=\sum_{\nu}d_{\nu} and d⊥=(S⊥d)1,2\mathbf{d}_{\perp}=(S_{\perp}\mathbf{d})_{1,2}. The allowed internal coordinates of an anchor in layer qq form

Ωq,F=⋂d∈F(Wq+qd−d⊥).(44)\Omega_{q,\mathcal{F}}=\bigcap_{\mathbf{d}\in\mathcal{F}}(W_{q+q_{d}}-\mathbf{d}_{\perp}). \tag*{(44)}

where an unoccupied layer has an empty window. Specifying an exact local pattern additionally excludes offsets that must be absent; these exclusions are represented by set differences of the same translated polygons. For Wq(v)=Wq+vW_{q}^{(\mathbf{v})}=W_{q}+\mathbf{v}, both intersections and differences obey

Ωq,F(v)=Ωq,F+v,∣Ωq,F(v)∣=∣Ωq,F∣.(45)\Omega_{q,\mathcal{F}}^{(\mathbf{v})}=\Omega_{q,\mathcal{F}}+\mathbf{v},\qquad\left|\Omega_{q,\mathcal{F}}^{(\mathbf{v})}\right|=\left|\Omega_{q,\mathcal{F}}\right|. \tag*{(45)}

The same identity holds for domains including absent offsets. More explicitly, the frequency per vertex of a specified magnetically labeled pattern p\mathbf{p} is

freq⁡(p)=∑q∈Qp∣Ωq,p∣∑q=−12∣Wq∣,(46)\operatorname{freq}(\mathbf{p}) = \frac{\sum_{q\in Q_{\mathbf{p}}} |\Omega_{q,\mathbf{p}}|}{\sum_{q=-1}^{2} |W_q|}, \tag*{(46)}

where QpQ_{\mathbf{p}} contains the anchor layers compatible with its magnetic labels and Ωq,p\Omega_{q,\mathbf{p}} includes all required occupancies and exclusions. The common coset density cancels between numerator and denominator. Equation (45) preserves both, so a common transverse translation preserves the frequency of every finite pattern, not only the populations of AA and BB.

Under Eq. (41), an offset d\boldsymbol{d} becomes −d-\boldsymbol{d}, and an anchor layer qq becomes 1−q1-q. The exchange field at the image site is restored by the independent spin flip, because χ′=−χ\chi'=-\chi and C2⊥sσz(C2⊥s)†=−σzC^{\mathrm{s}}_{2\perp}\sigma_z(C^{\mathrm{s}}_{2\perp})^\dagger=-\sigma_z. Applying Eq. (45) to Eq. (43) therefore establishes indistinguishability of the complete labeled nearest-neighbor magnetic network under the combined operation. This conclusion uses the stated regular window and its parity coloring; it is not inferred from the layer permutation alone, and it need not hold for independently modified magnetic acceptance domains.

Complex virtual hopping and the spectral consequence

For a virtual path i−j−ki-j-k, affine inversion reverses both outward bond vectors without reversing time or complex conjugating amplitudes. Hence

θij′=θij+π,θkj′=θkj+π(mod2π),Λ(θkj′−θij′)=Λ(θkj−θij).(47)\theta'_{ij}=\theta_{ij}+\pi,\qquad\theta'_{kj}=\theta_{kj}+\pi\pmod{2\pi},\qquad\Lambda(\theta'_{kj}-\theta'_{ij})=\Lambda(\theta_{kj}-\theta_{ij}). \tag*{(47)}

The coefficient iηΛi\eta\Lambda is therefore unchanged at fixed light helicity. The intermediate vertex is mapped together with the endpoints, so the acceptance domain contains the same transformed path constraints. All paths contributing to an endpoint pair are added coherently; their multiplicities and cancellations are carried into the image network. Because the nearest-neighbor and retained virtual-hopping amplitudes are deterministic local functions of this geometry, the acceptance-domain result establishes equal bulk frequencies for the full magnetically labeled hopping network of the order-1/ω1/\omega Hamiltonian, including its complex weights.

Finally, r′=aν−r\boldsymbol{r}'=\boldsymbol{a}_{\nu}-\boldsymbol{r} takes a plane-wave probe at k\boldsymbol{k} to one at −k-\boldsymbol{k}, up to the common phase eik⋅aνe^{i\boldsymbol{k}\cdot\boldsymbol{a}_{\nu}}. This phase cancels from spectral weights. The combined operation therefore gives

A‾s(k,ϵ)=A‾−s(−k,ϵ),\overline{A}_s(\boldsymbol{k},\epsilon)=\overline{A}_{-s}(-\boldsymbol{k},\epsilon),

as derived from spectral moments below. This is a bulk statistical relation for the canonical regular construction, not a claim that the inversion center aν/2\boldsymbol{a}_{\nu}/2 is an exact symmetry center of every finite realization.

Fivefold rotation

The noncrystallographic fivefold rotation is a crystallographic permutation in the five-dimensional parent lattice:

C5z: (e1,e2,e3,e4,e5)→(e2,e3,e4,e5,e1),(n1,n2,n3,n4,n5)→(n5,n1,n2,n3,n4).(48)C_{5z}:\ (e_1,e_2,e_3,e_4,e_5)\to(e_2,e_3,e_4,e_5,e_1), \qquad (n_1,n_2,n_3,n_4,n_5)\to(n_5,n_1,n_2,n_3,n_4). \tag*{(48)}

It leaves ∑ini\sum_i n_i invariant and does not exchange AA and BB.

Undriven and driven bulk spectral constraints

Undriven bulk spectral equality

Let XX denote a magnetically labeled tiling together with its hopping amplitudes, and gXgX its image under the combined spin-flip–inversion operation. Equations (D8)–(D12) establish that XX and gXgX have the same frequencies of every finite labeled hopping pattern for the regular Penrose construction specified here. We now derive the electronic spectral consequence of that result. Write A‾s\overline{A}_s for the thermodynamic vertex-averaged spectrum or the corresponding hull-averaged spectrum. For a different window or magnetic decoration, the labeled-network equivalence must be checked anew before applying the same argument.

The connection between pattern statistics and this electronic observable can be seen from the plane-wave spectral moments. Write HsH_s for the relevant single-particle spin block, either undriven or the retained driven HF,sH_{\mathrm{F},s}. For finite-range, bounded hopping they are

mp,s(k)=lim⁡Nsite→∞1Nsite∑ijeik⋅(rj−ri)(Hsp)ij.(49)m_{p,s}(\mathbf{k})=\lim_{N_{\mathrm{site}}\to\infty}\frac{1}{N_{\mathrm{site}}}\sum_{ij}e^{i\mathbf{k}\cdot(\mathbf{r}_j-\mathbf{r}_i)}(H_s^p)_{ij}. \tag*{(49)}

For each finite integer p≥0p\ge0, a term is a finite hopping walk, including on-site exchange insertions. Its displacement, magnetic labels, and amplitude are determined by a bounded local pattern. Equal frequencies of all such patterns therefore give the transformed moment relation, and hence the corresponding spectral-measure relation. Equivalently, the resolvent can be reconstructed from these moments for bounded HsH_s. This justifies the bulk constraint

A‾s(k,ϵ)=A‾−s(−k,ϵ).(50)\overline{A}_s(\mathbf{k},\epsilon)=\overline{A}_{-s}(-\mathbf{k},\epsilon). \tag*{(50)}

The broadening Γ>0\Gamma>0 in the spectral function smooths the same spectral measure.

Without light, HsH_s is real symmetric. A real eigenbasis gives equal squared plane-wave overlaps at kk and −k-k, so

As(k,ε)=As(−k,ε).(51)\mathcal{A}_s(\mathbf{k},\varepsilon)=\mathcal{A}_s(-\mathbf{k},\varepsilon). \tag*{(51)}

Unlike Eq. (50), this relation holds for any finite undriven patch with real hoppings and a real spatial mask. Combining the two relations gives

A‾↑(k,ε)=A‾↓(k,ε).(52)\overline{\mathcal{A}}_{\uparrow}(\mathbf{k},\varepsilon)=\overline{\mathcal{A}}_{\downarrow}(\mathbf{k},\varepsilon). \tag*{(52)}

This is equality of the opposite-spin bulk spectral measures at the same probe momentum, not a claim of exact Kramers pairing in every finite realization.

Pointwise symmetry and an independent bipartite check

A conventional real-space PT\mathcal{PT} symmetry requires a site permutation pp and a physical inversion center r0\mathbf{r}_0 such that

rp(i)=2r0−ri,χp(i)=−χi,tp(i)p(j)=tij\mathbf{r}_{p(i)}=2\mathbf{r}_0-\mathbf{r}_i,\qquad \chi_{p(i)}=-\chi_i,\qquad t_{p(i)p(j)}=t_{ij}

for the real undriven hoppings, with the boundary also mapped to itself. Only under these additional conditions does PT\mathcal{PT} act within the same finite Hamiltonian. For a real involutory inversion permutation and spin-1/21/2 time reversal, (PT)2=−1(\mathcal{PT})^2=-1, yielding exact eigenstate pairing. The representation iσyτxKi\sigma_y\tau_xK has this property in the two-sublattice quasi-band model, but is not a construction of the required site permutation for a finite Penrose patch.

