⚛️ quantum physics

Asymptotically Good Quantum Codes with Addressable Transversal T Gates

This paper presents an explicit construction of asymptotically good binary CSS codes that support fully addressable transversal $T$ gates by combining algebraic-geometry codes with optimized binary embeddings to achieve generalized divisibility.

Original authors: Tongyin Lin, Bujiao Wu, Bin Cheng

Published 2026-09-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tongyin Lin, Bujiao Wu, Bin Cheng

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 quest to build a working quantum computer, scientists face a paradox of protection and control. To make a quantum machine useful, it must be shielded from the slightest environmental noise that could scramble its delicate calculations. This is done by encoding information across many physical particles, creating a "quantum code" that can detect and fix errors without looking directly at the data. However, to actually perform calculations, the computer must also manipulate this protected information. The rules of quantum mechanics make this difficult: the very operations that protect the data often prevent the computer from performing the complex logic needed for universal computing. For decades, researchers have searched for a way to apply specific, powerful logic gates to these protected states without breaking the shield. The goal is to find a method where a simple, direct action on the physical particles automatically translates into a precise, complex action on the logical information, all while keeping the system robust against errors.

A team of researchers has now constructed a specific family of quantum codes that achieves this elusive balance. They have designed a system that not only protects information efficiently but also allows for the direct, individual control of every piece of logical data using a single type of physical operation. In the language of quantum computing, they have created codes that are "asymptotically good," meaning they can handle a large amount of information with a fixed overhead and can correct a number of errors that grows steadily as the system gets larger. More importantly, these codes admit "fully addressable transversal T gates." This means that if a computer needs to apply a specific, complex rotation to just one logical qubit while leaving all others untouched, it can do so by simply applying a corresponding rotation to the physical particles that make up that specific qubit. No complicated follow-up steps or corrections are needed; the physical action does the logical work perfectly.

The researchers built this system by combining two distinct mathematical tools. First, they used a sophisticated type of code known as an algebraic-geometry code, which is known for its excellent error-correcting properties. These codes are defined over a large mathematical field, but the researchers needed to translate them into a binary format that a real quantum computer could use. To do this, they devised a clever "embedding" process, a method of mapping the complex field elements into a sequence of binary bits. This translation was not arbitrary; it was carefully engineered to preserve a specific mathematical property called "multiplication." By ensuring that the product of five specific code words always summed to zero in a particular way, they created a structure where the physical operations naturally aligned with the logical requirements.

The second key ingredient was a method to control exactly which logical qubits received the operation. In many previous attempts, applying a gate to the physical layer would affect all logical qubits at once, or require a messy series of corrections afterward. The team solved this by using a technique called "parity lifting." They showed that within their constructed codes, it is possible to assign specific weights to different parts of the code. By carefully choosing these weights, they could ensure that the mathematical conditions for the gate were met for one specific logical qubit while being neutral for all others. This allowed them to "address" any single logical qubit individually, applying the necessary transformation without disturbing the rest of the system. The result is a fixed encoding where the physical hardware and the logical data are locked in a relationship that permits this precise, independent control.

The paper confirms that this construction is not just a theoretical possibility but an explicit, step-by-step recipe. The researchers provided the exact mathematical definitions for the codes and the embedding maps, proving that the system works for any size large enough to be useful. They demonstrated that the codes maintain a constant rate, meaning the ratio of useful information to total physical resources does not shrink as the system grows. They also proved that the distance, which measures how many errors the code can fix, grows linearly with the size of the system. This linear growth is a critical benchmark for scalability, suggesting that these codes could protect increasingly large quantum systems without requiring an explosion in the number of physical particles needed for each logical unit.

While the construction relies on complex mathematics, the physical implication is straightforward: it offers a new path toward fault-tolerant quantum computing that avoids the need for constant, error-prone correction cycles after every gate operation. The researchers noted that finding the most efficient version of their binary embedding is a problem of minimizing the length of the code, similar to finding the shortest path in a vast network of possibilities. They developed methods to solve this optimization problem, which improved the efficiency of their codes and tightened the bounds on how well they perform. This work does not claim to have solved every problem in quantum computing, but it establishes a concrete, proven family of codes that successfully combines high error correction with the ability to perform universal logical operations directly and individually. It moves the field closer to a reality where quantum computers can be both robust and programmable, handling the complex logic required for real-world applications.

Technical Summary: Asymptotically Good Quantum Codes with Addressable Transversal T Gates

Problem Statement
A central challenge in fault-tolerant quantum computation is designing quantum error-correcting codes that simultaneously possess asymptotically good parameters (constant rate $K/N = \Theta(1)$ and linear distance $D/N = \Theta(1)$ ) and admit transversal non-Clifford gates. The Eastin–Knill theorem prohibits universal transversal gate sets on a single code, necessitating that non-Clifford gates, such as the $T$ gate ( $T = \text{diag}(1, e^{i\pi/4})$ ), be implemented via transversal operations or magic-state distillation.

