Stochastic many-body perturbation theory for high-order calculations

This paper introduces perturbation theory quantum Monte Carlo (PTQMC), a stochastic method that efficiently computes high-order many-body perturbative corrections by using random walkers to avoid exponential scaling, while also providing a new complexity metric ($e^S$) to better assess the validity of perturbative expansions.

The Gist

Imagine you are trying to predict the weather for a city, but instead of looking at clouds, you are trying to calculate the behavior of trillions of tiny particles (nucleons) inside an atomic nucleus. This is the job of nuclear physicists.

To do this, they use a mathematical tool called Perturbation Theory. Think of this like trying to guess the final shape of a sculpture by starting with a rough block of clay and adding tiny, tiny chips of clay one by one.

• Order 1: You add the first chip.
• Order 2: You add a second chip based on the first.
• Order 100: You add a hundredth chip.

The problem? As you go to higher and higher orders (adding more chips), the number of possible ways the chips can be arranged explodes. It's like trying to count every possible path a ant could take through a maze that keeps growing larger every second. Traditional computers get stuck because the math becomes too heavy, and they can't even tell if they are getting closer to the right answer or just spinning in circles.

The New Solution: PTQMC (The "Random Walker" Method)

The authors of this paper introduced a new method called PTQMC (Perturbation Theory Quantum Monte Carlo). Here is how they solved the problem, using a simple analogy:

1. The "Crowd" vs. The "Map"

• The Old Way (Deterministic): Imagine trying to draw a complete map of every single road in a massive city to find the shortest route. You need to list every street, every intersection, and every dead end. As the city grows, your map becomes too big to hold.
• The New Way (PTQMC): Instead of drawing the whole map, imagine releasing a thousand random walkers (like people with blindfolds) into the city.
  • These walkers start at the "Home" (the reference state).
  • They take random steps to connected houses (configurations).
  • If a path leads to a dead end or a bad neighborhood, the walker disappears.
  • If a path leads to a good spot, the walker splits into more walkers.
  • By watching where the crowd of walkers gathers after many steps, you can figure out the most important paths without ever needing to draw the whole map.

2. The "Fake Convergence" Trap

One of the biggest discoveries in this paper is a warning about trickery.
Sometimes, when you calculate the weather (or the energy of a nucleus), the numbers look like they are settling down and becoming stable. You might think, "Great! We are done! The answer is 50."
But then, you take one more step, and the answer jumps to 100, then 10, then 200. It was a fake convergence. The math was lying to you.

The authors showed that in the old way, you can't easily tell if you've hit a fake convergence. But with their "Random Walker" method, they can see the complexity of the solution.

• The Analogy: Imagine the walkers are spreading out over a floor.
  • If the walkers stay in a small, cozy circle, the solution is simple and stable.
  • If the walkers suddenly start spreading out to cover the entire floor, running into every corner, the system is chaotic and the "fake convergence" is a lie.

They created a new "Complexity Meter" (called the Effective Number of Configurations) that counts how many different places the walkers are visiting. If the walkers keep spreading out, they know the calculation is unreliable, even if the final energy number looks calm.

3. The "Magic Fix" (Resummation)

Even when the math is chaotic and the numbers are bouncing around wildly, the authors showed that if you use their method to get the data, you can use a mathematical "magic trick" (called Padé Resummation) to smooth out the chaos.

• Analogy: Imagine a radio signal that is full of static and noise. You can't hear the music. But if you have a very good recording of the noise pattern, you can use a filter to remove the static and hear the music clearly.
• PTQMC provides the clean recording of the "noise" (the high-order data), allowing them to extract the true, stable answer even when the system is extremely difficult.

Why This Matters

• Speed: It allows scientists to calculate things that were previously impossible (like 16th-order calculations) without needing a supercomputer the size of a building.
• Truth: It helps scientists know when they are being tricked by "fake" stable answers.
• Future: This method can be used to understand the most extreme environments in the universe, like the inside of neutron stars or the Big Bang, where particles interact in incredibly complex ways.

In short: The authors built a "random walker" simulation that lets scientists peek at the most complex parts of the atomic nucleus without getting lost in the math, and it gives them a special tool to tell the difference between a real answer and a mathematical illusion.

Technical Summary

1. Problem Statement

High-order ab initio calculations in nuclear physics face two primary bottlenecks:

• Exponential Scaling: Conventional Many-Body Perturbation Theory (MBPT) requires the explicit construction and summation of high-rank excitation operators and Hugenholtz diagrams. The computational cost scales at least as $O(N^{2n})$ (where $N$ is the number of single-particle states and $n$ is the perturbation order), making calculations beyond the 4th order infeasible for realistic nuclear systems.
• Convergence Assessment: In strongly correlated systems, the perturbative series often becomes asymptotic or divergent. Low-order results may appear to converge (pseudo-convergence) but remain far from the exact solution. Without access to higher-order terms, it is impossible to reliably assess the accuracy or uncertainty of low-order MBPT results.