The particular nearest-neighbor model also admits an algebraic check that is independent of inversion. In the A/BA/B basis,

Hs=(sNIAMM†−sNIB)(53)H_s= \begin{pmatrix} sN I_A & M\\ M^\dagger& -sN I_B \end{pmatrix} \tag*{(53)}

where MM is the intersublattice hopping matrix. Each nonzero singular value vμv_\mu of MM produces a two-dimensional block

Hs(μ)=(sNvμvμ−sN),Eμ,±=±N2+vμ2.(54)H_s^{(\mu)}= \begin{pmatrix} sN & v_\mu\\ v_\mu& -sN \end{pmatrix}, \qquad E_{\mu,\pm}=\pm\sqrt{N^2+v_\mu^2}. \tag*{(54)}

These paired energies are spin-independent. Null states confined to AA have energy sNsN, and those confined to BB have energy −sN-sN; unequal sublattice populations can therefore leave unmatched finite-sample levels. Moreover, eigenvalue pairing does not imply equal plane-wave spectral weights. Equation (D17), rather than energy pairing alone, is the statement relevant to the bulk spectral spin contrast.

Driven bulk constraints

For the monochromatic vector potential of the main text, the polarization rotation is characterized by the dimensionless axial vector

κL=A(t)×A˙(t)ωA02=ηz^.(55)\boldsymbol{\kappa}_{L}=\frac{\mathbf{A}(t)\times\dot{\mathbf{A}}(t)}{\omega A_{0}^{2}}=\eta\hat{\mathbf{z}}. \tag*{(55)}

Here the dot denotes a time derivative, and the in-plane vector potential is embedded in three dimensions. This vector specifies the sense of rotation about the fixed sample normal. Under spatial inversion, both A\mathbf{A} and A˙\dot{\mathbf{A}} change sign, so their cross product is unchanged. Under time reversal, A(t)→−A(−t)\mathbf{A}(t)\to-\mathbf{A}(-t) and A˙(t)→A˙(−t)\dot{\mathbf{A}}(t)\to\dot{\mathbf{A}}(-t), so κL\boldsymbol{\kappa}_{L} changes sign. The fivefold rotation C5zC_{5z} preserves κL\boldsymbol{\kappa}_{L}. These are transformation properties of the polarization rotation in the spatially uniform orbital-coupling model.

The induced imaginary hopping has the same inversion-even, time-reversal-odd character. Equation (D12) verifies this inversion property for each virtual path, while the acceptance-domain proof includes the path’s occurrence frequency and intermediate-site constraint. Complex conjugation instead reverses the helicity-dependent imaginary hopping, so Eq. (51) is no longer a constraint at fixed light helicity. The combined spin-flip–inversion relation [C_{2\perp}\parallel\mathcal{P}] and the spin-preserving fivefold relation remain bulk constraints for the specified regular construction. The former enforces

ΔA‾(k,ϵ)=−ΔA‾(−k,ϵ),Eλs(k)=Eλ,−s(−k).(56)\Delta\overline{\mathcal{A}}(\mathbf{k},\epsilon)=-\Delta\overline{\mathcal{A}}(-\mathbf{k},\epsilon),\qquad E_{\lambda s}(\mathbf{k})=E_{\lambda,-s}(-\mathbf{k}). \tag*{(56)}

Here ΔA‾=A‾↑−A‾↓\Delta\overline{\mathcal{A}}=\overline{\mathcal{A}}_{\uparrow}-\overline{\mathcal{A}}_{\downarrow} is a bulk momentum-resolved electronic spin signal. The loss of the real-Hamiltonian relation removes the additional step that gave Eq. (52). A nonzero spin contrast is therefore allowed at generic k\mathbf{k}, while Eq. (56) fixes its odd parity. The energies in that equation denote matched bulk spectral or quasi-band branches.

Δλ(−k)=−Δλ(k),Δλ=Eλ,+−Eλ,−.\Delta_{\lambda}(-\mathbf{k})=-\Delta_{\lambda}(\mathbf{k}),\qquad\Delta_{\lambda}=E_{\lambda,+}-E_{\lambda,-}.

As a separate consistency check, spatial inversion in the reduced quasi-band model exchanges A/BA/B and sends k→−k\mathbf{k}\to-\mathbf{k}; its internal sublattice factor is τx\tau_x. The translation associated with its physical inversion center contributes only a common phase, which drops out of the Hamiltonian conjugation. Up to an overall spinor phase, the combined spin-flip–inversion factor is Uinv=σxτxU_{\mathrm{inv}}=\sigma_x\tau_x. Since B5(−k)=B5(k)B_5(-\mathbf{k})=B_5(\mathbf{k}) and D5(−k)=−D5(k)D_5(-\mathbf{k})=-D_5(\mathbf{k}), it satisfies

UinvH(0)(k)Uinv†=H(0)(−k).(57)U_{\mathrm{inv}}\mathcal{H}^{(0)}(\mathbf{k})U_{\mathrm{inv}}^\dagger=\mathcal{H}^{(0)}(-\mathbf{k}). \tag*{(57)}

This reproduces Eq. (56) in the projected theory. Spin–orbit coupling mixes the conserved-spin sectors and requires a separate material-specific symmetry analysis. Odd parity of the spectrum does not establish physical T\mathcal{T} symmetry of a state with fixed helicity and Néel order [72]. For a finite patch that does not implement the relevant operations exactly, these spectral constraints should be tested with size or phason averaging rather than assumed for every boundary.

The spatial rotation [Es∥C5z][E_s\parallel C_{5z}] leaves the spin sector unchanged. Combining it with [C2⊥s∥P][C_{2\perp}^{s}\parallel\mathcal{P}] yields

A‾s(R2π/5k,ε)=A‾s(k,ε),(58)\overline{A}_s(R_{2\pi/5}\mathbf{k},\varepsilon)=\overline{A}_s(\mathbf{k},\varepsilon), \tag*{(58)}
A‾s(Rπ/5k,ε)=A‾−s(k,ε).(59)\overline{A}_s(R_{\pi/5}\mathbf{k},\varepsilon)=\overline{A}_{-s}(\mathbf{k},\varepsilon). \tag*{(59)}

The second relation follows because Rπ/5=−R−4π/5R_{\pi/5}=-R_{-4\pi/5}, with R−4π/5R_{-4\pi/5} a fivefold rotation. Thus the spin sum is unchanged by 36∘36^\circ, while the spin difference reverses sign. This is the unitary spin-group operation [C2⊥s∥C10z][C_{2\perp}^{s}\parallel C_{10z}]; it need not be identified with physical C5zTC_{5z}\mathcal{T} or C10zTC_{10z}\mathcal{T} symmetry of the driven Hamiltonian. Expanding either the bulk spin contrast ΔA‾\overline{\Delta A} or the matched-branch splitting Δλ\Delta\lambda in eiℓθe^{i\ell\theta}, fivefold rotation requires ℓ∈5Z\ell\in5\mathbb{Z}, and odd parity requires odd ℓ\ell. The allowed indices are therefore ℓ=±5,±15,±25,…\ell=\pm5,\pm15,\pm25,\ldots The leading permitted angular form is hh-wave, while the relative weights at finite momentum are determined by the spectrum.

Successive acceptance-window partitions and quasi-band theory

The acceptance window selects the physical sites through their perpendicular-space coordinates. Partitioning this window according to local patches is an established geometric construction: regions associated with a given patch are called acceptance domains [74]. We use this geometric information to build progressively more detailed projected electronic Hamiltonians, in the spirit of earlier quasicrystal effective-model approaches [73]. Our approximation is specified by the partition rule, the associated basis states, and the projected hopping matrix elements, which we define below.

Window partitions and their statistical weights

Let W0W_0 be an acceptance window. A hyperlattice displacement d\mathbf{d} has projections r=Sd\mathbf{r}=S\mathbf{d} and r⊥=S⊥d\mathbf{r}_{\perp}=S_{\perp}\mathbf{d}. For a site with internal coordinate y\mathbf{y}, the shell signature is

Qm(y)={d∈Zdh:∥d∥1=m, y+S⊥d∈W0}.(60)Q_m(\mathbf{y})=\left\{\mathbf{d}\in\mathbb{Z}^{d_h}:\|\mathbf{d}\|_1=m,\ \mathbf{y}+S_{\perp}\mathbf{d}\in W_0\right\}. \tag*{(60)}

To make a nested hierarchy precise, retain the tuple (Q1,…,Ql)(Q_1,\ldots,Q_l) through order ll, or equivalently refine the previous partition by the new shell boundaries. Sites with the same retained signature belong to the same subwindow WljW_l^j. Disconnected regions may share a label if their signatures coincide. This definition records membership at allowed offsets; it does not require a connected hopping path to every offset.

