Unitary $k$-designs without independent Hamiltonian quenches

AI summary

Overview

Research area: Quantum chaos, random matrix theory, unitary $k$-designs, many-body quantum dynamics, and holography. The paper sits at the intersection of quantum information and theoretical high-energy physics.

Technical level: Advanced. It assumes familiarity with frame potentials, Haar randomness, Gaussian random matrix ensembles, Sachdev–Ye–Kitaev (SYK) models, and basic holographic ideas.

Scope: A theoretical and numerical study (arXiv:2607.20640v3, RIKEN-iTHEMS-Report-26) showing that a single fixed chaotic Hamiltonian, interrupted by two fixed Pauli operators, can generate approximate unitary $k$-designs without independent Hamiltonian quenches.

What This Paper Is About

Unitary $k$-designs reproduce the first $k$ moments of Haar-random unitaries using far fewer resources, but existing quench-based constructions require several independent Hamiltonian realizations. This paper asks whether one fixed chaotic Hamiltonian—even a spatially local, deterministic one—can generate such designs when randomness enters only through sampled evolution times. The proposed answer is the two-Pauli-kick (2PK) protocol, in which the same Hamiltonian and the same Pauli operator are reused throughout.

Key Contributions

1. A single-Hamiltonian protocol for unitary $k$-designs. The 2PK protocol uses $U_{\mathrm{2PK}}(t_1,t_2,t_3)=e^{-iHt_3}P e^{-iHt_2}P e^{-iHt_1}$ with a fixed chaotic $H$, a fixed non-identity Pauli string $P$ satisfying $[P,H]\neq 0$, and evolution times sampled uniformly from $[0,T]$. No independent Hamiltonian quenches are required.

2. Analytic control of the frame potential. The author shows that the $k$-th frame potential converges to the Haar value as $F^{(k)}_{\mathrm{2PK}}\to k![1+O(1/d)]$ for $T\to\infty$, with corrections controlled for $k^2\ll d$. The one-Pauli-kick (1PK) protocol is shown to fail, with an explicit Gaussian-ensemble formula for its limiting value.

3. Verification across random and deterministic systems. The protocol is tested on single realizations of Gaussian ensembles, Majorana and Spin SYK models, and deterministic mixed-field Ising models. In the deterministic spin chains, the author reports the first realization of designs in such settings, to the author's knowledge, in regimes where previous quench protocols are inapplicable or fail.

4. Finite-temperature and holographic extensions. A finite-temperature frame potential is defined using the Gibbs density matrix, and an analytic lower bound is established in terms of the equilibrium partition function. The protocol is also interpreted holographically as a pair of bulk shockwaves.

Main Findings

• One Hamiltonian is sufficient. The 2PK frame potential approaches $k!$, the Haar value, for a single Hamiltonian realization. Randomness comes only from sampling $t_1,t_2,t_3$, not from sampling Hamiltonians.

• Two kicks are necessary; one is not enough. The 1PK protocol does not converge to the Haar frame potential. For Gaussian ensembles it approaches $F^{(k)}{\mathrm{1PK,GXE}}=k!\sum{j=0}^{k}\binom{k}{j}!(k-j)\Lambda^j$, where $!n$ is the derangement number and $\Lambda$ is the kurtosis factor of Pauli matrix elements in the energy basis, taking value $2$ for GUE and $3$ for GOE. For large $k$ this behaves as $e^{\Lambda-1}(k!)^2$.

• The deviation scales as $1/d$. For $k=1$ the leading deviation is $\Delta^{(1)}{\mathrm{2PK}}=F^{(1)}{\mathrm{2PK}}-1=2(\Lambda-1)^2/d$. The general deviation $\Delta^{(k)}{\mathrm{2PK}}:=F^{(k)}{\mathrm{2PK}}-k!$ scales as $1/d$, so the design becomes exact in the large-$d$ limit.

• 2PK is related to, but distinct from, the three-step protocol. Proposition 1 shows that 2PK equals, up to a fixed left multiplication by $P^2$, a three-step protocol with three isospectral Hamiltonians $H_1=H$, $H_2=P^\dagger H P$, and $H_3=P^{\dagger 2}H P^2$. For a Hermitian Pauli string, $P^2=\mathbb{I}$ and $H_3=H_1=H$, leaving ${H,PHP}$. Unlike the earlier three-step protocol, these Hamiltonians are not statistically independent.

• Gaussian-ensemble benchmarks. For GUE/GOE with $d=2^8$, a Pauli operator of weight $6$, and $10^4$ independent temporal realizations, the 2PK frame potential approaches $k!$. The ratio $F^{(k)}_{\mathrm{2PK}}/k!$ stabilizes to unity for $T\ge 10^4$, and the long-time limit used is $T=10^6$.

