⚛️ quantum physics

Hamiltonian Eigenvalue Transformation by Tridiagonal Gadgets

This paper introduces a method to implement arbitrary polynomial transformations of a local Hamiltonian using a single time-independent local Hamiltonian coupled to short chains of ancilla qubits, thereby enabling efficient eigenstate filtering and adiabatic optimization without the sequential oracle calls required by the circuit model.

Original authors: Arthur Braida, Joseph Cunningham, Jérémie Roland

Published 2026-10-05

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Arthur Braida, Joseph Cunningham, Jérémie Roland

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

Imagine a machine built to solve a problem by letting a physical system evolve over time. This is the promise of analog computing, a field where the laws of physics themselves perform the calculation. In this world, the machine is governed by a Hamiltonian, a mathematical description of how energy flows through a system of interacting parts. The beauty of this approach is that if the machine is built from local interactions—where each part only talks to its immediate neighbors—the system remains manageable and physically realizable. However, the algorithms designed to solve the hardest problems often require the machine to perform operations that are not local. They ask the system to act as if every part is connected to every other part simultaneously, a feat that no physical device can actually build. This creates a gap between the elegant theory of what a computer should do and the messy reality of what a device can do.

The central question for researchers is whether we can bridge this gap. Can we take a simple, local machine and make it behave exactly like a complex, non-local one, without having to build the impossible connections? A new study by Arthur Braida, Joseph Cunningham, and Jérémie Roland answers this with a resounding yes, but with a specific trade-off. They have shown how to construct a local device that mimics the action of a complex mathematical function on a quantum system. Instead of trying to build the impossible connections directly, they attach short, simple chains of extra particles to the main system. These chains act as filters, reshaping the energy of the system in a precise way. The result is a single, static machine that performs a complex transformation instantly, rather than a sequence of steps that must be timed perfectly.

The researchers focused on a specific type of mathematical tool called a polynomial, which is a way of describing a curve or a transformation using a sum of powers. In quantum algorithms, these polynomials are used to amplify the signal of a correct answer while suppressing the noise of wrong answers. The problem is that applying such a polynomial to a physical system usually requires the system to become highly non-local, breaking the rules of what can be built. The team's solution involves attaching a series of small, open chains of particles to the main system. Each chain is a simple line of sites where particles can hop from one to the next. The researchers discovered that each chain has a unique, isolated energy level that depends on the input system in a very specific way.

The magic of these chains lies in their length. A chain with a certain number of sites produces an energy shift that starts with a specific power of the input. A longer chain produces a shift that starts with a higher power. Because the starting powers are different for chains of different lengths, the researchers can treat them like building blocks. By attaching chains of various lengths and weighting them with specific numbers, they can add up their effects to recreate any desired mathematical curve. It is similar to how a painter mixes primary colors to create any shade; here, the "colors" are the energy shifts from chains of different lengths, and the "mix" is the final local machine.

The team proved that this method works with mathematical certainty for any input system that is not too strong. They showed that the chains do not interfere with each other and that the resulting machine is still local, meaning it only requires connections between a few neighboring particles at a time. The cost of this transformation is not in the complexity of the connections, but in the number of extra particles needed and the energy scale of the machine. To achieve a high degree of precision, the machine requires a number of extra particles that grows with the square of the complexity of the task, and the energy required to run it increases as well. However, this is a significant improvement over previous methods, which would have required the machine to run a long sequence of operations, effectively turning the analog device into a digital one.

One of the most striking applications of this work is in the search for a specific state within a vast system, a problem known as analog search. In the ideal version of this algorithm, the machine must apply a projector, a mathematical operation that isolates a single correct answer out of billions of possibilities. This projector is the most non-local object imaginable, connecting every particle to every other. The researchers demonstrated that their chain-based construction can approximate this projector with high accuracy. They simulated the process on a computer for systems up to twenty particles and found that the local machine they built reproduced the exact energy spectrum and the critical gaps of the ideal, non-local algorithm. The machine successfully isolated the marked state, proving that the complex, global operation could be carried by a simple, local device.

