Evaluating Sample-Based Krylov Quantum Diagonalization for Heisenberg Models with Applications to Materials Science

This paper evaluates the Sample-based Krylov Quantum Diagonalization (SKQD) algorithm on one- and two-dimensional Heisenberg models, demonstrating its ability to accurately reproduce ground-state properties and magnetization curves across various regimes through both classical benchmarks and successful implementation on 18- and 30-qubit quantum hardware.

The Gist

Imagine you are trying to find the lowest point in a vast, foggy mountain range. This mountain range represents the "energy landscape" of a magnetic material. The goal is to find the absolute bottom (the "ground state"), which tells scientists how the material behaves.

For a long time, scientists have used powerful classical computers to map these mountains. But as the mountains get more complex (with many interacting spins), the fog gets so thick that even the best classical computers struggle. This is where quantum computers come in, promising to see through the fog.

This paper tests a specific new tool for these quantum computers called Sample-based Krylov Quantum Diagonalization (SKQD). Here is a simple breakdown of what they did and what they found:

1. The Challenge: The "Dense" Fog

The researchers were studying a specific type of magnetic model called the Heisenberg model. Think of this as a chain of tiny magnets (spins) that can point up or down and interact with their neighbors.

• The Problem: In some conditions (like when the magnets are all equally strong and there is no external magnetic field), the "lowest point" of the mountain is hidden in a very dense, crowded area.
• The Analogy: Imagine trying to find a specific person in a stadium. If the person is standing alone in an empty field, it's easy to spot them (this is a "sparse" state). But if the person is standing in the middle of a packed crowd of thousands, it's much harder to find them (this is a "dense" state).
• The Theory: The SKQD algorithm was originally designed to work best when the target is "sparse" (easy to find). The researchers wanted to see if it could still work when the target was "dense" (hard to find), which is common in real magnetic materials.

2. The Strategy: Using a "Flashlight" and a "Sweep"

To tackle this, the team used two clever tricks:

• The Right Starting Point (The Flashlight): Instead of starting their search from a random spot, they started with a "Singlet State." Imagine this as starting your search right next to where you think the person might be, based on physics. This gave them a huge head start.
• The Magnetization Sweep (The Search Pattern): They knew that changing the external magnetic field (like turning on a giant magnet nearby) changes how the tiny magnets align. To find the true lowest energy, they didn't just look in one spot. They systematically "swept" through different possible arrangements (particle sectors), checking each one to see which configuration was the most stable.

3. The Experiment: Real Hardware and Simulations

They tested this method in three ways:

• Real Quantum Computers: They ran the algorithm on actual IBM quantum processors with 18 and 30 qubits (the basic units of quantum computing). This is like testing a new car on a real, bumpy road rather than just in a wind tunnel.
• Classical Benchmarks: They compared their results against "exact" calculations (which are perfect but only work for small systems) and DMRG (a highly accurate classical simulation method).
• 2D Simulations: They also tested it on a 2D grid (like a checkerboard) to see if it works for more complex shapes, not just a single line.

4. The Results: It Works Better Than Expected

The findings were encouraging:

• Accuracy: The SKQD method successfully predicted the energy levels and magnetic behavior (magnetization) of the materials.
• The "Dense" Surprise: Even in the "dense" regions where the theory said the algorithm might struggle, it still worked. It didn't give the perfect number every time, but it got the shape of the curve right. It correctly predicted how the material would react as the magnetic field got stronger.
• Improvement with Anisotropy: The method worked even better when the material was "anisotropic" (meaning the magnets had a preferred direction, making the "crowd" less dense and easier to navigate).
• 2D Success: They showed it works on 2D grids, not just 1D lines, suggesting it can handle more complex material shapes.

5. The Bottom Line

This paper is essentially a stress test for a new quantum algorithm.

• What they proved: SKQD is a robust tool that can handle complex, "crowded" magnetic problems that were previously thought to be too difficult for this specific type of algorithm.
• Why it matters: It bridges the gap between theoretical quantum math and real-world materials science. It shows that we can use current, imperfect quantum computers to get useful, accurate insights into how magnetic materials behave, even when the math gets messy.

In short, they took a tool designed for simple puzzles and proved it can also solve the complex, crowded puzzles found in real magnetic materials.

