⚛️ lattice

Quantum Monte Carlo calculations in the nuclear shell model by the complex Langevin method

This paper demonstrates that the complex Langevin method can successfully overcome the sign problem in auxiliary-field quantum Monte Carlo calculations for small nuclear shell model systems at finite temperature, accurately reproducing exact energy results and offering a promising complementary approach to variational methods for larger systems.

Original authors: Yuhma Asano, Yuta Ito, Jun Nishimura, Noritaka Shimizu

Published 2026-10-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yuhma Asano, Yuta Ito, Jun Nishimura, Noritaka Shimizu

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

Atoms are the fundamental building blocks of the visible world, but inside the nucleus, the rules of physics become a chaotic tangle. To understand how protons and neutrons stick together to form the elements that make up our universe, physicists rely on a powerful framework called the nuclear shell model. This model treats the nucleus somewhat like a miniature solar system, where particles orbit in specific layers or shells around a dense core. While this picture works beautifully for simple cases, the math required to describe real, complex nuclei involves tracking the interactions of dozens of particles simultaneously. As the number of particles grows, the complexity of the calculations explodes, making it nearly impossible to solve the equations exactly for anything but the lightest atoms.

For decades, scientists have tried to bypass this mathematical wall using a technique called quantum Monte Carlo. Think of this as a way to simulate the behavior of a system by running millions of random trials to find the most likely outcome. However, when applied to realistic nuclear forces, this method hits a notorious roadblock known as the sign problem. In these simulations, the mathematical values representing different possibilities can become negative or even imaginary numbers. When this happens, the random trials cancel each other out in a way that destroys the accuracy of the result, leaving researchers with noise instead of a clear answer. This has forced scientists to rely on approximations that work well for some cases but fail to capture the full, messy reality of how atomic nuclei behave, especially when they are heated up or contain many particles.

A team of researchers has now taken a significant step toward solving this decades-old puzzle by applying a sophisticated new tool called the complex Langevin method. Instead of trying to force the simulation to stay within the realm of ordinary real numbers, this approach allows the mathematical variables to wander into the complex plane, a mathematical landscape that includes imaginary numbers. By letting the simulation explore this broader territory, the researchers found a way to navigate around the sign problem that has plagued nuclear physics for so long. They tested this method on a small but realistic model of a calcium nucleus, containing twenty available energy slots and a few neutrons moving within them. The goal was to see if this new technique could accurately predict the energy of the nucleus at different temperatures, a task that standard methods struggle with when the interactions between particles are strong and realistic.

The results were encouraging, though they came with a specific condition. The researchers found that the complex Langevin method worked perfectly when the nucleus was at a high temperature, producing energy values that matched the exact solutions known from other methods. However, as they cooled the system down to lower temperatures, the simulation began to drift away from the correct answer. This failure was not random; the team could actually see it coming by monitoring a specific mathematical signal within the simulation. When the simulation started to produce unreliable results, the distribution of a key variable changed shape, losing the rapid drop-off that indicates a stable calculation. This gave the researchers a built-in warning system, allowing them to know exactly when their method was trustworthy and when it was not.

To get around this temperature limit, the team devised a clever workaround. Instead of trying to force the simulation to work at low temperatures directly, they introduced a dial that could gradually turn down the strength of the interactions between the particles. They ran the simulation at high temperatures where the method was known to be reliable, but with the interactions weakened. By measuring the energy at these weaker settings, they could map out a smooth curve and then mathematically extend that curve back to the full, realistic strength of the interactions. This process, known as extrapolation, allowed them to predict the energy of the nucleus at low temperatures with high accuracy, matching the exact solutions without ever having to run the unstable simulation in the dangerous low-temperature zone.

The study confirms that this new approach is a viable path forward for understanding heavier nuclei, which are currently out of reach for exact calculations. While the method requires careful monitoring to ensure it stays within its valid range, the ability to combine high-temperature simulations with a controlled extrapolation offers a powerful new tool. It suggests that in the future, physicists may be able to use this technique to explore the properties of heavy elements and the behavior of nuclear matter under extreme conditions, filling in gaps in our understanding of the atomic world that have remained stubbornly empty for years. The work does not claim to have solved every problem in nuclear physics, but it provides a robust, complementary strategy that works alongside existing methods, offering a clearer view of the complex machinery at the heart of matter.

Technical Summary: Quantum Monte Carlo calculations in the nuclear shell model by the complex Langevin method

Problem Statement
The nuclear shell model is a highly effective theoretical framework for describing nuclear properties, including low-lying energy spectra and thermal properties. However, its application to large systems is hindered by the combinatorial explosion of the Hamiltonian matrix dimension. While the Auxiliary-Field Monte Carlo (AFMC) method, specifically the Shell Model Monte Carlo (SMMC) approach, was introduced to circumvent this dimensionality issue, its practical application to realistic effective interactions is severely restricted by the "sign problem." This problem arises because realistic interactions (e.g., those derived from G-matrix theory or VS-IMSRG) often lead to complex weights in the path integral, rendering standard stochastic sampling inefficient.