Here dh=5d_h=5 for Penrose and 44 for Ammann–Beenker geometry. The shell index mm is distinct from the layer index qq in Sec. D, and partition order ll is distinct from the angular harmonic ℓ\ell. In this general window construction, y\mathbf{y} denotes the full internal coordinate, including its layer for Penrose; the earlier y∈Wq\mathbf{y}\in W_q described only its transverse components.

Let μj=μ(Wlj)\mu_j=\mu(W_l^j). The measure μ(W)\mu(W) is the appropriate geometric window measure and μ0=μ(W0)\mu_0=\mu(W_0); it is unrelated to the chemical potential μ\mu in Sec. G. Uniform sampling of a regular cut-and-project window gives

pl,j=lim⁡Nl,jNsite=μjμ0,∑jpl,j=1.(61)p_{l,j}=\lim\frac{N_{l,j}}{N_{\mathrm{site}}}=\frac{\mu_j}{\mu_0},\qquad\sum_j p_{l,j}=1. \tag*{(61)}

For Ammann–Beenker tiling, μ\mu is area in the two-dimensional perpendicular plane. For the layered Penrose construction it is the induced measure on the occupied layers described above. Magnetic parities are kept as distinct labels at every order: the zeroth-order Penrose magnetic theory has one window for each of AA and BB, even before local-environment refinement. The geometric order ll is independent of the order in 1/ω1/\omega of the Floquet expansion.

Confined plane-wave states and the Gram matrix

For an arbitrary occupied window WW, define

∣k,W,s⟩=1NW∑yi∈Weik⋅ri∣i,s⟩.(62)\lvert k, W, s\rangle= \frac{1}{\sqrt{N_W}} \sum_{y_i\in W} e^{i\mathbf{k}\cdot\mathbf{r}_i}\lvert i,s\rangle. \tag*{(62)}

At the same k\mathbf{k}, disjoint windows have exactly zero overlap. More generally, their thermodynamic overlap is

⟨k,W1,s∣k,W2,s′⟩=δss′μ(W1∩W2)μ(W1)μ(W2).(63)\langle\mathbf{k},W_1,s\mid\mathbf{k},W_2,s'\rangle= \delta_{ss'}\frac{\mu(W_1\cap W_2)}{\sqrt{\mu(W_1)\mu(W_2)}}. \tag*{(63)}

For a disjoint window partition, this identity supplies an orthonormal finite-dimensional basis at each k\mathbf{k}; overlapping windows instead require an overlap matrix. It does not imply orthogonality between different momenta:

G(k,j),(p,b)=∑i∈Wlj∩Wlbei(p−k)⋅riNl,jNl,b.(64)G_{(\mathbf{k},j),(\mathbf{p},b)} = \frac{\sum_{i\in W_l^j\cap W_l^b}e^{i(\mathbf{p}-\mathbf{k})\cdot\mathbf{r}_i}}{\sqrt{N_{l,j}N_{l,b}}}. \tag*{(64)}

Bragg-related components can give nonzero off-diagonal Gram entries. We use each fixed-k\mathbf{k} subspace as a projected quasi-band model. If several momenta are combined, their Gram matrix must be retained and the generalized eigenproblem solved on its non-null subspace. Increasing ll alone is not a proof of cross-momentum orthogonality or completeness in the thermodynamic limit.

The Fourier module labels diffraction vectors. A finite-sample plane-wave spectral function, in contrast, can be evaluated for any continuous k∈R2\mathbf{k}\in\mathbb{R}^2. A projected hyperlattice momentum sampling is a computational choice rather than a restriction on the photoelectron momentum. We use a finite decagonal region for Penrose plots, called the first pseudo-Brillouin zone (1PBZ) in the source figures. It does not tile reciprocal space as an exact primitive cell, and spectral information outside it remains physical.

Projected hopping and the zeroth-order Patterson limit

Write a directed hopping as Hr=tr∑i∣i+r⟩⟨i∣H_{\mathbf{r}}=t_{\mathbf{r}}\sum_i\lvert i+\mathbf{r}\rangle\langle i\rvert, restricted to occupied endpoints. A source coordinate y∈Wlb\mathbf{y}\in W_l^b must obey y+r⊥∈Wla\mathbf{y}+\mathbf{r}_{\perp}\in W_l^a. With the phases in Eq. (62),

Hab(l)(k)=∑rtre−ik⋅rμ[Wla∩(Wlb+r⊥)]μaμb.(65)\mathcal{H}^{(l)}_{ab}(\mathbf{k})=\sum_{\mathbf{r}}t_{\mathbf{r}}e^{-i\mathbf{k}\cdot\mathbf{r}}\frac{\mu\left[W_l^a\cap\left(W_l^b+\mathbf{r}_{\perp}\right)\right]}{\sqrt{\mu_a\mu_b}}. \tag*{(65)}

The phase follows directly from e−ik⋅(ri+r)eik⋅rie^{-i\mathbf{k}\cdot(\mathbf{r}_i+\mathbf{r})}e^{i\mathbf{k}\cdot\mathbf{r}_i}. Since t−r=tr∗t_{-\mathbf{r}}=t_{\mathbf{r}}^* and overlap measures are translation invariant, Hba(l)=Hab(l)∗\mathcal{H}^{(l)}_{ba}=\mathcal{H}^{(l)*}_{ab}. The on-site magnetic term is sNχaδabsN\chi_a\delta_{ab}.

In the Floquet problem a hopping amplitude can depend on the intermediate vertex, not only on its endpoint displacement. The general matrix element is therefore

Hab,s(l)(k)=sNχaδab+∑re−ik⋅rμaμb∫Wlb∩(Wla−r⊥)dμ(y) tr(y).(66)\mathcal{H}^{(l)}_{ab,s}(\mathbf{k})=sN\chi_a\delta_{ab}+\sum_{\mathbf{r}}\frac{e^{-i\mathbf{k}\cdot\mathbf{r}}}{\sqrt{\mu_a\mu_b}}\int_{W_l^b\cap(W_l^a-\mathbf{r}_{\perp})}d\mu(\mathbf{y})\,t_{\mathbf{r}}(\mathbf{y}). \tag*{(66)}

Here tr(y)t_{\mathbf{r}}(\mathbf{y}) sums all allowed paths with that displacement. For two successive steps d1,d2\mathbf{d}_1,\mathbf{d}_2, the integration domain also requires y+S⊥d1∈W0\mathbf{y}+S_{\perp}\mathbf{d}_1\in W_0. Their complex amplitudes are accumulated before diagonalization. This preserves the rhombus cancellation automatically. Refining by path signatures makes these amplitudes piecewise constant.

For a single nonmagnetic window, Eq. (65) reduces to

H(0)(k)=∑rtre−ik⋅rΠW(r),ΠW(r)=μ[W0∩(W0+r⊥)]μ0.(67)\mathcal{H}^{(0)}(\mathbf{k})=\sum_{\mathbf{r}}t_{\mathbf{r}}e^{-i\mathbf{k}\cdot\mathbf{r}}\Pi_W(\mathbf{r}),\qquad\Pi_W(\mathbf{r})=\frac{\mu[W_0\cap(W_0+\mathbf{r}_{\perp})]}{\mu_0}. \tag*{(67)}

Here ΠW\Pi_W denotes the normalized window-overlap (Patterson) weight. For a pair of partition regions, μa=μ(Wla)\mu_a=\mu(W_l^a) and μb=μ(Wlb)\mu_b=\mu(W_l^b). For two magnetic windows the normalization is μαμβ\sqrt{\mu_\alpha\mu_\beta}. A conditional bond frequency Nβα(r)/NαN_{\beta\alpha}(\mathbf{r})/N_\alpha equals the normalized matrix weight only for equal sublattice populations; otherwise the factor Nα/Nβ\sqrt{N_\alpha/N_\beta} must be included.

Spectral weights and quasicrystal benchmarks

If H(l)(k)uf=Efuf\mathcal{H}^{(l)}(\mathbf{k})u_f=E_fu_f, the unpartitioned plane wave satisfies

∣k,W0⟩=∑jpl,j∣k,Wlj⟩.(68)|\mathbf{k},W_0\rangle=\sum_j\sqrt{p_{l,j}}|\mathbf{k},W_l^j\rangle. \tag*{(68)}

Consequently, its projected spectral function is

A(l)(k,ϵ)=−1πIm⁡∑f∣∑jpl,juf,j∣2ϵ−Ef+iΓ.(69)\mathcal{A}^{(l)}(\mathbf{k},\epsilon)=-\frac{1}{\pi}\operatorname{Im}\sum_f\frac{\left|\sum_j\sqrt{p_{l,j}}u_{f,j}\right|^2}{\epsilon-E_f+i\Gamma}. \tag*{(69)}

There is one sum over bands ff: the window components combine coherently inside the modulus. Completeness of the eigenvectors of the projected matrix gives a total weight of one. For a spinful model the same construction is performed in each conserved-spin block.

The Ammann–Beenker (AB) benchmark uses Dl=1,41,185,329D_l=1,41,185,329 distinct subwindows for orders l=0,1,2,3l=0,1,2,3, as shown in Figure 6. These partition sizes refer to AB geometry.

