⚛️ quantum physics

Low-gate-count block encodings for second-quantized fermionic Hamiltonians

This paper introduces new explicit block encoding constructions for second-quantized fermionic Hamiltonians that significantly reduce Clifford+T gate complexity and ancilla overhead by leveraging SWAP-based architectures and targeting specific particle subspaces, thereby enabling more resource-efficient early fault-tolerant quantum simulations.

Original authors: Diyi Liu, Shuchen Zhu, Lin Lin, Guang Hao Low, Chao Yang

Published 2026-09-17

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Diyi Liu, Shuchen Zhu, Lin Lin, Guang Hao Low, Chao Yang

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 understand the material world, scientists often turn to the behavior of electrons trapped within atoms and molecules. These tiny particles do not move independently; they interact in complex, collective ways that determine the properties of everything from the air we breathe to the chips in our computers. To predict these properties, researchers use mathematical models called Hamiltonians, which act as a complete instruction manual for how every electron in a system should behave. However, calculating the outcome of these instructions for even a modest number of electrons is impossible for classical computers, which process information in a linear, step-by-step fashion. The sheer number of possible arrangements of electrons grows so rapidly that the memory required to store the calculation exceeds the capacity of all the world's computers combined.

To solve this, scientists are developing quantum computers, machines that use the strange laws of quantum mechanics to process information in a fundamentally different way. Instead of following a single path, a quantum computer can explore many possibilities simultaneously. To make this work, researchers must translate the complex mathematical rules of electron interactions into a format a quantum computer can read and execute. This translation process is known as "block encoding." It is a method of packaging the Hamiltonian's instructions into a larger, manageable quantum circuit. The efficiency of this packaging is critical: if the instructions are too bulky or require too many steps to execute, the quantum computer will run out of time and coherence before it can finish the calculation. The goal is to find the most compact, efficient way to encode these instructions so that the machine can simulate nature with the fewest possible resources.

A team of researchers has developed a new, highly efficient method for this packaging process, specifically tailored for systems where the number of electrons remains constant. In their work, they introduced a construction that significantly reduces the number of complex operations required to load the data into the quantum computer. Previous methods treated every possible interaction between electrons as a separate item to be loaded, regardless of whether that interaction was actually possible in the specific system being studied. This approach was like trying to find a specific book in a library by checking every single shelf in the building, even if the book was known to be in only one specific room. The new method, however, acts more like a librarian who knows exactly which shelves contain the books relevant to the current reader, skipping the irrelevant sections entirely.

The researchers achieved this by designing a system that dynamically checks which electron states are actually occupied before loading the data. They created a set of logical tools, or "oracles," that act as gatekeepers. One tool determines which electron positions are valid for a given state, while another tool loads the specific strength of the interaction for those valid positions. By using a technique that swaps data into place only when needed, they avoided the heavy computational cost of loading every possible interaction at once. This approach allows the quantum computer to focus its energy on the interactions that actually matter for the system at hand, rather than wasting resources on impossible scenarios.

The results of this work show a dramatic reduction in the computational cost. For a general system of electrons, the number of complex steps required to perform the simulation scales with the square root of the number of possible interactions, rather than linearly. This is a substantial improvement, meaning that as the system grows larger, the new method becomes increasingly more efficient compared to older techniques. Furthermore, by restricting the simulation to a fixed number of particles, the researchers were able to reduce the "subnormalization factor," a measure of how much the quantum signal is diluted during the process. In simpler terms, this means the quantum computer can extract the correct answer with much higher fidelity and fewer repetitions.

The team also demonstrated that their method works exceptionally well for systems with specific structures, such as those where electrons only interact with their immediate neighbors or where the interaction strength follows a predictable pattern based on distance. In these cases, the efficiency gains are even more pronounced. The researchers provided detailed blueprints for how to build these circuits, showing that the number of physical components required is significantly lower than what was previously thought necessary. This work does not just offer a theoretical improvement; it provides a practical path forward for simulating complex chemical and physical systems on early fault-tolerant quantum computers. By lowering the resource overhead, the new method brings the simulation of real-world materials closer to reality, potentially accelerating the discovery of new drugs, materials, and energy solutions.

Technical Summary: Low-gate-count block encodings for second-quantized fermionic Hamiltonians

Problem Statement
Efficient block encoding of many-body Hamiltonians is a prerequisite for quantum algorithms in scientific computing, particularly for early fault-tolerant quantum simulation. A primary bottleneck in existing methods, such as the Linear Combination of Unitaries (LCU) approach, is the resource cost associated with loading classical Hamiltonian coefficients into quantum circuits. For general second-quantized electronic-structure Hamiltonians, the number of interaction terms $L$ scales as $\Theta(n^4)$ (where $n$ is the number of spin orbitals). Standard constructions result in a T-gate complexity scaling linearly with $L$ (i.e., $O(n^4)$ ) and a subnormalization factor $\alpha$ that also scales as $O(n^4)$ . Furthermore, many applications require simulating systems with a fixed number of particles $\eta$ , yet standard block encodings often diffuse over the full Hilbert space, failing to exploit particle-number conservation to reduce resource overheads.

Methodology
The authors propose new explicit constructions for block encoding second-quantized fermionic Hamiltonians that act directly on the occupation-number representation rather than first mapping fermionic operators to Pauli strings. The methodology relies on two primary oracles: a sparsity oracle ( $O_C$ ) and an amplitude oracle ( $O_A$ ).