The researchers also explored a more efficient way to build this filter for specific tasks. Instead of synthesizing the entire curve in one go, they showed that iterating a simple two-particle block could achieve the same result. This method uses fewer extra particles and keeps the energy scale manageable, growing only polynomially with the size of the problem. In simulations, this iterative approach successfully mimicked the behavior of the ideal search algorithm, maintaining the crucial energy gaps that allow the system to find the solution efficiently. The work suggests that the exact behavior of these simple quantum chains is a powerful primitive, capable of performing complex transformations without the need for the intricate, time-dependent sequences that usually plague analog computing.

This research does not claim to have solved every problem in quantum computing, nor does it suggest that these machines are ready to be built in a lab tomorrow. The energy scales required are large, and the number of extra particles needed grows with the difficulty of the task. However, the study provides a rigorous proof that the gap between the ideal algorithms and physical devices can be closed. It demonstrates that a local, time-independent Hamiltonian can be constructed to perform the action of a complex polynomial, offering a new path for designing analog quantum computers. By turning a sequence of operations into a single, static structure, the work brings the theoretical power of quantum algorithms closer to the physical reality of what can be built.

Technical Summary: Hamiltonian Eigenvalue Transformation by Tridiagonal Gadgets

Problem Statement
Analog quantum devices implement local Hamiltonians $H$ , but many continuous-time algorithms require operators that are polynomials $P(H)$ of the input Hamiltonian. Generally, $P(H)$ is non-local, making it impossible to implement directly on hardware that supports only local, few-body interactions. While the circuit model addresses this via Quantum Singular Value Transformation (QSVT) by interleaving oracle calls (paying the degree in time/depth), the analog model lacks a mechanism to synthesize arbitrary polynomials as a single, time-independent local operator. The specific challenge addressed is: Can we find a local Hamiltonian whose restriction to an invariant subspace equals a prescribed polynomial of the original Hamiltonian?

Methodology
The authors propose a construction using "gadgets"—short chains of ancilla qubits attached to the input Hamiltonian $H$ . The core mechanism involves:

The Gadget Structure: A chain of $2m$ sites (for even polynomials) or $2n+1$ sites (for odd polynomials) is constructed. The couplings alternate between the input $H$ (or scalar $x$ ) and unit couplings. The chain terminates with a specific boundary condition: a detuning $-\Delta$ for even chains, or a coupling proportional to the input $\Delta_o x$ for odd chains.
Exact Eigenvalue Analysis: Unlike traditional perturbative gadgets that approximate effective interactions, this work analyzes the exact isolated eigenvalue branch $\lambda(x)$ $λ (x)$ of the gadget.
- For a chain of length $2m$ , the isolated eigenvalue $\lambda^{(m)}_0(x)$ vanishes at $x=0$ to an order of exactly $2m$ .
- The function is proven to be analytic within a complex disc $|x| < r$ (where $r < 1$ ) provided the detuning $\Delta$ exceeds a threshold $\Delta_\star(r)$ .
- The Taylor expansion of this branch is $\lambda^{(m)}_0(x) = \sum c^{(m)}_k x^{2m+2k}$ , with coefficients $c^{(m)}_k$ bounded uniformly in $m$ .
Triangular Synthesis: Because chains of different lengths produce eigenvalues with distinct leading orders ( $x^{2m}$ $x^{2 m}$ vs $x^{2(m+1)}$ $x^{2 (m + 1)}$ ), they form a triangular family. By attaching multiple chains of varying lengths to the same input $H$ $H$ with specific real weights $\alpha_j$ $α_{j}$ , the authors construct a weighted sum $\Sigma = \sum \alpha_j E_{2j}(H)$ $Σ = \sum α_{j} E_{2 j} (H)$ .
- The weights are determined by solving a lower-triangular linear system to match the coefficients of a target polynomial $P(x)$ term-by-term.
- The truncation error (terms beyond the target degree) decays geometrically with the number of chains, provided the input norm $\|H\|$ is strictly less than the analyticity radius $r$ .
Local Realization: The resulting operator $\Sigma$ acts on the system plus ancillas. It is shown to be a single, time-independent Hamiltonian with locality $(\kappa + 2)$ (where $\kappa$ is the locality of $H$ ), acting on an invariant subspace spanned by the tensor product of the system eigenvectors and the gadget's branch eigenvectors.

