Random Quantum Circuits

• Random quantum circuits are unitary operators constructed from randomly-selected two-qubit (or qudit) gates that efficiently approximate Haar-random distributions.
• They use sequential, parallel, and local architectures to enable precise benchmarking, error-correcting code design, and simulation of many-body quantum dynamics.
• These circuits serve as minimal models for studying operator growth, entanglement dynamics, and computational complexity in quantum information science.

Random quantum circuits are unitary operators constructed as sequential or parallel compositions of randomly selected quantum gates, typically two-qubit unitaries, acting on a register of $n$ qubits (or more generally, qudits). These circuit ensembles underpin a wide spectrum of central results and applications in quantum information science, including quantum computational complexity, quantum chaos, entanglement and operator spreading, benchmarking, and quantum coding. Their significance arises from both their rapidly mixing properties—approximating Haar-randomness efficiently—and the theoretical control that randomness provides, mapping quantum many-body dynamics onto tractable classical statistical frameworks.

1. Models and Architecture of Random Quantum Circuits

Random quantum circuits are defined by the choice of spatial geometry, connectivity, and rules for gate randomization and placement. Common models include:

• Sequential random circuits: At each time step, an unordered pair of distinct qubits $(i,j)$ is chosen uniformly, and a two-qubit Haar-random unitary is applied (Brown et al., 2013). This process is iterated for $t$ steps.
• Parallel (brickwork) circuits: Disjoint pairs corresponding to even/odd bonds are updated in alternating layers, facilitating parallelization and efficient depth scaling (Haferkamp, 2022, Brown et al., 2012, Fisher et al., 2022).
• Local random circuits (LRC): Gates are constrained to act on neighboring qubits along a 1D chain or higher-dimensional lattice, respecting locality (Haferkamp, 2022, Gullans et al., 2020).
• Gate ensembles: Either true Haar-random $U(4)$ gates, Clifford-random gates (exact 2-designs), or subgroup-restricted gates enforcing symmetries (orthogonal, symplectic, etc.) (Hunter-Jones, 2018).

The circuit size may be specified in terms of the number of gates ($t$) or the parallel circuit depth (number of layers $d$). Random quantum circuit models admit efficient parallelization: with $t=O(n \log^2 n)$ random two-qubit gates, the typical circuit can be implemented with depth $O(\log^3 n)$ (Brown et al., 2013, Brown et al., 2012).

2. Unitary Designs and Randomness Generation

A central property is the ability of random circuits to approximate the uniform (Haar) measure on the unitary group, at least up to low finite moments. An ensemble $\nu$ of unitaries forms an $\epsilon$-approximate unitary $t$-design if its $t$-th moment superoperator matches that of the Haar measure up to diamond-norm error $\epsilon/d^t$ (Haferkamp, 2022). Depth bounds are:

• 2-designs: Brickwork or local circuits on $n$ qubits achieve $\epsilon$-approximate 2-designs in depth $O(n)$ (with logarithmic corrections), as established both by direct moment analysis and via “domain wall” partition function reformulations (Hunter-Jones, 2019, Brown et al., 2012).
• $t$-designs: The depth to achieve a $t$-design is $O(n t^{5+o(1)})$, nearly saturating known lower bounds. This scaling improves over the previous $O(n t^{10.5})$ and relies on sharp spectral gap estimates for the moment operator Hamiltonian and the effectiveness of Clifford-group mixing (Haferkamp, 2022).
• Physical mechanism: Statistical mechanics mappings interpret the approach to a $t$-design in terms of proliferating domain walls or defects in effective classical spin (permutation) models that capture the spreading of operator weight and local randomness (Hunter-Jones, 2019).

3. Scrambling, Decoupling, and Quantum Error Correction

Random circuits are minimal models for quantum information scrambling—rapid dispersal of local quantum information across the global state. Key results include:

• Decoupling: Random quantum circuits with $O(n\log^2 n)$ gates achieve near-optimal decoupling, making subsystems $B$ and $E$ nearly independent, as measured by the trace distance $$\left\|\mathcal{T}(U\rho_{AE}U^\dagger)-\tau_B\otimes\rho_E\right\|_1$$, with error bounds proceeding directly from Pauli-weight Markov chain mixing rather than general two-design character (Brown et al., 2013).
• Scrambling time and depth: In all-to-all architectures, depth $O(\log n)$ suffices for scrambling small ($O(1)$-size) subsystems; on local lattices, light-cone bounds yield depth scaling as $O(n^{1/d})$ in $d$ dimensions (Brown et al., 2012).
• Quantum codes from random circuits: Low-depth random circuits serve as efficient encoders for approximate quantum codes with linear rate and distance. In $D\ge2$, Clifford-random local circuits of depth $O(\log N)$ yield capacity-achieving erasure codes. In $D=2$, expurgation by targeted measurements produces codes with sub-logarithmic depth and high performance for near-term scalable devices (Gullans et al., 2020).