Existing alternatives, such as variational methods (e.g., VAMPIR, Monte Carlo Shell Model), provide upper bounds or extrapolated values dependent on specific wave function ansatzes and are generally inapplicable to finite-temperature physics. Furthermore, previous attempts to bypass the sign problem using artificial interactions and extrapolation (e.g., $g$ -extrapolation) often introduce large uncertainties. Consequently, there is a need for a quantum Monte Carlo technique that can handle realistic interactions without relying on variational ansatzes.

Methodology
This paper applies the Complex Langevin Method (CLM) to the nuclear shell model to address the sign problem. The authors utilize two formalisms:

Canonical Formalism: Directly projects onto a fixed number of active particles ( $N_v$ ). While exact, the implementation becomes computationally complex as $N_v$ increases.
Grand-Canonical Formalism: Introduces a chemical potential and treats the particle number projection via an integral over a phase variable $\phi$ . This formalism is computationally more convenient for varying $N_v$ and is the primary focus of the simulations.

The authors simulate the system at finite temperature using the imaginary-time formalism. The partition function is linearized using the Hubbard-Stratonovich transformation, introducing auxiliary scalar fields $\sigma_{\alpha n}$ . In the CLM, these fields and the phase variable $\phi$ are complexified. The dynamics are governed by the discretized complex Langevin equation, where the drift term is derived from the effective action $S$ .

To ensure the validity of the CLM, the authors strictly monitor the "validity criterion" proposed in Ref. [25]. This criterion requires that the probability distribution of the magnitude of the drift term $v(z)$ falls off exponentially or faster at large magnitudes. If this condition is met, the boundary terms in the integration by parts vanish, ensuring the CLM converges to the correct expectation values. The authors employ an adaptive step-size for the time evolution to control instabilities.

Key Contributions and Results
The study focuses on small systems ( $N_s = 20$ single-particle states in the $pf$ shell) with realistic GXPF1A interactions. The key findings are:

Temperature Dependence: In the grand-canonical formalism, the CLM results for the energy expectation value agree with exact Hamiltonian diagonalization only at sufficiently high temperatures (low inverse temperature $\beta$ ). For $N_v=2$ , agreement holds for $\beta \lesssim 0.6$ ; for $N_v=10$ , agreement holds for $\beta \lesssim 0.4$ .
Validity Criterion Correlation: The breakdown of agreement at lower temperatures correlates with the failure of the validity criterion. Specifically, the histogram of the drift term magnitude for the $\sigma$ field ceases to show a fast exponential fall-off when $\beta \gtrsim 0.4$ , indicating that the CLM is no longer guaranteed to yield correct results in this region.
Extrapolation Strategy: To access lower temperatures (larger $\beta$ $β$ ) where the CLM fails directly, the authors introduce a tunable parameter $t$ $t$ in the interaction term of the Hamiltonian ( $\hat{H}_t$ $\hat{H}_{t}$ ). They perform simulations in the region where the validity criterion is satisfied (small $t$ $t$ ) and extrapolate the results to the physical point $t=1$ $t = 1$ .
- For $N_v=2$ and $\beta=1$ , the CLM is valid for $t \lesssim 0.4$ (grand-canonical) and $t \lesssim 0.5$ (canonical).
- A quadratic fit of the energy data in the valid region successfully reproduces the exact diagonalization result at $t=1$ within fitting errors.
- The extrapolation is found to be slightly more robust in the canonical formalism due to a marginally larger validity region.

Significance and Claims
The paper concludes that the Complex Langevin Method can serve as a complementary approach to variational methods for nuclear shell model calculations, particularly for large systems where variational ansatzes are difficult to construct.

Complementarity: Unlike variational methods, the CLM does not rely on a specific wave function ansatz and can, in principle, access finite-temperature physics.
Scalability: The computational cost scales as $O((N_s)^5 n_\beta)$ , which is polynomial. The authors suggest that while the sign problem becomes more severe in heavier nuclei, the validity region of the CLM may not be significantly affected by system size, making it a promising tool for heavy nuclei.
Benchmarking: The authors note that polynomial-scaling classical methods like the one investigated here could provide useful benchmarks for future quantum computing algorithms applied to nuclear structure problems.

The study acknowledges limitations: the direct CLM application is restricted to high temperatures, and the extrapolation method relies on the assumption that the energy is a smooth function of the interaction parameter $t$ in the valid region. For very low temperatures (large $\beta$ ), higher-order fitting ansatzes might be required, potentially reducing reliability. However, the successful demonstration of the extrapolation technique for small systems establishes a pathway for tackling larger, more complex nuclei.

More like this