While recent works have established the existence of asymptotically good codes with transversal logical $CCZ$ gates or addressable $CCZ$ gates, a specific gap remained for binary codes. Existing constructions for transversal $T$ either resulted in vanishing rates, growing but sub-linear distances, or induced only logical identities or Clifford gates ( $S^\dagger$ ) rather than the logical $T$ itself. Furthermore, achieving "full addressability"—the ability to apply a logical $T$ (or any power thereof) to any specific subset of logical qubits while acting as the identity on the rest, using a fixed encoding and without subsequent Clifford correction—had not been demonstrated for asymptotically good binary codes.

Methodology
The authors construct an explicit family of binary Calderbank–Shor–Steane (CSS) codes by combining algebraic-geometry (AG) codes with a novel binary embedding technique. The methodology proceeds through the following logical steps:

Generalized Divisibility Framework: The paper utilizes a weighted divisibility criterion. A binary code $C$ is $(a, b)$ -weighted 8-divisible if $\sum a_j c_j \equiv 0 \pmod 8$ for all codewords. The authors show that if a code admits specific divisibility coefficients on a set of information coordinates, puncturing and shortening these coordinates yields a CSS code with fully addressable transversal $T$ gates (Lemma 3.1).
From Multiplication to Divisibility: To satisfy the divisibility requirements, the authors leverage the "multiplication property" of classical codes. Specifically, they utilize the fact that if a binary code satisfies the 4-multiplication property (the component-wise product of any 5 codewords has even weight), one can lift vectors from the dual of the 4-fold product code to obtain the required weighted 8-divisibility coefficients with prescribed parities (Lemma 4.1 and 4.2).
Algebraic-Geometry Codes: The construction begins with AG codes over the field $\mathbb{F}_{1024}$ . The authors select a specific Galois tower and divisor configuration to ensure the codes possess the 15-multiplication property. This property implies that the coordinate sum of the component-wise product of any 16 codewords vanishes (Lemma 4.3).
Binary Embedding: A critical step is mapping the $\mathbb{F}_{1024}$ -based AG codes to binary codes while preserving the necessary algebraic structure. The authors define a symmetric, $\mathbb{F}_2$ -linear map $\sigma: \mathbb{F}_{1024} \to \mathbb{F}_2^L$ based on a five-input specialization of an alphabet-reduction map. This map ensures that the 15-multiplication property of the AG codes (vanishing sum of 16-codeword products) translates into the binary 4-multiplication property (vanishing parity of 5-codeword products) for the resulting binary codes (Lemma 4.4).
Addressability via Restriction: To achieve addressability, the authors restrict the binary code to a subspace generated by a specific element of trace one. This restriction preserves the 4-multiplication property and allows for the selection of a coordinate set $S$ where the divisibility coefficients can be independently prescribed.
Optimization of Embedding Length: The paper formulates the problem of finding the shortest binary embedding satisfying the required fivefold identity as a minimum-weight problem over an affine space defined by a linear system over $\mathbb{F}_2$ . Solving this optimization problem allows for the improvement of the asymptotic constants in the rate and distance bounds (Section 5).

Key Contributions and Results

Main Theorem (Theorem 1.2 / Theorem 4.5): The authors construct an explicit family of binary CSS codes with parameters $[[N_i, K_i = \Theta(N_i), D_i = \Theta(N_i)]]$ that admit fully addressable transversal logical $T$ gates.
Exact Implementation: For any prescribed tensor product of logical powers of $T$ (defined by a vector $c \in \mathbb{Z}_8^K$ ), the logical operation is implemented exactly by a tensor product of physical powers of $T$ . No subsequent Clifford correction is required, and the encoding is fixed independently of the logical operation.
Asymptotic Parameters: The constructed codes achieve a constant rate and linear distance. Specifically, the rate converges to $1/(991L)$ and the relative distances are bounded below by constants proportional to $1/L$ , where $L$ is the length of the binary embedding (optimized in Section 5).
Addressability: The construction allows for the independent application of $T$ gates to any chosen logical qubit (or subset thereof) while acting as the identity on others, a property not previously established for asymptotically good binary codes with transversal $T$ .

Significance
The paper claims to resolve the open question of whether explicit asymptotically good binary quantum codes can admit fully addressable transversal $T$ gates. By combining AG codes with a tailored binary embedding and generalized divisibility arguments, the work provides a theoretical foundation for fault-tolerant architectures that rely on transversal non-Clifford gates without the overhead of magic-state distillation for every operation. The construction demonstrates that the constraints of the Eastin–Knill theorem can be navigated to achieve universal gate sets (via $T$ and Clifford) with constant encoding overhead and linear error-correction capability, provided the specific structure of addressable transversal $T$ is utilized. The paper also highlights that the minimum embedding length problem can be treated as a combinatorial optimization problem, offering a pathway to further improve the concrete parameters of these codes.

Technical Summary: Asymptotically Good Quantum Codes with Addressable Transversal T Gates

More like this