💻 computer science

Ramanujan quantum expanders from the Weil representation

This paper presents an exact construction of infinite families of Ramanujan quantum expanders for any odd prime power $q$ by transferring Morgenstern's Ramanujan Cayley graphs through the Weil representation, achieving the optimal singular value bound without additive error while using logarithmic gate complexity.

Original authors: Siddhartha Jain

Published 2026-09-30

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Siddhartha Jain

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the vast landscape of modern physics, there is a constant tension between the chaotic nature of randomness and the rigid structure required for reliable technology. Scientists often rely on random processes to solve problems that are too complex for step-by-step logic, much like how a hiker might wander through a dense forest to find a clearing rather than trying to map every tree in advance. In the realm of quantum computing, this idea translates to "quantum expanders," which are special mathematical tools that mix information together efficiently. Imagine a room full of people where everyone whispers a secret to a neighbor; if the connections are random and well-distributed, the secrets spread quickly and evenly throughout the crowd. Quantum expanders do this with the delicate states of quantum particles, ensuring that information disperses rapidly and uniformly. This speed is vital for building powerful quantum computers, but creating these tools is notoriously difficult because they must be constructed with extreme precision to avoid errors that could destroy the delicate quantum information.

For years, researchers have struggled to build these quantum mixers with the perfect efficiency known as the "Ramanujan" bound. This is a theoretical limit that represents the absolute best possible performance, where the mixing happens as fast as the laws of physics allow. Previous attempts could get very close to this ideal, but they always left a tiny margin of error, or they required such complex machinery that they were impractical to build. A researcher at The University of Texas at Austin has now solved this puzzle by constructing a new, infinite family of these perfect quantum expanders. Their work proves that it is possible to create these highly efficient mixers for a wide range of sizes, and crucially, they can be built using a specific, manageable set of quantum operations that do not introduce any extra error.

The path to this solution involved bridging two very different worlds of mathematics: the study of symmetrical shapes and the behavior of quantum waves. The researcher started with a known structure from classical mathematics, a type of network called a Cayley graph, which was already known to be a perfect mixer for classical information. The challenge was to translate this classical network into the quantum world without losing its perfect properties. To do this, they used a mathematical tool called the Weil representation, which acts like a translator, converting the movements of the classical network into operations on quantum states. They focused on a specific, hidden part of this translation process, a "subspace" where the quantum states behave in a very particular way. By isolating this specific part, they found that the complex quantum operations simplified into just three basic actions: shifting the phase of a wave, scaling it up or down, and performing a Fourier transform, which is a way of rearranging information based on patterns.

What makes this discovery significant is not just that they found a way to mix the information, but how efficiently they did it. The researcher showed that for a quantum system of a certain size, they could build the entire mixing machine using a number of basic steps that grows very slowly as the system gets larger. Specifically, the number of steps required is proportional to the square of the logarithm of the system's size. This means that even as the quantum computer grows to handle massive amounts of data, the effort to build the mixer remains surprisingly small. Furthermore, the construction uses a fixed set of tools, meaning the same basic instructions work regardless of how large the system becomes. This is a major improvement over previous methods, which often required increasingly complex and error-prone instructions as the system scaled up.

The researcher also addressed a critical concern regarding precision. In the real world, quantum computers are noisy, and small mistakes can accumulate. However, the team demonstrated that if the quantum circuit is built exactly as designed, using their specific set of tools, the mixing performance hits the theoretical limit perfectly. There is no leftover error or "additive noise" that pushes the performance slightly below the ideal. While building such a perfect circuit in a physical laboratory is a formidable challenge due to the fragility of quantum states, the mathematical proof shows that the ideal is achievable in principle. The work relies on a deterministic process, meaning that for any given size, the researcher can calculate the exact instructions needed to build the mixer without any guesswork or random searching.