Figure 6. Successive acceptance-window partitions in the AB benchmark. The diagram shows the unpartitioned octagon and the boundaries at orders 1, 2, 3, with reported region counts 41, 185, 329. These data illustrate the quasi-band method on an octagonal geometry.

Figure 7 compares their spectra to exact diagonalization of a 1433-site AB cluster along Γ−M1−M2−Γ\Gamma- M_{1} - M_{2} - \Gamma. The zeroth order captures a broad dispersion; refinement produces additional branches and weight redistribution closer to the finite-system spectrum. Figure 8 further shows the spectral function distribution at ε=−0.8 eV\varepsilon= -0.8\ \mathrm{eV} in the 2D momentum plane for orders l=0,1,2l = 0, 1, 2. In both figures, the refinement resolves additional local-environment features of the octagonal AB quasicrystal.

Figure 7. AB spectral benchmark. From top to bottom: window orders 0, 1, 2, 3, followed by exact diagonalization for Nsite=1433N_{\mathrm{site}} = 1433. The path is indicated in the inset. The vertical symbol ω\omega in the original artwork denotes spectral energy in the benchmark units, not the CPL frequency of the Penrose model.

AB spectral benchmark with spectral function distribution in 2D momentum plane

Figure 8. AB spectral benchmark with spectral function distribution in 2D momentum plane. From left to right: window orders l=0,1,2l = 0, 1, 2, followed by exact diagonalization for Nsite=897N_{\mathrm{site}} = 897.

In addition, we examine the selection window partition hierarchy in Penrose quasicrystals. Figure 9 compares the spin-difference spectra to exact diagonalization of a 1387-site Penrose quasicrystal cluster. With increasing order ll, refinement produces additional peaks and weight redistribution closer to the exact cluster spectrum.

Penrose spectral benchmark with spin-difference spectral function distribution in 2D momentum plane

Figure 9. Penrose spectral benchmark with spin-difference spectral function distribution in 2D momentum plane. From left to right: window orders l=0,1,2l = 0, 1, 2, followed by exact diagonalization for Nsite=1387N_{\mathrm{site}} = 1387.

Fivefold structure factors

Let aj=a(cos⁡2πj/5,sin⁡2πj/5)\boldsymbol{a}_j = a(\cos2\pi j/5,\sin2\pi j/5), with indices understood modulo five. The nearest-neighbor part is summarized by

C5(k)=15∑j=04cos⁡(k⋅aj),B5(k)=4t1J0(α0)C5(k).(70)C_5(\boldsymbol{k}) = \frac{1}{5}\sum_{j=0}^{4}\cos(\boldsymbol{k}\cdot\boldsymbol{a}_j), \qquad B_5(\boldsymbol{k}) = 4t_1J_0(\alpha_0)C_5(\boldsymbol{k}). \tag*{(70)}

For a turning angle δ\delta, the two-step displacement vectors form a fivefold star of length Lδ=2asin⁡(δ/2)L_\delta= 2a\sin(\delta/2). We define

S5(k;δ)=15∑j=04sin⁡[2asin⁡(δ2)k⋅uj],S_5(\boldsymbol{k};\delta) = \frac{1}{5}\sum_{j=0}^{4}\sin\left[2a\sin\left(\frac{\delta}{2}\right)\boldsymbol{k}\cdot\boldsymbol{u}_j\right],

where

uj=(cos⁡((4j+1)π10),sin⁡((4j+1)π10))\boldsymbol{u}_j = \left(\cos\left(\frac{(4j+1)\pi}{10}\right),\sin\left(\frac{(4j+1)\pi}{10}\right)\right)

for the orientation convention of the main text. Summing the vertex-class contributions with Eq. (19) gives

D5(k)=(55−11)∑δ=2π/5, 4π/5Λ(δ)S5(k;δ),(71)D_5(\boldsymbol{k}) = (5\sqrt{5}-11)\sum_{\delta=2\pi/5,\,4\pi/5}\Lambda(\delta)S_5(\boldsymbol{k};\delta), \tag*{(71)}
I(δ)=2Λ(δ)=4t12ℏω∑n=1∞Jn2(α0)nsin⁡(nδ).(72)I(\delta)=2\Lambda(\delta)=\frac{4t_1^2}{\hbar\omega}\sum_{n=1}^{\infty}\frac{J_n^2(\alpha_0)}{n}\sin(n\delta). \tag*{(72)}

One can check the geometric coefficient directly. With X=I(2π/5)S5(k;2π/5)X=I(2\pi/5)S_5(\mathbf{k};2\pi/5) and Y=I(4π/5)S5(k;4π/5)Y=I(4\pi/5)S_5(\mathbf{k};4\pi/5), the contributions to the BB–BB block at η=+1\eta=+1 are

HBBD,Q=0,HBBK=2ρKY,HBBS=5ρSY,HBBS5=−5ρS5Y,HBBJ=ρJY,HBBT=ρT(X−5Y),HBBV=ρV(X−5Y).(73)\begin{aligned} H_{BB}^{D,Q}&=0, & H_{BB}^{K}&=2\rho_KY,\\ H_{BB}^{S}&=5\rho_SY, & H_{BB}^{S5}&=-5\rho_{S5}Y, & H_{BB}^{J}&=\rho_JY,\\ H_{BB}^{T}&=\rho_T(X-5Y), & H_{BB}^{V}&=\rho_V(X-5Y). \tag*{(73)} \end{aligned}

Equation (A3) gives ρT+ρV=ξ\rho_T+\rho_V=\xi and 2ρK+5ρS−5ρS5+ρJ−5ρT−5ρV=ξ2\rho_K+5\rho_S-5\rho_{S5}+\rho_J-5\rho_T-5\rho_V=\xi. Here ξ=(55−11)/2\xi=(5\sqrt{5}-11)/2, and I(δ)=2Λ(δ)I(\delta)=2\Lambda(\delta), with Λ\Lambda defined by the virtual hopping Eq. (3) in the main text. Thus HBB=ηD5H_{BB}=\eta D_5 and HAA=−ηD5H_{AA}=-\eta D_5. Here D5D_5, Λ\Lambda, and II are helicity independent; the only helicity factor is the explicit η\eta. The resulting quasi-band Hamiltonian is

H(0)(k)=Nσzτz−B5(k)τx−ηD5(k)τz.(74)\mathcal{H}^{(0)}(\mathbf{k})=N\sigma_z\tau_z-B_5(\mathbf{k})\tau_x-\eta D_5(\mathbf{k})\tau_z. \tag*{(74)}

Its eigenvalues are

Eλs(k)=λB52(k)+[sN−ηD5(k)]2.E_{\lambda s}(\mathbf{k})=\lambda\sqrt{B_5^2(\mathbf{k})+[sN-\eta D_5(\mathbf{k})]^2}.

Here s=+1s=+1 and −1-1 correspond to ↑\uparrow and ↓\downarrow, while λ=+1\lambda=+1 and −1-1 label the upper and lower quasi-bands independently of spin.

Long-wavelength hh-wave expansion

The angular selection rule follows most transparently from complex coordinates k±=kx±ikyk_\pm=k_x\pm ik_y. Under a rotation by 2π/52\pi/5, k±→e±i2π/5k±k_\pm\to e^{\pm i2\pi/5}k_\pm. An odd scalar harmonic invariant under fivefold rotations must have total odd degree and angular momentum equal to a multiple of five. Degrees one and three cannot satisfy this condition. At degree five the two real basis functions are

Re⁡k+5=kx5−10kx3ky2+5kxky4=k5cos⁡5θ,(75)\operatorname{Re} k_+^5=k_x^5-10k_x^3k_y^2+5k_xk_y^4=k^5\cos5\theta, \tag*{(75)}
Im⁡k+5=5kx4ky−10kx2ky3+ky5=k5sin⁡5θ.(76)\operatorname{Im} k_+^5=5k_x^4k_y-10k_x^2k_y^3+k_y^5=k^5\sin5\theta. \tag*{(76)}

Here k=k(cos⁡θ,sin⁡θ)\mathbf{k}=k(\cos\theta,\sin\theta); θ\theta is the momentum polar angle, not a directed real-space bond angle θij\theta_{ij}.