Technical Summary

Technical Summary: Evaluating Sample-Based Krylov Quantum Diagonalization for Heisenberg Models

Problem Statement
The Heisenberg model is a fundamental framework for understanding quantum magnetism, entanglement, and phase transitions in both low- and high-dimensional systems. However, numerically simulating strongly correlated regimes, particularly where the ground state is dense (non-sparse), remains computationally demanding for classical methods. While variational quantum algorithms like the Variational Quantum Eigensolver (VQE) have been applied to these systems, they often rely on ansatze tailored for sparse representations and suffer from issues such as barren plateaus and initialization sensitivity. The Sample-based Krylov Quantum Diagonalization (SKQD) algorithm was developed to address spectral estimation on quantum-centric supercomputing architectures, but its theoretical efficiency analysis assumes a sparse ground state. This paper investigates whether SKQD remains effective when applied to the Heisenberg model, specifically in regimes where the ground state violates these sparsity assumptions.

Methodology
The study evaluates the SKQD algorithm on one- and two-dimensional spin-1/2 Heisenberg models with open boundary conditions. The core methodology involves:

• Algorithmic Framework: SKQD constructs an effective Hamiltonian within a quantum Krylov subspace generated by time-evolved reference states ($|\psi_k\rangle = e^{-ikH\Delta t}|\psi_0\rangle$). Matrix elements of the effective Hamiltonian and the overlap matrix are obtained directly from sampled quantum measurements, avoiding full state tomography. The low-energy spectrum is derived by diagonalizing the generalized eigenvalue problem $H_{eff}v = \lambda S_{eff}v$.
• Initial State Strategies: To address the challenge of dense ground states and varying magnetization sectors, the authors employ two specific initialization strategies:
  1. Singlet Product States: Used as a physically motivated guess for antiferromagnetic (AFM) regimes, constructed as a product of singlet pairs.
  2. Magnetization-Sector Sweeps: To handle external magnetic fields (Zeeman terms) that break half-filling symmetry, the authors sweep through different particle sectors. Initial states for specific particle numbers $k$ are constructed as products of W-states distributed across qubit groups. This allows the algorithm to sample bitstrings from various sectors and identify the global ground state by comparing energies across sectors.
• Benchmarks: The algorithm is benchmarked against exact diagonalization (ED), Density Matrix Renormalization Group (DMRG), and Bethe ansatz solutions. Experiments are conducted on IBM quantum hardware (18- and 30-qubit Heisenberg chains) and via simulation on small 2D square-lattice systems.

Key Results

• Accuracy in Sparse and Dense Regimes: SKQD accurately reproduces ground-state energies and field-dependent magnetization across a range of anisotropies ($\Delta$). The method shows consistent qualitative agreement with DMRG and exact diagonalization.
• Dependence on Sparsity: The results confirm that accuracy is highest in regimes with sparse ground states (e.g., strong magnetic fields or high anisotropy/Ising limits). In the low-field, isotropic regime (where the ground state is highly entangled and non-sparse), the method still reproduces correct qualitative trends, though quantitative accuracy requires increased sampling and post-processing.
• Hardware Implementation: The authors successfully implemented SKQD on 18- and 30-qubit IBM Heron processors. The resulting magnetization curves for the material Copper Pyrazine Dinitrate (CuPzN) closely match theoretical Bethe ansatz curves, including nonlinear rises and saturation behaviors.
• 2D Extension: Simulations on a 6x4 square lattice demonstrate that SKQD extends effectively to two-dimensional geometries, capturing the monotonic dependence of ground-state energy on anisotropy and matching exact diagonalization trends.
• Role of Initialization: The study highlights that while a poor initial state does not necessarily worsen the theoretical error bound, it significantly increases the sampling cost required to reach that bound. The magnetization-sector sweep using W-states proved critical for accurately extracting magnetization in the presence of external fields.

Significance and Claims
The paper claims that SKQD is a robust, flexible quantum-classical method for correlated spin systems that can operate effectively beyond the strict sparsity assumptions underlying its formal efficiency analysis. By connecting algorithmic innovation with physically rich test systems, the work bridges current quantum computational techniques and foundational models of quantum magnetism.

The authors emphasize that SKQD offers a promising alternative to variational methods by eliminating the need for parameter optimization, thereby reducing vulnerability to barren plateaus. The successful demonstration on real quantum hardware for magnetization curves and field-driven transitions represents a significant step toward practical quantum simulation in materials science. The study concludes that SKQD remains robust even when sparsity assumptions are not strictly satisfied, making it a viable tool for near-term quantum devices in both algorithmic development and materials science applications.