This achievement opens a new door for the design of quantum algorithms. By providing a clear, exact blueprint for creating these perfect mixers, the researcher has removed a major theoretical obstacle that had long hindered progress. The construction works for a specific family of sizes determined by prime numbers, but it covers an infinite range of possibilities, suggesting that the method is robust and scalable. The paper does not claim to have built a physical device, but rather to have solved the mathematical and algorithmic problem of how such a device could be constructed. It establishes that the perfect quantum mixer is not just a theoretical dream, but a concrete reality that can be described with simple, efficient rules. This clarity allows other scientists to focus on the engineering challenges of building the hardware, knowing that the underlying logic is sound and optimal.

The work also clarifies what is not necessary to achieve this goal. For a long time, it was thought that building these mixers required a full "quantum Fourier transform" over a complex group, a massive and difficult operation that decomposes a system into all its possible parts. The researcher showed that this heavy machinery is unnecessary. Instead, they only needed to implement a single, large, irreducible representation, which is a much simpler and more direct approach. This insight simplifies the entire field, showing that the path to perfect quantum mixing does not require solving the hardest problems in quantum mathematics, but rather finding the right, simpler perspective on the problem.

In the end, the paper presents a complete and self-contained solution. It starts with a classical network, translates it into a quantum language using a specific representation, isolates the most efficient part of that language, and proves that the resulting machine works perfectly. The result is a family of quantum expanders that are as good as they can possibly be, built with a number of steps that scales efficiently, and defined by a set of rules that are exact and free of error. This provides a solid foundation for future developments in quantum computing, offering a clear target for engineers and a new tool for theorists to explore the limits of information processing.

Technical Summary: Ramanujan Quantum Expanders from the Weil Representation

Problem Statement

Quantum expanders are the quantum analogs of classical random walks on expander graphs. They are defined as unital quantum channels where the second singular value (the contraction factor on traceless operators) is small. Specifically, a degree- $D$ mixed-unitary channel $\Phi$ is a Ramanujan quantum expander if its second singular value $\sigma_2(\Phi)$ satisfies the bound $\lambda_D = \frac{2\sqrt{D-1}}{D}$ .

While Hastings (2007) showed that random unitaries achieve this bound asymptotically, the challenge lies in explicitly constructing such channels with efficient quantum circuits. Harrow (2008) proposed a transfer principle: if a classical Cayley graph on a group $G$ is a Ramanujan expander, and $\mu$ is an irreducible representation of $G$ , then the channel $\Phi_\mu(X) = \frac{1}{|S|} \sum_{s \in S} \mu(s) X \mu(s)^*$ inherits the spectral bound. However, implementing $\mu(s)$ efficiently has been a bottleneck. For the group $PSL_2(\mathbb{F}_p)$ , which admits Ramanujan Cayley graphs (Morgenstern, 1994), no efficient non-abelian Quantum Fourier Transform (QFT) is known to decompose the regular representation. Existing approaches either fail to achieve the optimal fixed-degree bound or require approximations that introduce additive errors.

Methodology

This paper constructs an infinite family of Ramanujan quantum expanders by combining three components:

Classical Ramanujan Graphs: Utilizing Morgenstern's construction of Ramanujan Cayley graphs on $PSL_2(\mathbb{F}_p)$ for $p = q^n$ (where $q$ is an odd prime power and $n$ is even).
The Odd Weil Representation: Instead of attempting a full group QFT, the author utilizes the odd irreducible subrepresentation of the Weil representation of $SL_2(\mathbb{F}_p)$ $S L_{2} (F_{p})$ .
- The Weil representation acts on the Hilbert space $\mathbb{C}^{\mathbb{F}_p}$ spanned by basis states $|x\rangle$ for $x \in \mathbb{F}_p$ .
- The representation is generated by three elementary operations: a quadratic phase (chirp), field scaling, and the additive Fourier transform over $\mathbb{F}_p$ .
- The author restricts this representation to the odd parity subspace $H^-_p = \text{ker}(R+1)$ , where $R|x\rangle = |-x\rangle$ . This subspace has dimension $N = (p-1)/2$ and is proven to be irreducible under the projective action of $PSL_2(\mathbb{F}_p)$ .
Efficient Circuit Implementation:
- Any element $g \in SL_2(\mathbb{F}_p)$ can be decomposed into at most four elementary matrices (Bruhat factorization).
- Under the Weil representation, these matrices correspond to the three elementary quantum operations mentioned above.
- The author demonstrates that these operations can be implemented using $O(\log^2 N)$ elementary gates and $O(\log N)$ ancilla qudits, using a fixed finite gate set dependent only on $q$ .