The same result follows by expanding Eq. (71). For a fivefold star uj\mathbf{u}_j,

∑j(k⋅uj)p=0\sum_j(\mathbf{k}\cdot\mathbf{u}_j)^p=0

for odd p=1,3p = 1,3, while the p=5p = 5 sum is proportional to cos⁡5(θ−θ0)\cos5(\theta-\theta_0). Hence

D5(k)=g5(α0,ω)k5sin⁡(5θ)+O(k7).(77)D_5(\mathbf{k}) = g_5(\alpha_0,\omega)k^5\sin(5\theta) + O(k^7). \tag*{(77)}

For the bond stars defined above, θ0=π/10\theta_0=\pi/10, so cos⁡[5(θ−θ0)]=sin⁡(5θ)\cos[5(\theta-\theta_0)]=\sin(5\theta). This fifth harmonic changes sign under a 36∘36^\circ rotation. More explicitly, the elementary identity cos⁡5x=(10cos⁡x+5cos⁡3x+cos⁡5x)/16\cos^5x=(10\cos x+5\cos3x+\cos5x)/16 gives

15∑jsin⁡(Lδk⋅uj)=Lδ5k51920cos⁡[5(θ−θ0)]+O(k7).(78)\frac{1}{5}\sum_j\sin(L_\delta\mathbf{k}\cdot\mathbf{u}_j)=\frac{L_\delta^5k^5}{1920}\cos[5(\theta-\theta_0)]+O(k^7). \tag*{(78)}

The positive fifth-order sign follows from the Taylor series of sine. For the stars defined above, θ0=π/10\theta_0=\pi/10, and therefore

g5=55−111920∑δ=2π/5, 4π/5Λ(δ)Lδ5.(79)g_5=\frac{5\sqrt{5}-11}{1920}\sum_{\delta=2\pi/5,\,4\pi/5}\Lambda(\delta)L_\delta^5. \tag*{(79)}

For weak fields, J1(α0)≃α0/2J_1(\alpha_0)\simeq\alpha_0/2, so

g5∝t12α05ℏω.g_5\propto\frac{t_1^2\alpha_0^5}{\hbar\omega}.

Symmetry permits this coefficient but does not require it to be nonzero at every drive amplitude: Bessel-weighted contributions can cancel at special parameters. Allowed angular indices are ℓ=5,15,25,…\ell=5,15,25,\ldots; radial corrections to ℓ=5\ell=5 already enter at degree seven. A 36∘36^\circ rotation maps a fivefold star onto its negative, giving D5(Rπ/5k)=−D5(k)D_5(R_{\pi/5}\mathbf{k})=-D_5(\mathbf{k}), while C5C_5 is unchanged. Thus Eλs(Rπ/5k)=Eλ,−s(k)E_{\lambda s}(R_{\pi/5}\mathbf{k})=E_{\lambda,-s}(\mathbf{k}).

Writing E0(k)=N2+B52(k)E_0(\mathbf{k})=\sqrt{N^2+B_5^2(\mathbf{k})}, expansion of the matched-branch difference Δλ=Eλ,↑−Eλ,↓\Delta_\lambda=E_{\lambda,\uparrow}-E_{\lambda,\downarrow} for ∣D5∣≪E0\lvert D_5\rvert\ll E_0 gives

Δλ(k)=−2ληND5(k)E0(k)+O(D53/E02).(80)\Delta_\lambda(\mathbf{k})=-\frac{2\lambda\eta N D_5(\mathbf{k})}{E_0(\mathbf{k})}+O(D_5^3/E_0^2). \tag*{(80)}

This explicitly demonstrates the cooperative nature of the effect: the nonrelativistic spin splitting vanishes if either NN or the light-induced D5D_5 term is absent.

Since C5(0)=1C_5(0)=1 and C5(k)=1+O(k2)C_5(\mathbf{k})=1+O(k^2), substituting Eq. (77) into Eq. (80) gives, in the θ0=π/10\theta_0=\pi/10 convention,

Δλ(k)=−2ληNg5N2+16t12J02(α0)k5sin⁡(5θ)+O(k7).(81)\Delta_\lambda(\mathbf{k})=-\frac{2\lambda\eta N g_5}{\sqrt{N^2+16t_1^2J_0^2(\alpha_0)}}k^5\sin(5\theta)+O(k^7). \tag*{(81)}

Corrections to the even denominator enter at degree seven, whereas the next odd power of D5D_5 in the energy difference starts at degree fifteen. This expression assumes an analytic, nondegenerate spin-independent branch near the origin; the finite exchange N≠0N\ne0 ensures a nonzero denominator in the present model.

Time-dependent pump and spin-resolved trARPES

We use a finite-duration vector potential

Apump(t)=A0e−t2/Tpump2[cos⁡ωt,ηsin⁡ωt].\mathbf{A}_{\mathrm{pump}}(t) = A_0 e^{-t^2/T_{\mathrm{pump}}^2} [\cos\omega t, \eta\sin\omega t].

The pump maximum defines t=0t=0, as in the main text. The probe envelope centered at delay tdt_d is

spr(t,td)=e−(t−td)2/Tprobe2.s_{\mathrm{pr}}(t,t_d) = e^{-(t-t_d)^2/T_{\mathrm{probe}}^2}.

For independent electrons initially in thermal equilibrium, let Hinit∣n⟩=ϵn∣n⟩H_{\mathrm{init}}|n\rangle= \epsilon_n|n\rangle, where HinitH_{\mathrm{init}} is the single-particle Hamiltonian specifying the preparation at time t0t_0. The initial one-particle correlation matrix is

C0=f(Hinit)=∑nf(ϵn)∣n⟩⟨n∣,f(E)=[e(E−μ)/(kBT)+1]−1.(82)C_0 = f(H_{\mathrm{init}}) = \sum_n f(\epsilon_n)|n\rangle\langle n|,\qquad f(E) = \left[e^{(E-\mu)/(k_{\mathrm{B}}T)}+1\right]^{-1}. \tag*{(82)}

The matrix function applies the Fermi distribution to the eigenvalues, not to individual matrix elements. In a site–spin basis, (C0)ab=⟨cb†(t0)ca(t0)⟩(C_0)_{ab}=\langle c_b^\dagger(t_0)c_a(t_0)\rangle. For Hinit=Vdiag⁡(ϵn)V†H_{\mathrm{init}}=V\operatorname{diag}(\epsilon_n)V^\dagger, it is evaluated as C0=Vdiag⁡[f(ϵn)]V†C_0=V\operatorname{diag}[f(\epsilon_n)]V^\dagger. At zero temperature it is the projector onto states below μ\mu, with occupations at ϵn=μ\epsilon_n=\mu specified by the chosen limiting prescription. Unlike the many-body thermal density operator ρ^0=Z−1e−(H^init−μN^)/(kBT)\hat{\rho}_0=Z^{-1}e^{-(\hat{H}_{\mathrm{init}}-\mu\hat{N})/(k_{\mathrm{B}}T)}, whose trace is one, C0C_0 has trace equal to the mean particle number.

The single-particle propagator satisfies iℏ∂tU(t,t0)=H(t)U(t,t0)i\hbar\partial_tU(t,t_0)=H(t)U(t,t_0) and U(t0,t0)=IU(t_0,t_0)=I, where H(t)H(t) includes the pump. The Heisenberg operators obey c(t)=U(t,t0)c(t0)\mathbf{c}(t)=U(t,t_0)\mathbf{c}(t_0). Substitution into Gab<(t1,t2)=i⟨cb†(t2)ca(t1)⟩G^{<}_{ab}(t_1,t_2)=i\langle c_b^\dagger(t_2)c_a(t_1)\rangle gives

G<(t1,t2)=iU(t1,t0)f(Hinit)U†(t2,t0),G^{<}(t_1,t_2)=iU(t_1,t_0)f(H_{\mathrm{init}})U^\dagger(t_2,t_0),
Gab<(t1,t2)=i∑nf(ϵn)ψn,a(t1)ψn,b∗(t2).(83)G^{<}_{ab}(t_1,t_2)=i\sum_n f(\epsilon_n)\psi_{n,a}(t_1)\psi^*_{n,b}(t_2). \tag*{(83)}

The first line is the main-text propagation formula, with C0=f(Hinit)C_0=f(H_{\mathrm{init}}). Here ∣ψn(t)⟩=U(t,t0)∣n⟩|\psi_n(t)\rangle=U(t,t_0)|n\rangle. Each initial state retains its statistical weight while its wavefunction evolves. For a field-free equilibrium preparation, HinitH_{\mathrm{init}} is the undriven Hamiltonian; it need not coincide with H(t0)H(t_0) if the pump is already appreciable at the chosen starting time. Here and in the propagation formulas, H(t)H(t), HinitH_{\mathrm{init}}, and the evolution operators act in single-particle space. The matrix H(t)H(t) is obtained from the coefficients of the quadratic Hamiltonian in the main text, with the finite pump substituted for the monochromatic field.

