Determining the ground state of the hole-doped Hubbard model, a key to understanding high-temperature superconductivity, has been hampered by uncertainty in computational methods. Ao Chen, Johnnie Gray, Si-Jing Du and Garnet Kin-Lic Chan achieved unprecedented accuracy using advanced simulations on lattices up to 1024 sites; these calculations reach a precision of a few milli-t.

Simulations have revealed a surprising level of competition between different electronic states within materials thought to hold the key to high-temperature superconductivity; modelling up to 1024 sites achieves this unprecedented accuracy. Rather than one dominant arrangement of electrons, results indicate that superconductivity strongly contends with “stripes”, an alternative ordering pattern.

Current understanding of these complex materials requires further refinement and challenges existing theories about how they function. The California Institute of Technology team has achieved a breakthrough in understanding high-temperature superconductivity by simulating materials at an unprecedented level of accuracy; their calculations modelled up to 1024 sites, reaching precision within incredibly small energy differences, akin to distinguishing almost identical shades of colour. These new simulations reveal surprising competition between different electronic states within these complex materials, challenging existing theories about how they function.

Results indicate that superconductivity strongly contends with “stripes”, an alternative ordering pattern where electron density forms waves. This is similar to adding seasoning to food: introducing impurities (“doping”) alters the material’s properties in specific ways. The team’s work suggests a more nuanced picture than previously thought and raises questions regarding which state ultimately prevails under varying conditions; further details on their computational methods are presented below.

High Precision Simulations Reveal Competing Orders Near Degeneracy

Variational accuracy has reached unprecedented levels. Simulations on lattices with up to 1024 sites achieve precision down to a few milli-t, improving upon previous methods limited by uncertainties exceeding the energy scale of competing orders, roughly 0.01 t. This breakthrough enabled researchers at the California Institute of Technology to reveal near degeneracy between a low-energy uniform blank”>d-wave superconducting state, superconducting stripes and non-superconducting stripe states at 1/8 doping in the Hubbard model; all occurring within an incredibly small energy range.

The findings suggest extreme competition between superconductivity and different types of striped orders, fundamentally shifting understanding away from a single dominant order towards this subtle interaction in relevant parameter regimes. In particular, they employed neural tensor network and backflow state methods on lattices up to 1024 sites, achieving unprecedented precision exceeding previous calculations by a factor of ten, reducing uncertainties from approximately 0.01 t down to a few milli-t. While these results demonstrate extreme sensitivity between competing orders, predicting which ultimately prevails at finite temperatures or under external pressures remains vital for practical applications.

Energetically Similar Phases Demand Rethinking High Temperature Superconductivity Models

The California Institute of Technology researchers have demonstrated surprising competition between superconductivity and striped orders within the Hubbard model; however, their work doesn’t resolve *which* arrangement dominates, it merely establishes they exist in close energetic proximity. What subtle factors, perhaps imperfections or external pressures not included in this simplified model, might tip the balance towards one phase over another is now a critical question. Even though these results haven’t definitively declared a winner between superconductivity and competing “striped” orders within materials, this work isn’t inconclusive, it establishes that these states are remarkably similar in energy. At $1/8 doping, an electron concentration relevant to high-temperature superconductors, uniform d-wave superconductivity competes directly with both superconducting and non-superconducting striped states all within an incredibly small energy range measured in milli-t. This finding fundamentally alters expectations surrounding the Hubbard model, a key theoretical framework for understanding high-temperature superconductors. The team achieved such precision by combining advanced neural tensor network and backflow state techniques with large-scale computing power; simulating materials containing up to 1024 sites represents a leap in accuracy over prior methods, allowing them to explore how slight variations in parameters could favour which phase.

Using sophisticated computational simulations on lattices of up to 1024 sites, researchers found that at 1/8 doping within the Hubbard model, uniform d-wave superconductivity competes closely with both superconducting and non-superconducting striped states, all differing in energy by only milli-t$. This result demonstrates extreme sensitivity between these competing orders and challenges previous expectations about this theoretical framework for high-temperature superconductors.

The work reveals an energetic near degeneracy among these phases rather than a single dominant order, suggesting subtle factors may determine the ultimate ground state. Researchers indicate further investigation is needed to understand which phase prevails under different conditions.

👉 More information
🗞 Extreme competition between stripes and superconductivity in the hole-doped Hubbard model
✍️ Ao Chen, Johnnie Gray, Si-Jing Du and Garnet Kin-Lic Chan (California Institute of Technology)
🧠 ArXiv: https://arxiv.org/abs/2610.30321

More like this