Key Contributions

1. Exact Satisfaction of the Ramanujan Bound

The primary theoretical contribution is the construction of a quantum channel that satisfies the Ramanujan bound $\lambda_D = \frac{2\sqrt{D-1}}{D}$ exactly, without additive error, provided the circuit is implemented exactly.

Previous work (e.g., Iyer et al., 2026) achieved near-Ramanujan bounds with additive error $\epsilon$ in time polylogarithmic in $1/\epsilon$ .
This construction relies on the fact that the two lifts of an element from $PSL_2(\mathbb{F}_p)$ to $SL_2(\mathbb{F}_p)$ differ only by a scalar phase on the odd subspace. Since the quantum channel involves conjugation ( $U X U^\dagger$ ), these scalar phases cancel out, making the channel well-defined and independent of the lift choice.

2. Efficient Circuit Complexity

The paper provides an explicit circuit construction for a degree- $D$ quantum expander of dimension $N$ :

Gate Count: $O(\log^2 N)$ elementary gates.
Ancilla: $O(\log N)$ ancilla qudits.
Gate Set: A fixed finite set of gates depending on $q$ (including field arithmetic, controlled permutations, and Fourier transforms over the base field $\mathbb{F}_\ell$ ).
Preprocessing: The classical preprocessing required to determine the field model, generator matrices, and circuit coefficients takes $\tilde{O}(\log^5 N)$ bit operations.

3. Avoidance of Full Group QFT

The construction bypasses the need for a full non-abelian QFT over $PSL_2(\mathbb{F}_p)$ . By identifying that a single large irreducible representation (the odd Weil representation) is sufficient for Harrow's transfer lemma, the author avoids the obstruction of decomposing the entire regular representation.

Results

The paper establishes the following theorem:
For every odd prime power $q$ and degree $D = q+1$ , there exists an infinite family of Ramanujan quantum expanders $\{\Phi_{p_k}\}$ acting on Hilbert spaces of dimension $N_k = (q^{2k}-1)/2$ .

The channel is a mixed-unitary channel of degree $D$ .
The second singular value satisfies $\sigma_2(\Phi_{p_k}) \le \frac{2\sqrt{D-1}}{D}$ .
The implementation uses $O(\log^2 N)$ gates and $O(\log N)$ ancillas.

The paper also addresses approximation: if the exact gate set is replaced by a universal gate set with Solovay-Kitaev synthesis, the singular value bound incurs an additive error proportional to the approximation error, consistent with standard quantum circuit synthesis limits.

Significance and Claims

The paper claims significance primarily in the exactness of the spectral bound and the efficiency of the construction.

Exactness: Unlike previous constructions that approach the Ramanujan bound asymptotically or with error, this method satisfies the bound exactly for the idealized circuit. This is a direct consequence of the representation-theoretic properties of the odd Weil representation and the specific choice of Morgenstern's graphs.
Efficiency: The construction achieves the optimal logarithmic scaling in circuit depth and size relative to the dimension $N$ , matching the efficiency of classical expander constructions while maintaining the quantum spectral properties.
Theoretical Bridge: It successfully bridges the gap between classical Ramanujan graphs (Morgenstern) and quantum expanders (Harrow) by utilizing a specific, efficiently implementable representation (Weil) that was previously underutilized for this specific purpose.

The author notes that while exact implementation is theoretically sound, practical implementation would rely on the Solovay-Kitaev theorem, introducing the standard overhead for approximating unitaries, but the theoretical framework provides a clean, error-free baseline. The work leaves open the problem of constructing such expanders for all degrees and for all sufficiently large Hilbert space dimensions, rather than just the specific sequence $N_k = (q^{2k}-1)/2$ .