The spin-resolved photoemission intensity, neglecting matrix-element variations, is

Is(k,ε,td)∝Im⁡∫dt1dt2 spr(t1,td)spr(t2,td)eiε(t1−t2)/ℏGss<(k;t1,t2).(84)I_s(\mathbf{k},\varepsilon,t_d) \propto\operatorname{Im}\int dt_1dt_2\,s_{\mathrm{pr}}(t_1,t_d)s_{\mathrm{pr}}(t_2,t_d)e^{i\varepsilon(t_1-t_2)/\hbar}G^{<}_{ss}(\mathbf{k};t_1,t_2). \tag*{(84)}

An equivalent positive expression convenient for implementation is

Is(k,ε,td)∝∑nf(ϵn)∣∫dt spr(t,td)eiεt/ℏ⟨k,s∣U(t,t0)∣n⟩∣2.(85)I_s(\mathbf{k},\varepsilon,t_d) \propto\sum_n f(\epsilon_n)\left|\int dt\,s_{\mathrm{pr}}(t,t_d)e^{i\varepsilon t/\hbar}\langle\mathbf{k},s|U(t,t_0)|n\rangle\right|^2. \tag*{(85)}

This follows by inserting the occupied-state expansion of G<G^{<}; it supplies a positivity check on the double-time calculation. The time integral in Eq. (85) is a probe-filtered amplitude for each initially occupied state. Squaring it gives that state’s photoemission contribution, and summing with the weights f(ϵn)f(\epsilon_n) gives the total current. In the absence of the pump, U(t,t0)∣n⟩=e−iϵn(t−t0)/ℏ∣n⟩U(t,t_0)|n\rangle=e^{-i\epsilon_n(t-t_0)/\hbar}|n\rangle, so the current reduces to occupied plane-wave spectral weights broadened by the Fourier transform of the probe envelope. During the pump, the evolving wavefunctions carry both the changing electronic structure and the coherent time dependence. This is why the finite-pulse photocurrent need not coincide with the static van Vleck spectral function.

The initial-state factors f(ϵn)f(\epsilon_n) do not amount to assigning a Fermi distribution to Floquet quasienergies. Occupations here are propagated unitarily from the specified initial density matrix. By contrast, Li, Shao, and Kovalev determine driven steady-state spin responses with explicit bath coupling [31]. Their spin-conserving bosonic-bath and fermionic-reservoir settings give different spin-generation and transport behavior. Consequently, a spin-polarized photoemission pattern in our closed-system protocol does not establish a steady net spin accumulation or spin current. Such claims would require a specified relaxation model and a separate calculation of the corresponding observable.

A midpoint propagation step is

U(tj+1,tj)≃exp⁡[−iH((tj+1+tj)/2)δt/ℏ].U(t_{j+1},t_j)\simeq\exp\left[-iH\left((t_{j+1}+t_j)/2\right)\delta t/\hbar\right].

Refinement of δt=tj+1−tj\delta t=t_{j+1}-t_j controls the time-discretization error; U†U=IU^\dagger U=I provides an independent unitarity check. Equation (85) evaluates the photocurrent through probe-filtered amplitudes of initially occupied states and is manifestly nonnegative.

Parameters and initialization

The main-text photoemission map uses

QuantityValue
t1t_1, NN1 eV, 1 eV
α0\alpha_01.5
ω\omega20 Prad/s = 20 rad/fs
ℏ\hbar0.6582 eV fs
ℏω\hbar\omega, 2π/ω2\pi/\omega13.164 eV, 0.31416 fs
TpumpT_{\mathrm{pump}}, TprobeT_{\mathrm{probe}}150 fs, 10 fs
tdt_d0
TT, μ\mu0, 0
Energy of the momentum map−1.20 eV

Table 1.

The probe is centered at the pump maximum, td=0t_d = 0. Initial thermal occupations are set by HinitH_{\mathrm{init}} and are subsequently propagated, rather than rethermalized at each time. A field-free preparation and an equilibrium preparation in a finite initial vector potential are physically different protocols. The optical period 2π/ω=0.314162\pi/\omega= 0.31416 fs sets the rapid timescale of the propagation.

Drive regime and limitations

The van Vleck Hamiltonian is retained through order 1/ω1/\omega, with corrections of order ω−2\omega^{-2} omitted [62]. Its high-frequency regime requires

t1ℏω≪1,Nℏω≪1,(86)\frac{t_1}{\hbar\omega} \ll1,\qquad\frac{N}{\hbar\omega} \ll1, \tag*{(86)}

together with an off-resonant separation of the photon energy from the electronic spectrum relevant to the driven model. The local ratios in Eq. (86) are useful scale estimates, but do not by themselves bound the truncation error: coordination and the full bandwidth also enter. For the photoemission parameters, ℏω=13.164\hbar\omega= 13.164 eV and both ratios are approximately 0.0760. Quantitative accuracy of the effective spectrum is assessed by comparison with higher-order terms or a converged Floquet calculation, rather than by these ratios alone. No weak-field expansion in α0\alpha_0 is required when the Bessel sums are retained. At fixed α0\alpha_0, increasing $\omega improves the high-frequency hierarchy but reduces the leading virtual-hopping scale t12/(ℏω)t_1^2/(\hbar\omega), and hence the spin splitting.

The pulsed trARPES calculation instead propagates the time-dependent Hamiltonian and includes micromotion within the specified independent-electron model; it does not replace the pulse by a static HFH_{\mathrm{F}}. For the chosen envelopes, ωTpump=3000\omega T_{\mathrm{pump}} = 3000 and ωTprobe=200\omega T_{\mathrm{probe}} = 200, so both envelopes vary slowly over an optical cycle. This separation supports comparison with the effective spectrum near the pump maximum, but does not imply adiabatic electronic occupations or a thermal Floquet steady state. Time-step and initial-time convergence are separate from convergence of the high-frequency expansion.

The probe duration also sets an energy-resolution scale. For the Gaussian amplitude envelope used here, an isolated stationary level has the probe-limited intensity full width at half maximum

δεpr=22ln⁡2 ℏTprobe≃0.155 eV(Tprobe=10 fs).(87)\delta\varepsilon_{\mathrm{pr}} = \frac{2\sqrt{2\ln2}\,\hbar}{T_{\mathrm{probe}}} \simeq0.155\,\mathrm{eV}\qquad(T_{\mathrm{probe}} = 10\,\mathrm{fs}). \tag*{(87)}

This follows from the squared Fourier transform of the probe envelope. Pump-induced evolution, scattering, and instrumental resolution can further modify the line shape. A nonzero spin-difference intensity therefore need not imply two separately resolved opposite-spin peaks. The probe width is distinct from the Lorentzian broadening Γ\Gamma used in the static spectral function.

Converting the dimensionless drive to a physical field requires the edge length aa. The electric-field amplitude at the pump center is E0=ωA0\mathcal{E}_0 = \omega A_0, distinct from the quasi-band energy scale E0E_0, and satisfies

E0a=α0ℏωe=19.746 V(α0=1.5).(88)\mathcal{E}_0 a = \frac{\alpha_0\hbar\omega}{e} = 19.746\,\mathrm{V}\qquad(\alpha_0 = 1.5). \tag*{(88)}

The quoted parameters specify a model calculation, not an established experimental operating window. In a material, optical absorption, transitions outside the retained electronic subspace, and the stability of the magnetic moments must also be considered. Being off-resonant within the model does not establish negligible heating in the material.

The present treatment assumes conserved spin, independent electrons coupled to a fixed staggered exchange field, and no bath or self-consistent magnetic dynamics. It therefore addresses transient electronic spin splitting and photoemission contrast, rather than the formation or long-time stability of magnetic order. The coarsest quasi-band model additionally averages local environments, whereas the finite real-space calculations retain their individual hopping paths and diffraction structure. Refining the window partition addresses the former approximation; increasing the patch size and controlling its boundary address the latter finite-size effects. The bulk symmetry relations constrain the allowed angular harmonics as long as the assumed spin-group relations hold, but do not determine their relative weights, linewidths, or experimental visibility. Spin–orbit coupling, interaction-induced relaxation, and changes of magnetic order require a corresponding extension of the model and its symmetry analysis.