• SYK models work with a single realization. For Majorana SYK with $N_{\mathrm{Maj}}=16$ (GOE) and Spin SYK with $N_{\mathrm{Spin}}=8$ (GUE), using a Pauli operator of weight $6$ and $10^5$ temporal realizations, the frame potentials approach the Haar value. This includes Spin SYK$_2$, whose interaction structure is simpler than SYK$_4$, and no averaging over the SYK ensemble is performed.

• Deterministic local spin chains work. For mixed-field Ising models with $L=10$, $(h_x,h_z)=(-1.05,0.5)$, and a level-spacing ratio $\langle r\rangle\simeq 0.53$ confirming GOE statistics, the full-weight kick $P=ZXYZYXYXYY$ produces an approximate design. The protocol is typically effective for Pauli weights $\omega\gtrsim L/2$ and fails for low-weight strings.

• A geometric criterion for the kick. The kick rotates $H$ into $PHP$, with overlap ratio $c=\mathrm{Tr}(H,PHP)/\mathrm{Tr}(H^2)=1-2f$, where $f$ is the fraction of the Frobenius weight of $H$ whose sign the kick flips. An extensive fraction must be rotated, so $|c|\ll 1$ is necessary, though not sufficient.

• 2PK succeeds where 3SP fails under weak disorder. With weakly disordered MFIM couplings $g_i=W\mathcal{N}(0,1)$ and $W=0.2$, 2PK reproduces the Haar distribution $e^{-x}$ for $x=|\mathrm{Tr}(U^\dagger V)|^2$, while 3SP fails because weak disorder does not sufficiently rotate the eigenbases of independent local Hamiltonians ($|c|\approx 1$). The 2PK distribution of $y=\log_{10}x$ is a scaled Gumbel density peaking at $y=0$ with height $\ln 10/e=0.847$.

• A design diagnostic requiring no temporal sampling. For any non-degenerate $H$, $F^{(1)}{\mathrm{2PK}}=\mathrm{Tr}(w^4)$ exactly, with $w{ab}=|\langle E_a|P|E_b\rangle|^2$. Thus $\Delta^{(1)}{\mathrm{2PK}}\to 0$ once $P$ is delocalized in the energy eigenbasis. More generally, an approximate $k$-design implies $\Delta^{(1)}{\mathcal{E}}\to 0$ and $\kappa_{\mathcal{E}}=F^{(2)}{\mathcal{E}}/[2(F^{(1)}{\mathcal{E}})^2]-1\to 0$.

• Finite-temperature frame potential. With $\rho_\beta=e^{-\beta H}$ unnormalized, $F^{(k,\beta)}{\mathrm{2PK}}=\mathbb{E}{U,V}|\mathrm{Tr}(\rho_\beta U^\dagger V)|^{2k}$ obeys $F^{(k,\beta)}{\mathrm{2PK}}=k!(\mathcal{Z}(2\beta)/d)^k[1+O(1/d)]$ for all $\beta$, reducing to $k!$ at $\beta=0$. A universal lower bound $F^{(k,\beta)}{\mathcal{E}}\ge k!(\mathcal{Z}(2\beta)/d)^k[1+O(1/d)]$ is proved, with equality at $\beta=0$.

• Low-temperature growth. As $\beta\to\infty$, $F^{(k,\beta)}_{\mathrm{2PK}}\sim (k!/d^k)e^{2k\beta|E_0|}$, giving exponent $\alpha_k=2k|E_0|$. For GUE with spectral support in $[-2,2]$, $E_0\simeq -2$, so $\alpha_k=4k$, matching the numerical slopes.

• Comparison with prior protocols. The 2PK protocol uses a fixed $H$ and fixed $P$ with temporal randomness only, is applicable to local $H$, and admits finite-temperature treatment. The earlier three-Hamiltonian protocol uses temporal randomness plus Hamiltonian randomness, is not applicable to local $H$, and lacks a finite-temperature treatment. A two-Hamiltonian plus random-Pauli protocol leaves locality not established and fails at finite temperature.

• Holographic interpretation. In the interaction picture, $U_{\mathrm{2PK}}=e^{-iH(t_1+t_2+t_3)}P(t_1+t_2)P(t_1)$, where each precursor $P(\tau)=e^{iH\tau}Pe^{-iH\tau}$ is a bulk shockwave. The 2PK protocol therefore creates two shockwaves, and the frame potential quantifies how effectively repeated shockwaves randomize quantum information.

• Note added. The author notes that a contemporaneous reference requires Hamiltonian randomness, while 2PK is fully deterministic, and that a later reference confirmed the GUE results for 1PK and 2PK but did not address deterministic, spatially local chaotic Hamiltonians.