Key Contributions and Results

Theorem A (Analyticity and Bounds): Proves that the target eigenvalue branch is unique, analytic on a controlled complex disc, and possesses Taylor coefficients that are uniformly bounded. This establishes the mathematical foundation for the synthesis, ensuring the series converges for inputs with $\|H\| < r$ .
Theorem B (Polynomial Synthesis): Demonstrates that any even polynomial (and by extension, any polynomial using odd gadgets) can be approximated to accuracy $\epsilon$ $ϵ$ by a single local Hamiltonian.
- Resource Cost: For a polynomial of degree $2\ell$ , the construction requires $O(\ell^2 + \log^2(1/\epsilon))$ ancilla qubits.
- Locality: The output Hamiltonian is $(\kappa + 2)$ -local.
- Energy Scale: The norm of the Hamiltonian scales with the total weight of the synthesis coefficients, which grows quasi-polynomially with the degree for general polynomials.
Theorem C (Filtering via Iteration): For the specific application of filtering a marked eigenstate (e.g., in analog search), the authors propose a recursive approach using the exact eigenvalue map of a two-site chain ( $m=1$ $m = 1$ ) rather than synthesizing a high-degree Chebyshev polynomial in one pass.
- Mechanism: Iterating the map $k$ times folds the spectrum, isolating the marked state.
- Efficiency: This method uses only $O(\log n)$ ancilla qubits and $O(\log n)$ -local terms for an $n$ -qubit system.
- Energy Scale: Crucially, the energy scale of this iterative filter is polynomial in the inverse passband width ( $\sigma/d$ ) and independent of the number of iterations $k$ , unlike the one-pass synthesis where energy scales quasi-polynomially with degree.
- Convergence: The error decays geometrically in $k$ (as $\varrho^{-k}$ ), which is sufficient for analog search applications.

Significance and Claims
The paper claims to bridge the gap between idealized continuous-time algorithms (which often use non-local projectors or high-degree polynomials) and physically realizable local devices.

Static vs. Sequential: Unlike QSVT, which pays the degree in sequential depth (time), this method pays in width (ancilla qubits) and energy scale, but the transformation is a single static operator. This allows the transformed Hamiltonian to be used as a term in a larger Hamiltonian, a feature difficult to achieve with sequential circuit protocols.
Exact vs. Perturbative: The construction relies on exact eigenvalue analysis rather than perturbative expansions, avoiding the need for large detunings to suppress errors in the leading order.
Application to Analog Search: The authors simulate the construction for analog search (finding a marked state) up to $n=20$ . The results show that the gadget-based Hamiltonian reproduces the spectrum and the minimum gap scaling ( $2^{-n/2}$ ) of the ideal analog search algorithm.
Trade-offs: The paper acknowledges that the energy scale (and thus the time overhead for a fixed energy budget) is a cost. For general polynomials, this cost is quasi-polynomial in the degree. However, for the specific case of filtering (Theorem C), the energy scale is polynomial in the problem parameters and independent of the filter depth, making it a viable alternative to the rank-one projector in adiabatic optimization.

The work concludes that while the construction does not eliminate the energy cost inherent in simulating non-local operators, it provides a rigorous, static, and local realization of polynomial transformations, offering a concrete path to implementing idealized analog algorithms on hardware with limited locality.

Technical Summary: Hamiltonian Eigenvalue Transformation by Tridiagonal Gadgets

More like this