References

  1. [1]T. Jungwirth, R. M. Fernandes, E. Fradkin, A. H. MacDonald, J. Sinova, and L. Šmejkal, Altermagnetism: An unconventional spin-ordered phase of matter, Newton 1, 100162 (2025).arxiv.org/abs/2411.00717
  2. [2]C. Song, H. Bai, Z. Zhou, L. Han, H. Reichlova, J. H. Dil, J. Liu, X. Chen, and F. Pan, Altermagnets as a new class of functional materials, Nat. Rev. Mater. 10, 473 (2025).DOI
  3. [3]L. Bai, W. Feng, S. Liu, L. Smejkal, Y. Mokrousov, and Y. Yao, Altermagnetism: Exploring new frontiers in magnetism and spintronics, Adv. Funct. Mater. 34, 2409327 (2024).DOI
  4. [4]X.-J. Luo, D. Li, R.-C. Xiao, D.-F. Shao, L. Li, M. Tian, and Y. Yao, Unconventional magnetism: Symmetry classification, hybrid-parity and unconstrained-parity classes (2026), arXiv:2605.21336 [cond-mat.mtrl-sci].arXiv
  5. [5]L. Smejkal, R. González-Hernández, T. Jungwirth, and J. Sinova, Crystal time-reversal symmetry breaking and spontaneous hall effect in collinear antiferromagnets, Sci. Adv. 6, eaaz8809 (2020).DOI
  6. [6]H.-Y. Ma, M. Hu, N. Li, J. Liu, W. Yao, J.-F. Jia, and J. Liu, Multifunctional antiferromagnetic materials with giant piezomagnetism and noncollinear spin current, Nat. Commun. 12, 2846 (2021).DOI
  7. [7]L. Smejkal, J. Sinova, and T. Jungwirth, Beyond conventional ferromagnetism and antiferromagnetism: A phase with nonrelativistic spin and crystal rotation symmetry, Phys. Rev. X 12, 031042 (2022).DOI
  8. [8]L. Smejkal, J. Sinova, and T. Jungwirth, Emerging research landscape of altermagnetism, Phys. Rev. X 12, 040501 (2022).arxiv.org/abs/2204.10844
  9. [9]L. Smejkal, A. B. Hellenes, R. González-Hernández, J. Sinova, and T. Jungwirth, Giant and tunneling magnetoresistance in unconventional collinear antiferromagnets with nonrelativistic spin-momentum coupling, Phys. Rev. X 12, 011028 (2022).DOI
  10. [10]L. Smejkal, A. Marmodoro, K.-H. Ahn, R. González-Hernández, I. Turek, S. Mankovsky, H. Ebert, S. W. D’Souza, O. c. v. Šipr, J. Sinova, and T. c. v. Jungwirth, Chiral magnons in altermagnetic RuO₂, Phys. Rev. Lett. 131, 256703 (2023).arxiv.org/abs/2211.13806
  11. [11]S. Lee, S. Lee, S. Jung, J. Jung, D. Kim, Y. Lee, B. Seok, J. Kim, B. G. Park, L. Šmejkal, C.-J. Kang, and C. Kim, Broken kramers degeneracy in altermagnetic MnTe, Phys. Rev. Lett. 132, 036702 (2024).DOI
  12. [12]J. Krempaský, L. Šmejkal, S. W. D’Souza, M. Hajlaoui, G. Springholz, K. Uhlířová, F. Alarab, P. C. Constantinou, V. Strocov, D. Usanov, W. R. Pudelko, R. González-Hernández, A. Birk Hellenes, Z. Jansa, H. Reichlová, Z. Šobáň, R. D. Gonzalez Betancourt, P. Wadley, J. Sinova, D. Kriegner, J. Minár, J. H. Dil, and T. Jungwirth, Altermagnetic lifting of kramers spin degeneracy, Nature 626, 517 (2024).DOI
  13. [13]H. Reichlova, R. Lopes Seeger, R. González-Hernández, I. Kounta, R. Schlitz, D. Kriegner, P. Ritzinger, M. Lammel, M. Leiviskä, A. Birk Hellenes, K. Olejník, V. Petříček, P. Doležal, L. Horak, E. Schmoranzerova, A. Badura, S. Bertaina, A. Thomas, V. Baltz, L. Michez, J. Sinova, S. T. B. Goennenwein, T. Jungwirth, and L. Šmejkal, Observation of a spontaneous anomalous hall response in the Mn5Si3 d-wave altermagnet candidate, Nat. Commun. 15, 4961 (2024).DOI
  14. [14]M. Hu, X. Cheng, Z. Huang, and J. Liu, Catalog of c-paired spin-momentum locking in antiferromagnetic systems, Phys. Rev. X 15, 021083 (2025).DOI
  15. [15]A. B. Hellenes, T. Jungwirth, R. Jaeschke-Ubiergo, A. Chakraborty, J. Sinova, and L. Šmejkal, P-wave magnets (2024), arXiv:2309.01607 [cond-mat.mes-hall].arXiv
  16. [16]B. Brekke, P. Sukhachov, H. G. Giil, A. Brataas, and J. Linder, Minimal models and transport properties of unconventional p-wave magnets, Phys. Rev. Lett. 133, 236703 (2024).DOI
  17. [17]Y. Yu, M. B. Lyngby, T. Shishidou, M. Roig, A. Kreisel, M. Weinert, B. M. Andersen, and D. F. Agterberg, Odd-parity magnetism driven by antiferromagnetic exchange, Phys. Rev. Lett. 135, 046701 (2025).DOI
  18. [18]Q. Song, S. Stavrić, P. Barone, A. Droghetti, D. S. Antonenko, J. W. F. Venderbos, C. A. Occhialini, B. Ilyas, E. Ergeçen, N. Gedik, S.-W. Cheong, R. M. Fernandes, S. Picozzi, and R. Comin, Electrical switching of a p-wave magnet, Nature 642, 64 (2025).DOI
  19. [19]Yamada, R., Birch, M. T., Baral, P. R., Okumura, S., Nakano, R., Gao, S., Ezawa, M., Nomoto, T., Masell, J., Ishihara, Y., & al., . et . (2025). A metallic p-wave magnet with commensurate spin helix. Nature, 646, 837.
  20. [20]A. Chakraborty, A. Birk Hellenes, R. Jaeschke-Ubiergo, T. Jungwirth, L. Smejkal, and J. Sinova, Highly efficient non-relativistic edelstein effect in nodal p-wave magnets, Nat. Commun. 16, 7270 (2025).DOI
  21. [21]J.-K. Yuan, Z. Pan, and C. Wu, Unconventional magnetism in spin-orbit coupled systems, Phys. Rev. B 113, 014426 (2026).arxiv.org/abs/2504.14577
  22. [22]V. Leeb and J. Knolle, Collinear p-wave magnetism and hidden orbital ferrimagnetism, Phys. Rev. Lett. 137, 106701 (2026).DOI
  23. [23]G. Sim and S. Rachel, Quantum spin models of commensurate p-wave magnets, Phys. Rev. Lett. 137, 026503 (2026).DOI
  24. [24]K. Kudo and Y. Yanase, Odd-parity magnetism by quantum geometry (2026), arXiv:2505.20907 [cond-mat.str-el].arxiv.org/abs/2505.20907
  25. [25]Zhuang, Z.-Y., Zhu, D., Liu, D., Wu, Z., & Yan, Z. (2026). Odd-parity altermagnetism originated from orbital orders.arXiv
  26. [26]T. Jungwirth, R. M. Fernandes, J. Sinova, and L. Smejkal, Altermagnets and beyond: Nodal magnetically-ordered phases (2024), arXiv:2409.10034 [cond-mat.mtrl-sci].arxiv.org/abs/2409.10034
  27. [27]T. Oka and S. Kitamura, Floquet engineering of quantum materials, Annu. Rev. Condens. Matter Phys. 10, 387–408 (2019).DOI
  28. [28]F. Zhan, R. Chen, Z. Ning, D.-S. Ma, Z. Wang, D.-H. Xu, and R. Wang, Perspective: Floquet engineering topological states from effective models towards realistic materials, Quantum Frontiers 3, 21 (2024).arxiv.org/abs/2409.02774
  29. [29]T. Zhu, D. Zhou, H. Wang, S.-H. Wei, and J. Ruan, Floquet odd-parity collinear magnets, Phys. Rev. Lett. 136, 126704 (2026).DOI
  30. [30]S. Huang, Z. Qin, F. Zhan, D.-H. Xu, D.-S. Ma, and R. Wang, Light-induced odd-parity magnetism in conventional antiferromagnetism, Phys. Rev. Lett. 136, 126703 (2026).DOI
  31. [31]B. Li, D.-F. Shao, and A. A. Kovalev, Floquet spin splitting and spin generation in antiferromagnets, Phys. Rev. Lett. 136, 166701 (2026).DOI
  32. [32]T. Zhang, Z. Xiao, T. Fang, and Z. Wang, Light-induced switchable odd-parity altermagnetism in one- and two-dimensional triangulene crystals, Nano Lett. 26, 4889 (2026).DOI
  33. [33]D. Liu, Z.-Y. Zhuang, D. Zhu, Z. Wu, and Z. Yan, Light-induced odd-parity altermagnets on dimerized lattices, Phys. Rev. B 113, L060409 (2026).DOI
  34. [34]M. Zeng, Z. Qin, L. Qin, S. Feng, L. Wu, D.-H. Xu, and R. Wang, Odd-parity altermagnetism: A spin group study, Phys. Rev. B 113, L220412 (2026).DOI
  35. [35]M. Zeng, Z. Qin, L. Qin, S. Feng, J.-C. He, D.-H. Xu, and R. Wang, Odd-parity altermagnetism induced reconstruction of the chern insulating phase in the haldane-hubbard model, Phys. Rev. B 114, 175137 (2026).DOI
  36. [36]Y.-P. Lin and M. Vila, Odd-parity altermagnetism through sublattice currents: From Haldane-Hubbard model to general bipartite lattices, Phys. Rev. Lett. 137, 066702 (2026).DOI
  37. [37]T. Janssen, G. Chapuis, and M. De Boissieu, Aperiodic crystals: from modulated phases to quasicrystals: structure and properties (Oxford University Press, Oxford, 2018).DOI
