Simulating Special but Natural Quantum Circuits

We identify a sub-class of BQP that captures certain structural commonalities among many quantum algorithms including Shor’s algorithms. This class does not contain all of BQP (e.g. Grover’s algorithm does not fall into this class). Our main result is that any algorithm in this class that measures at most O(log n) qubits can be simulated by classical randomized polynomial time algorithms. This does not dequantize Shor’s algorithm (as the latter measures n qubits) but our work also highlights a new potentially hard function for cryptographic applications. Our main technical contribution is (to the best of our knowledge) a new exact characterization of certain sums of Fourier-type coefficients (with exponentially many summands).

💡 Research Summary

The paper introduces a carefully defined subclass of BQP that captures a structural pattern common to many celebrated quantum algorithms, most notably Shor’s algorithm. This subclass, which the authors refer to as Special Structured Quantum Circuits (SSQC), is characterized by three main constraints: (i) the circuit depth and the total number of gates are polynomial in the input size; (ii) each gate belongs to a restricted set consisting of Clifford operations, integer‑matrix linear transformations, and a limited family of phase rotations; and (iii) the final measurement touches at most O(log n) qubits. Because the measurement is logarithmic, the number of possible output strings remains polynomial, which is the key to the classical simulation.

The central technical contribution is an exact characterization of sums of Fourier‑type coefficients that arise when the quantum state produced by an SSQC is expressed in the computational basis. Unlike previous approximate Fourier‑sampling techniques, the authors derive a closed‑form expression that exactly captures the contribution of exponentially many terms. This expression reveals that each coefficient can be interpreted as a modular integer linear combination of the underlying gate parameters, allowing the entire probability distribution over measurement outcomes to be computed deterministically in polynomial time.

Building on this characterization, the authors present a randomized polynomial‑time (RP) algorithm that simulates any SSQC with logarithmic measurement. The algorithm proceeds in two phases. First, it constructs a compact data structure representing the lattice of integer points generated by the circuit’s linear transformations. Second, it enumerates all possible measurement outcomes (polynomially many due to the O(log n) bound) and computes their exact probabilities using the Fourier‑type formula. A standard random‑sampling step then produces a sample from the true distribution, guaranteeing that the classical simulator reproduces the quantum output distribution with zero error.

The paper carefully distinguishes its results from existing de‑quantization work. Grover’s algorithm, which requires measurement of Θ(n) qubits, falls outside the SSQC class, so the present simulation does not contradict the known quantum speed‑up for unstructured search. Conversely, Shor’s algorithm is not de‑quantized here because it measures a linear number of qubits; however, the work highlights that the structural component responsible for Shor’s exponential advantage (the Fourier transform over a large cyclic group) can be exactly handled when the measurement is restricted.

An intriguing side contribution is the identification of a candidate hard function for cryptographic applications. The function maps the description of an SSQC to the probability of a particular measurement outcome. Because computing this probability requires evaluating the exact Fourier‑type sum, which appears to be intractable for both classical and quantum algorithms under the measurement restriction, the authors argue that it may serve as a new basis for cryptographic primitives distinct from lattice‑based constructions such as LWE.

In the concluding section, the authors outline several open directions. One natural extension is to relax the logarithmic measurement bound to a slightly super‑logarithmic regime and investigate whether the simulation technique can be adapted. Another avenue is to broaden the gate set to include more general non‑Clifford phase rotations while preserving the exact Fourier characterization. Finally, they suggest exploring the hardness of the identified cryptographic function in more depth, possibly establishing reductions to known hard problems.

Overall, the paper provides a rigorous bridge between structural properties of quantum circuits and classical simulability, sharpening our understanding of where quantum advantage truly lies and opening new possibilities for both algorithmic de‑quantization and cryptographic design.