2. Methodology: Perturbation Theory Quantum Monte Carlo (PTQMC)

The authors introduce PTQMC, a stochastic approach designed to compute high-order MBPT corrections without explicitly constructing diagrams.

• Core Concept: The $n$-th order MBPT energy correction is reformulated as a summation over all possible "routes" in configuration space starting and ending at the reference state ($|0\rangle \to |I_1\rangle \to \dots \to |I_{n-1}\rangle \to |0\rangle$).
• Stochastic Representation: Instead of tracking all configurations deterministically, PTQMC uses random walkers distributed over many-body configurations. The population of walkers on a specific configuration is proportional to its perturbative coefficient.
• Algorithm Steps:
  1. Initialization: Walkers are initialized based on the first-order amplitudes derived from the Hamiltonian perturbation $\hat{H}_1$.
  2. Spawning: Walkers propagate stochastically. A walker on configuration $|J\rangle$ spawns a new walker on a connected configuration $|I\rangle$ with a probability proportional to the matrix element $|\langle \Phi_I | \hat{H}_1 | \Phi_J \rangle| / |\Delta_{I0}|$, where $\Delta_{I0}$ is the Møller-Plesset energy denominator. The sign of the walker is determined by the sign of the matrix element and the parent walker's phase.
  3. Annihilation and Evaluation: Walkers on the same configuration are summed (annihilated) to update the weight distribution. The energy correction for order $n+1$ is estimated using the standard MBPT formula involving the reference state and the updated weights.
• Key Distinction: Unlike projection-based Quantum Monte Carlo (which evolves toward the ground state), PTQMC samples perturbative orders independently, allowing direct access to the order-resolved structure of the wave function.

3. Key Contributions

• Algorithm Development: The formulation of a stochastic algorithm that bypasses the exponential scaling of conventional diagrammatic MBPT, enabling calculations up to very high orders (demonstrated up to 16th order).
• Resummation Integration: The demonstration that PTQMC data can be effectively combined with series resummation techniques (specifically Padé approximants) to extract stable physical predictions from divergent or oscillatory series.
• New Diagnostic Metric: The proposal of the Effective Number of Configurations ($e_S$), derived from the Shannon entropy of the walker distribution, as a global measure of perturbative wave-function complexity.
• Benchmarking: Extensive validation of the method against exact Full Configuration Interaction (FCI) and deterministic MBPT results using the Richardson pairing model.

4. Key Results

• Accuracy: In the Richardson pairing model, PTQMC accurately reproduces exact MBPT coefficients up to the 16th order, even in regimes where the series is strongly divergent.
• Statistical Scaling: The method exhibits the expected statistical convergence behavior, with relative error scaling as $N_w^{-1/2}$ (where $N_w$ is the number of walkers).
• Resummation Success:
  • In regions where the raw perturbative series oscillates wildly or diverges (e.g., coupling strength $g > 1.5$), the raw series fails to predict the exact energy.
  • However, applying Padé resummation to the high-order PTQMC data yields stable estimates that converge to the exact FCI energy, outperforming several non-perturbative methods (like ADC(2) and IMSRG(2)) in strong-coupling regimes.
• Pseudo-Convergence Detection: The study reveals that low-order energies can appear converged while the wave function is still spreading uncontrollably.
  • The $e_S$ Metric: The effective number of configurations ($e_S$) serves as a superior indicator of validity.
    • Saturation: If $e_S$ saturates as the order increases, the perturbative expansion is well-behaved and amenable to resummation.
    • Divergence: If $e_S$ grows continuously, it signals uncontrolled wave-function spreading and unreliable perturbative descriptions, even if the energy seems stable.

5. Significance

• Overcoming the "Order Barrier": PTQMC provides a scalable route to access high-order perturbative physics in nuclear systems, a task previously blocked by the combinatorial explosion of diagrams.
• Reliability in Strong Correlation: By enabling the calculation of high-order terms, the method allows for the diagnosis of convergence issues that are invisible at low orders. This is crucial for nuclear matter and finite nuclei where correlations are strong.
• Wave Function Complexity: The introduction of $e_S$ shifts the focus from mere energy convergence to the structural stability of the wave function, offering a physically motivated criterion to distinguish between genuine convergence and spurious pseudo-convergence.
• Future Applications: The framework is model-agnostic (relying only on Hamiltonian matrix elements) and is poised for application to realistic nuclear Hamiltonians, including those with residual three-nucleon interactions, paving the way for more precise ab initio nuclear theory.