Sparsity Oracle ( $O_C$ ): Instead of using multi-controlled gates to determine matrix element locations, the authors employ a SWAP-based data-lookup strategy. They utilize SWAP-UP networks to move specific qubits representing occupied orbitals to the least significant position, allowing for efficient controlled bit-flips and phase encoding. This approach constructs the column-sparsity oracle using SWAP-UP, ladder circuits (staggered CNOTs), and targeted Pauli-Z gates to handle fermionic signs.
Amplitude Oracle ( $O_A$ ): To encode the numerical values of coefficients, the authors combine a SELECT-SWAP data-lookup circuit with a direct sampling method. The SELECT-SWAP technique groups data items to reduce the T-gate count for loading classical data from $O(L)$ to $\tilde{O}(\sqrt{L})$ . The direct sampling method converts the binary representation of coefficients into probability amplitudes without requiring complex arithmetic or alias sampling, provided coefficients are $O(1)$ or decay with indices.
Fixed- $\eta$ Particle Construction: For systems with a fixed particle number $\eta$ , the authors introduce an indirect diffusion strategy. Rather than diffusing over all formal orbital indices, the diffusion is conditioned on the input state $|j\rangle$ to generate superpositions only over indices corresponding to occupied orbitals. This requires an occupation detection oracle ( $O_{occ}$ ) to compute the address of the $i$ -th occupied qubit transiently in an ancilla register. This state-dependent approach restricts the diffusion to the relevant subspace, significantly reducing the subnormalization factor.
Structured Hamiltonians: The framework is extended to Hamiltonians with translation invariance, nearest-neighbor interactions, or coefficient decay. In these cases, the number of distinct coefficients to be loaded is reduced, and the indirect diffusion is combined with arithmetic circuits to generate superpositions of valid index pairs (e.g., $|p\rangle|p \pm m\rangle$ ) rather than all possible pairs.

Key Contributions and Results

Quadratic Improvement in Gate Complexity: For general second-quantized Hamiltonians with $L$ interaction terms, the proposed construction achieves a T-gate count scaling of $\tilde{O}(\sqrt{L})$ (specifically $\tilde{O}(\sqrt{L \log(L/\epsilon)})$ ), a quadratic improvement over the linear $O(L)$ scaling of standard LCU/QROM methods. For electronic-structure Hamiltonians ( $L \sim n^4$ ), this reduces the T-count from $O(n^4)$ to $\tilde{O}(n^2)$ .
Reduced Subnormalization Factor in Fixed- $\eta$ Sector: While the subnormalization factor remains $O(L) \sim O(n^4)$ in the general full-Hilbert-space setting, the construction targeting the fixed- $\eta$ particle subspace reduces the factor from $O(L)$ to $O(n^2 \eta^2)$ for general two-body interactions. When $\eta \ll n$ , this represents a substantial reduction in the qubitization cost for time evolution.
Optimized Resource Estimates for Structured Models:
- Hubbard Models: For the standard Hubbard model (only two unique coefficients), the data-lookup cost vanishes, and the complexity is dominated by the sparsity oracle, yielding $O(n + \log(n/\epsilon))$ T-gates for the full space and $O(n \log \eta + \log(\eta/\epsilon))$ for the $\eta$ -particle sector.
- Translation-Invariant and Localized Models: For Hamiltonians with translation invariance or localized orbitals (where coefficients decay), the number of distinct coefficients $L_{distinct}$ is reduced. The construction adapts to these structures, further lowering the T-count and T-depth compared to unstructured cases.
Comparison with Existing Models: The paper compares its results against the Babbush et al. [28] LCU model and the Kivlichan et al. [49] Trotter-based approach. The new method offers lower T-counts and T-depths for qubitization-based simulation. Notably, while related fixed-particle input models exist (e.g., Tong, An et al. [38]), the authors' construction is the first explicit block encoding that directly targets the fixed- $\eta$ sector of a general second-quantized Hamiltonian (including off-diagonal one- and two-body interactions) and achieves optimal subnormalization scaling through circuit construction itself, without relying on Hamiltonian shifts or instance-dependent classical optimization.

Significance
The authors claim that this work provides a practical path toward early fault-tolerant quantum simulation of many-body systems. The significance lies in:

Substantial Resource Reduction: The reduction of T-gate complexity from linear to square-root in the number of terms ( $L$ ) and the reduction of the subnormalization factor for fixed-particle systems directly lower the fault-tolerant overhead for simulating time evolution and eigenvalue problems.
Direct Subspace Targeting: The construction is the first explicit block encoding that directly targets the fixed- $\eta$ sector of a general second-quantized Hamiltonian, achieving optimal subnormalization scaling for off-diagonal interactions through circuit construction rather than Hamiltonian shifts or classical optimization.
Scalability: The approach scales efficiently with system size $n$ and particle number $\eta$ , making it suitable for simulating larger molecular and condensed matter systems than previously feasible with standard input models.
Modularity: The framework is general and can be adapted to various Hamiltonian structures (nearest-neighbor, translation-invariant, factorized), demonstrating that exploiting physical symmetries and locality can be systematically integrated into the block-encoding framework to further reduce costs.

The paper concludes that while the improvements are most pronounced for structured Hamiltonians and fixed-particle sectors, the general construction already offers significant advantages over previous methods by leveraging SWAP-UP networks and SELECT-SWAP data lookups to minimize Clifford+T gate overheads.

Technical Summary: Low-gate-count block encodings for second-quantized fermionic Hamiltonians

More like this