Methodology in Plain English

The author starts with a fixed chaotic Hamiltonian and a fixed Pauli operator. The system evolves under that Hamiltonian for a sampled time, receives a Pauli kick, evolves again, receives the same Pauli kick again, and evolves a third time. By sampling the three evolution times many times, one builds an ensemble of unitaries. The test of whether this ensemble mimics Haar randomness is the frame potential, an average overlap between pairs of unitaries raised to the $2k$-th power; the Haar ensemble gives the minimum value $k!$.

The calculations are done by diagonalizing the Hamiltonian once and rewriting the frame potential in terms of the eigenbasis and the Pauli matrix elements in that basis. This makes the protocol numerically cheaper than protocols requiring multiple independent Hamiltonians. The author then compares numerical frame potentials with analytic predictions for Gaussian random matrices, SYK models, and spin chains. For finite temperature, the Gibbs density matrix is inserted into the trace, and the result is related to the partition function.

Why This Matters

This work removes a major resource requirement from the construction of unitary $k$-designs: multiple independent Hamiltonian realizations. It shows that a single chaotic Hamiltonian plus two simple Pauli insertions can suffice, including in deterministic local spin chains. That makes design-like randomness more accessible to experimentally relevant settings and provides a finite-temperature framework tied to the partition function, which earlier multi-Hamiltonian constructions did not naturally offer.

Potential applications, which follow from the paper's framing rather than from hardware demonstrations reported here:

• Benchmarking quantum chaos and scrambling in a single quantum device or simulator, using one Hamiltonian and simple Pauli insertions rather than many independently sampled Hamiltonians.
• Finite-temperature quantum information protocols that need thermal weighting, since the frame potential is explicitly bounded by the equilibrium partition function.
• Holographic and quantum-gravity analogue studies, where the protocol maps onto repeated bulk shockwaves and may offer a controlled setting for shockwave geometry.
• Resource-efficient randomized quantum protocols that rely on design-like ensembles, where reducing the number of independent Hamiltonian realizations lowers experimental and numerical overhead.

Industry relevance: The result is primarily foundational, but it is relevant to quantum-computing hardware and software groups interested in benchmarking scrambling, validating chaotic dynamics, and implementing design-based protocols with minimal control overhead. The paper does not report an industrial implementation or application-specific benchmark.

Future Directions

• Chaos-to-integrability transitions and other ensembles. The author proposes studying spin chains across chaos-to-integrability transitions, SYK models, and Rosenzweig–Porter-type ensembles, where fractality may connect quantitatively to $k$-design formation.
• Continuous symmetries and conserved charges. If a charge is conserved, the kick must lie within the corresponding commuting subalgebra. The open question is when such symmetry-restricted kicks can still generate designs.
• The role of $k$-freeness. It remains to be understood whether $k$-design formation is governed by $k$-freeness, and the 2PK protocol is proposed as a natural testbed.
• A rigorous holographic treatment and $n$-Pauli-kick generalizations. The author suggests that a detailed study of the shockwave interpretation, and of the generalized $n$PK protocol with cumulative times $\tau_j=\sum_{i=1}^{j}t_i$, may reveal multi-shockwave or caterpillar geometries in bulk gravity.

Target Audience

This paper is best suited to researchers in quantum information, quantum chaos, random matrix theory, and holography who already know what a frame potential and a unitary $k$-design are. It will be especially useful to theorists and numerical many-body physicists interested in deterministic local systems, single-realization protocols, and finite-temperature generalizations. Experimentalists working on quantum simulators may find the protocol conceptually accessible, but the paper is written at an advanced theoretical level.

Authors’ abstract

Unitary $k$-designs provide a resource-efficient framework for emulating Haar randomness up to the $k$-th order. Quench-based protocols have recently been shown to generate such designs, but achieving this typically requires multiple Hamiltonian realizations, even when using temporal ensembles. Here, we show that no independent Hamiltonians are required: a single chaotic Hamiltonian is sufficient to generate approximate unitary $k$-designs, even when that Hamiltonian is spatially local. We introduce a two-Pauli-kick (2PK) protocol, in which unitary evolution under a fixed Hamiltonian is interspersed with two Pauli operator insertions (kicks). We find that the resulting frame potential approaches the Haar value at long times, with deviations suppressed by the inverse Hilbert-space dimension. We verify this for single realizations of Gaussian random matrices, the Majorana and Spin Sachdev-Ye-Kitaev models, and deterministic local quantum spin chains. Remarkably, in deterministic spin systems, the 2PK protocol generates approximate unitary designs in regimes where quench-based protocols are either inapplicable or fail to converge. Furthermore, our protocol provides a finite-temperature extension of the frame potential and establishes an analytic bound in terms of the equilibrium partition function. We discuss a holographic perspective on this mechanism.