4. Operator Growth, Entanglement, and Transport

Random circuits capture universal features of many-body quantum chaos, including ballistic operator spreading ("butterfly velocity" $v_B$) and entanglement growth:

• Pauli-string Markov process: The spreading of operator weight is described by a Markov chain on Pauli strings. The operator's edge position exhibits biased diffusion, with analytically computable $v_B$ and diffusion constant $D$ depending on gate symmetry class (Hunter-Jones, 2018).
• Entanglement growth: In unitary circuits without measurements, local initial states evolve to states whose entanglement entropy grows ballistically (rate $v_E$) with KPZ-type (cubic-root) temporal fluctuations. Minimal membrane or "directed polymer" mappings yield exact predictions in the $q\to\infty$ or $n\to0$ limits (Fisher et al., 2022).
• Addition of conservation laws or measurements: Circuit models with additional constraints (e.g., U(1) symmetry) or stochastic mid-circuit measurements display rich dynamical phase diagrams, including entanglement phase transitions between volume-law and area-law regimes, described by percolation or membrane theory (Fisher et al., 2022).

5. Classical Simulation, Complexity, and Quantum Advantage

Random quantum circuits are critical in benchmarking quantum hardware, simulating quantum supremacy experiments, and establishing the computational intractability of quantum tasks:

• Classical hardness of sampling: Random circuit sampling (RCS) tasks are #P-Hard to simulate on average, not just in the worst case; the hardness is formalized via interpolation and worst-to-average-case reductions (using rational function reconstructions) (Movassagh, 2019). Any classical simulation within exponential additive error would collapse the Polynomial Hierarchy.
• Simulation algorithms: Efficient simulation of low-depth random circuits has been achieved with teleportation-inspired logical-qubit folding, enabling the contraction and sampling of circuits with thousands of qubits provided the depth is modest ($\sim 40$)—thereby pushing the boundary of classically simulable quantum dynamics (Chen et al., 2019).
• Frame potential diagnostics: Tensor network algorithms are used to numerically estimate the $k$-th frame potential $F^{(k)}$, which quantifies convergence to $k$-designs, expressibility of variational circuits, and the onset of barren plateaus in optimization (Liu et al., 2022). Experimental and numerical results confirm the linear-in-$n$ depth scaling for design generation.

6. Random Quantum Circuits in Quantum Information and Applications

The practical and conceptual utility of random quantum circuits extends across several domains:

• Benchmarking and certification: Randomized benchmarking and cross-entropy benchmarking protocols leverage random circuit ensembles for robust, sample-efficient evaluation of quantum gate fidelity and noise characteristics. Filtered randomized benchmarking schemes are provably implemented with linear-depth random circuits for arbitrary target symmetry groups (Heinrich et al., 2022).
• Random projection and quantum data science: Quantum random circuits act as efficient Johnson–Lindenstrauss projectors, embedding high-dimensional classical or quantum data into smaller spaces with guaranteed distance preservation. Local random circuits of linear depth suffice to attain near-Haar embedding quality (Kumaran et al., 2023).
• Statistical universality and quantum chaos: Effective field theory approaches (replica sigma models) analytically recover universal Wigner–Dyson spectral statistics and capture operator/entanglement spreading, connecting random circuits to quantum chaotic dynamics and thermalization mechanisms (Liao et al., 2022).

7. Complexity Diagnostics and Majorization

Alternative complexity indicators for random quantum circuits, such as the principle of decreasing majorization and the fluctuations of Lorenz curves of output probability distributions, serve as practical diagnostics:

• Majorization as complexity witness: All random circuit families satisfy decreasing majorization on average. However, the fluctuations of Lorenz curves sharply discriminate between universal (complexity-hard or sampling-hard) families and classically simulatable classes, corresponding to the emergence of atypical randomness in universal circuits (Vallejos et al., 2021).
• Comparison with entanglement spectrum and OTOCs: The distribution of entanglement-spectrum gap ratios and out-of-time-ordered correlators (OTOCs) serve as corroborating measures, but majorization-based diagnostics often yield sharper separation of complexity classes.

---

In summary, random quantum circuits provide a rigorous, flexible platform for realizing and analyzing fast quantum information scrambling, design generation, error-correcting coding, operator growth, benchmarking, and computational complexity in quantum many-body systems. Their dynamics, analyzed through both direct combinatorial, spectral, and statistical mechanics methods, have illuminated the structure of quantum chaos, established nearly optimal depth bounds for algorithmic tasks, and become foundational in both theoretical and experimental quantum information science (Brown et al., 2013, Haferkamp, 2022, Hunter-Jones, 2019, Gullans et al., 2020, Fisher et al., 2022, Movassagh, 2019, Liu et al., 2022, Kumaran et al., 2023, Vallejos et al., 2021, Heinrich et al., 2022, Liao et al., 2022).