  38. [38]Walter, S., & Deloudi, S. (2009). Crystallography of quasicrystals: concepts, methods and structures (Vol. 126).
  39. [39]Fukamichi, K. (1999). Magnetic properties of quasicrystals. In Physical Properties of Quasicrystals (pp. 295–326).
  40. [40]I. Fisher, Z. Islam, J. Zarestky, C. Stassis, M. Kramer, A. Goldman, and P. Canfield, Magnetic properties of icosahedral R–Mg–Zn quasicrystals (R=Y, Tb, Dy, Ho and Er), J. Alloys and Compd. 303-304, 223 (2000).DOI
  41. [41]E. Y. Vedmedenko, Magnetic ordering in quasicrystals, Mod. Phys. Lett. B 19, 1367 (2005).DOI
  42. [42]E. Y. Vedmedenko, U. Grimm, and R. Wiesendanger, Interplay between magnetic and spatial order in quasicrystals, Philos. Mag. 86, 733 (2006).
  43. [43]F. Hippert and J. Préjean, Magnetism in quasicrystals, Philos. Mag. 88, 2175 (2008).DOI
  44. [44]Z. M. Stadnik, Magnetic properties of quasicrystals and their approximants, in Handbook of Magnetic Materials, Vol. 21 (Elsevier, 2013) pp. 77–130.DOI
  45. [45]R. Tamura, A. Ishikawa, S. Suzuki, T. Kotajima, Y. Tanaka, T. Seki, N. Shibata, T. Yamada, T. Fujii, C.-W. Wang, M. Avdeev, K. Nawa, D. Okuyama, and T. J. Sato, Experimental observation of long-range magnetic order in icosahedral quasicrystals, J. Am. Chem. Soc. 143, 19938 (2021).DOI
  46. [46]Tamura, R., Abe, T., & Yoshida, S. e . a . (2025). Observation of antiferromagnetic order in a quasicrystal. Nat. Phys., 21, 974.
  47. [47]Godreche, C., Luck, J., & Orland, H. (1986). Magnetic phase structure on the Penrose lattice. J. Stat. Phys., 45, 777.
  48. [48]E. Y. Vedmedenko, H. P. Oepen, and J. Kirschner, Decagonal quasiferromagnetic microstructure on the Penrose tiling, Phys. Rev. Lett. 90, 137203 (2003).pubmed.ncbi.nlm.nih.gov/12689322
  49. [49]S. Wessel, A. Jagannathan, and S. Haas, Quantum antiferromagnetism in quasicrystals, Phys. Rev. Lett. 90, 177205 (2003).DOI
  50. [50]A. Jagannathan, Ground state of a two-dimensional quasiperiodic quantum antiferromagnet, Phys. Rev. B 71, 115101 (2005).DOI
  51. [51]A. Jagannathan, A. Szallas, S. Wessel, and M. Duneau, Penrose quantum antiferromagnet, Phys. Rev. B 75, 212407 (2007).DOI
  52. [52]A. Jagannathan and A. Szallas, Antiferromagnetism in two-dimensional quasicrystals, Z. Kristallogr. Cryst. Mater. 224, 53 (2009).arxiv.org/abs/0810.2312
  53. [53]A. Koga and H. Tsunetsugu, Antiferromagnetic order in the hubbard model on the Penrose lattice, Phys. Rev. B 96, 214402 (2017).DOI
  54. [54]Y. Li, M. Pan, J. Leng, Y. Chen, and H. Huang, Unconventional altermagnetism in quasicrystals: A hyperspatial projective construction, Phys. Rev. Lett. 137, 046704 (2026).DOI
  55. [55]R. Chen, B. Zhou, and D.-H. Xu, Altermagnetism in quasicrystals, Phys. Rev. Lett. 137, 046705 (2026).DOI
  56. [56]Z.-Y. Shao, J.-H. Ji, C. Lu, Z. Pan, Y.-B. Liu, and F. Yang, Classification of magnetism and altermagnetism in quasicrystals, npj Comput. Mater. (2026).arxiv.org/abs/2508.15702
  57. [57]J.-H. Ji, Z.-Y. Shao, Y.-B. Liu, W. Li, and F. Yang, Néel ordered magnetic phases in bipartite quasicrystals (2026), arXiv:2511.22204 [cond-mat.str-el].arXiv
  58. [58]N. de Bruijn, Algebraic theory of Penrose’s non-periodic tilings of the plane. II, Indag. Math. 84, 53 (1981).DOI
  59. [59]See Supplemental Material for the Penrose cut-and-project construction, Floquet-theory calculations, symmetry analysis, successive acceptance-window partitions and quasi-band theory, long-wavelength $h$-wave expansion, and time-resolved photoemission formalism, which includes Refs. [29–31, 34, 36, 58, 60–62, 65, 70, 72–74, 80, 81].. (n.d.). See Supplemental Material for the Penrose cut-and-project construction, Floquet-theory calculations, symmetry analysis, successive acceptance-window partitions and quasi-band theory, long-wavelength $h$-wave expansion, and time-resolved photoemission formalism, which includes Refs. [29–31, 34, 36, 58, 60–62, 65, 70, 72–74, 80, 81]..
  60. [60]V. Kumar, D. Sahoo, and G. Athithan, Characterization and decoration of the two-dimensional penrose lattice, Phys. Rev. B 34, 6924 (1986).pubmed.ncbi.nlm.nih.gov/9939344
  61. [61]H. Patel and D. Sherrington, Propagating modes in quasicrystals, Phys. Rev. B 40, 11185 (1989).pubmed.ncbi.nlm.nih.gov/9991682
  62. [62]A. Eckardt and E. Anisimovas, High-frequency approximation for periodically driven quantum systems from a Floquet-space perspective, New J. Phys. 17, 093039 (2015).DOI
  63. [63]F. D. M. Haldane, Model for a quantum Hall effect without Landau levels: Condensed-matter realization of the “parity anomaly”, Phys. Rev. Lett. 61, 2015 (1988).pubmed.ncbi.nlm.nih.gov/10038961
  64. [64]R. Eto and J. Knolle, Odd-parity magnons in the Haldane-Hubbard model from topological exciton condensation (2026), arXiv:2606.05837 [cond-mat.str-el].arXiv
  65. [65]R. Lifshitz, Symmetry of magnetically ordered quasicrystals, Phys. Rev. Lett. 80, 2717 (1998).DOI
  66. [66]R. Lifshitz, The symmetry of quasiperiodic crystals, Physica A Stat. Mech. Appl. 232, 633 (1996).DOI
  67. [67]D. A. Rabson, N. D. Mermin, D. S. Rokhsar, and D. C. Wright, The space groups of axial crystals and quasicrystals, Rev. Mod. Phys. 63, 699 (1991).DOI
  68. [68]R. Lifshitz, Theory of color symmetry for periodic and quasiperiodic crystals, Rev. Mod. Phys. 69, 1181 (1997).DOI
  69. [69]R. Lifshitz, Quasiperiodic spin space groups (2001).
  70. [70]R. Lifshitz and S. Even-Dar Mandel, Magnetically ordered quasicrystals: enumeration of spin groups and calculation of magnetic selection rules, Acta Crystallogr. A 60, 167 (2004).pubmed.ncbi.nlm.nih.gov/14966329
  71. [71]R. Lifshitz, Magnetic point groups and space groups, in Encyclopedia of Condensed Matter Physics, edited by F. Bassani, G. L. Liedl, and P. Wyder (Elsevier, Oxford, 2005) pp. 219–226.DOI
  72. [72]X.-J. Luo, Z.-T. Sun, X. Feng, M. Tian, and K. T. Law, Hidden Zeeman field in odd-parity magnets: An ideal platform for topological superconductivity, Phys. Rev. Lett. 137, 136003 (2026).DOI
  73. [73]C. Wang, F. Liu, and H. Huang, Effective model for fractional topological corner modes in quasicrystals, Phys. Rev. Lett. 129, 056403 (2022).DOI
  74. [74]H. Koivusalo and J. J. Walton, Cut and project sets with polytopal window I: Complexity, Ergod. Theory Dyn. Syst. 41, 1431 (2021).DOI
  75. [75]J. K. Freericks, H. R. Krishnamurthy, and T. Pruschke, Theoretical description of time-resolved photoemission spectroscopy: Application to pump-probe experiments, Phys. Rev. Lett. 102, 136401 (2009).DOI
  76. [76]M. A. Sentef, M. Claassen, A. F. Kemper, B. Moritz, T. Oka, J. K. Freericks, and T. P. Devereaux, Theory of floquet band formation and local pseudospin textures in pump-probe photoemission of graphene, Nat. Commun. 6, 7047 (2015).DOI
  77. [77]A. Farrell, A. Arsenault, and T. Pereg-Barnea, Dirac cones, floquet side bands, and theory of time-resolved angle-resolved photoemission, Phys. Rev. B 94, 155304 (2016).DOI
  78. [78]L. Broers and L. Mathey, Detecting light-induced Floquet band gaps of graphene via trARPES, Phys. Rev. Research 4, 013057 (2022).DOI
  79. [79]F. Boschini, M. Zonno, and A. Damascelli, Time-resolved ARPES studies of quantum materials, Rev. Mod. Phys. 96, 015003 (2024).DOI
  80. [80]Y. Yu, Tunable mixed parity spin splittings in altermagnets (2026), arXiv:2605.03026v2arXiv
  81. [81]R. V. Moody, Uniform distribution in model sets, Canadian Mathematical Bulletin 45, 123 (2002).DOI

Paper details

vlnc:W_pkw6jyps
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents