A new quantum algorithm efficiently determines both the spectrum, the set of distinct eigenvalues, of a sparse Hermitian matrix and prepares corresponding eigenvectors as quantum states; it avoids fully reconstructing those vectors classically. This enables direct calculation of expectation values which are vital in many computational applications. The method allows direct calculation of expectation values essential in many areas of science and computation.
The work focuses on tackling large, sparse matrices that can be thought of like mostly blank spreadsheets containing key pieces of information. This is achieved through an approach using Quantum Phase Estimation, but what does it mean when spectral gaps, the separation between energy levels akin to distinguishing notes on a musical scale, become small. Further details outlining the algorithm’s mechanics and performance characteristics follow.
Quantum Algorithm Scales Linearly With Matrix Size For Small Spectral Gaps
A novel algorithm from the Indian Statistical Institute achieves a runtime of tildemathcalO(n s2 log2 n / Δmin), marking a significant improvement over classical methods scaling at n3. This speed-up is especially pronounced for matrices exhibiting small spectral gaps, inverse gaps up to O(n). Previously, accurately computing both eigenvalues *and* their associated eigenvectors for large matrices was computationally prohibitive; full eigenvector recovery demanded time scaling of order n3, even when leveraging sparsity.
The new quantum approach circumvents this limitation by encoding eigenvalue information as binary strings and preserving eigenvectors purely as quantum states, enabling direct expectation value calculations without costly classical reconstruction. Achieving runtime scaling of tildemathcalO(n s2 log2 n / Δmin), the algorithm offers a substantial advantage over classical methods requiring n3 operations in scenarios with small spectral gaps, the difference between eigenvalue levels, up to an inverse gap of O(n). Eigenvalues are represented as concise binary strings while maintaining eigenvector information solely within quantum states.
Direct calculation of expectation values, a measure of observable properties, becomes possible, eliminating the need to convert those eigenvector states into lengthy classical data. For matrices with nearly uniformly spaced spectra, such as those found in perfect-state-transfer spin chains or Sylvester-Kac matrices and up to certain sparsity levels, their approach outperforms existing techniques. However, these runtime benefits currently depend on efficient access to matrix elements; demonstrating this speedup on real hardware still requires overcoming challenges related to qubit coherence and gate fidelity.
Eigenvalue calculations accelerate idealized physics models with uniformly distributed spectra
The Indian Statistical Institute team and collaborators have demonstrated a potential acceleration in calculating eigenvalues and eigenvectors, values vital for both physics simulations *and* data science applications. This new technique could significantly accelerate specific calculations where energy levels are evenly spaced, offering benefits as quantum computers mature in these areas. This approach matters because it potentially offers faster computations than existing techniques when dealing with large datasets exhibiting small spectral gaps, crucial properties utilised in fields like materials science or graph analysis.
By representing eigenvalues as binary strings and keeping eigenvectors as quantum states, the method avoids reconstructing lengthy classical data sets, and achieves a runtime scaling of approximately tildemathcalO(n s2 log2 n / Δmin). The technique enables direct calculation of expectation values without fully realizing eigenvector information.
👉 More information
🗞 A Quantum Algorithm for Computing All Eigenvalues and Eigenstates of Sparse Hermitian Matrices
✍️ Krishnakanta Barik and Goutam Paul (Indian Statistical Institute)
🧠 ArXiv: https://arxiv.org/abs/2611.16610
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
