Numerical study of the two-boson bound-state problem with and without partial-wave decomposition

This paper establishes a high-precision numerical benchmark for the two-boson bound-state problem by demonstrating the equivalence between standard partial-wave and vector-variable formulations while deriving analytical error bounds to validate algorithms for future few-body calculations.

The Gist

Imagine you are trying to predict how two dancers (bosons) will hold hands and spin together in a perfect, stable circle (a bound state). In the world of quantum physics, this is a fundamental problem. But to predict their dance, you need a very precise set of rules (mathematical equations) and a way to calculate their moves without making a single mistake.

This paper is essentially a high-stakes "stress test" for the computer programs physicists use to solve these dance routines. The author, Wolfgang Schadow, wants to make sure the software is so accurate that it can be trusted to handle even more complex dances involving three or four particles later on.

Here is the breakdown of the paper using simple analogies:

1. The Two Ways to Describe the Dance

The paper compares two different ways of calculating the dancers' moves:

• Method A: The "Layer Cake" Approach (Partial-Wave Decomposition)
  Imagine describing the dance by breaking it down into layers. You first look at the simplest spin (the "s-wave"), then the next spin, and so on. It's like peeling an onion or stacking layers of a cake.

  • Pros: It's very fast and easy if the dance is simple and slow (low energy).
  • Cons: If the dancers start spinning wildly fast (high energy), you need so many layers that the cake becomes impossible to manage. The math gets messy and slow.

• Method B: The "3D Map" Approach (Vector Variables)
  Instead of peeling layers, this method looks at the dance as a whole, 3D object. It tracks the dancers' positions and angles directly, without breaking them into layers.

  • Pros: It handles fast, wild spins much better. It's the "modern" way to do it.
  • Cons: It's computationally heavy and harder to check for errors because it's so complex.

The Goal: The author wanted to prove that Method B is just as accurate as the trusted Method A, even though they look at the problem differently.

2. The Test Subjects: Two Types of "Dance Partners"

To test the software, the author used two different types of "dance partners" (mathematical potentials):

• The Yamaguchi Partner (The Predictable Robot):
  This is a simplified, idealized partner. Its moves are so simple that we can calculate the exact answer with a pen and paper.

  • Why use it? It's the "gold standard." If the computer can't match the pen-and-paper answer for this simple partner, the software is broken.
  • The Bonus: Because it's so simple, the author could derive exact formulas to show exactly how much error is introduced if you stop the calculation too early (like cutting off the dance floor).

• The Malfliet-Tjon Partner (The Realistic Human):
  This partner is more like a real person. They have a "hard core"—they push away strongly if you get too close (repulsion) but pull together from a distance (attraction). This is much harder to calculate because the math gets "stiff" and jagged.

  • Why use it? Real atoms and nuclei behave like this. If the software works for this partner, it's ready for the real world.

3. The "Cut-Off" Problem (The Truncated Dance Floor)

Imagine you are trying to calculate the dance, but your computer screen is too small to show the whole dance floor. You have to draw a line and say, "We will only calculate moves within this circle."

• The Problem: If you cut the floor off too early, you miss important moves happening at the edge, and your answer is wrong.
• The Solution: The author used the "Predictable Robot" (Yamaguchi) to write down exact mathematical formulas that tell you exactly how wrong your answer is based on how small your dance floor is. This allows scientists to know exactly how big their "floor" needs to be to get a perfect result.

4. The Results: A Perfect Match

After running thousands of calculations, the author found:

• The Two Methods Agree: The "Layer Cake" method and the "3D Map" method produced results that were identical up to the 10th decimal place. This is like two different chefs following different recipes and ending up with a cake that tastes exactly the same.
• The Software is Ready: The "3D Map" method (which is needed for complex, fast-moving particles) is proven to be incredibly precise.
• Error Control: The author provided a "ruler" (the analytical formulas) that tells future scientists exactly how much error they have if they don't use enough computer power.

Why Does This Matter?

Think of this paper as the calibration certificate for a new, high-tech microscope.
Before scientists can use this microscope to look at complex things (like how three or four particles interact in a nucleus), they need to know the lens is perfect. This paper says, "We tested the lens on simple objects and complex objects, and it is accurate to within a billionth of a unit."

This gives physicists the confidence to use these new, faster "3D Map" methods to solve the much harder problems of the future, knowing their results won't be ruined by hidden math errors.

Technical Summary

1. Problem Statement

The accurate numerical solution of few-body systems (three or more particles) in atomic and nuclear physics relies on solving Faddeev or Faddeev-Yakubovsky equations. Validating these complex codes requires rigorous benchmarks on simpler subsystems.

• The Challenge: While the standard Partial-Wave Decomposition (PWD) method is efficient for low-energy bound states, it becomes computationally prohibitive for scattering at intermediate and high energies due to the slow convergence of the partial-wave series.
• The Alternative: Formulations based directly on vector variables (avoiding angular momentum expansion) are advantageous for high-energy regimes but require robust validation against established methods.
• The Goal: To establish a high-precision benchmark comparing the standard 1D PWD approach with a 2D vector-variable formulation for the two-boson bound-state problem, quantifying numerical errors, and validating the vector approach for future few-body applications.

2. Methodology

The study solves the Lippmann–Schwinger equation for the two-boson ground state using two complementary formulations:

A. Formulations

1. 1D Partial-Wave Approach: Projects the equation onto momentum states $|p l m\rangle$, reducing the problem to a set of uncoupled one-dimensional integral equations. For the systems studied, only the $l=0$ (s-wave) channel supports a bound state.
2. 2D Vector-Variable Approach: Projects directly onto vector momentum states $|\mathbf{p}\rangle$. This treats the equation as a two-dimensional integral equation involving the momentum magnitude ($p$) and the angle between momenta (represented by $x = \cos \theta$). This formulation mirrors the structure of three-body Faddeev equations.

B. Interaction Potentials

Two classes of potentials were used to test robustness:

1. Yamaguchi Potential: A rank-one separable potential. It allows for exact analytical solutions for binding energies and wave functions. Crucially, it enables the derivation of exact analytical expressions for systematic errors introduced by finite momentum ($p_{cut}$) and coordinate ($r_{cut}$) cut-offs.
2. Malfliet–Tjon (MT) Potentials: Local, non-separable Yukawa-type potentials with a strong short-range repulsive core (hard core). These pose significant numerical challenges due to high-momentum components and lack closed-form solutions, serving as a realistic test for numerical stability.

C. Numerical Techniques

• Discretization: Both formulations utilize Gauss–Legendre quadrature. The 2D formulation maps the semi-infinite momentum domain and the angular domain $[-1, 1]$ to discrete grids.
• Eigenvalue Solver: The problem is cast as finding the unit eigenvalue of a non-symmetric kernel matrix $K(E)$. Due to the large matrix sizes (up to $N \approx 10^4$ for 2D) and the presence of large negative eigenvalues from repulsive cores, standard power iteration fails. The authors employ the Restarted Arnoldi Method (RAM), a Krylov subspace technique, to efficiently isolate the physical bound state (the algebraically largest positive eigenvalue) without full matrix inversion.
• Fourier Transformation: Coordinate-space observables are obtained via numerical Fourier transformation of the momentum-space solutions, utilizing analytical derivatives of spherical Bessel functions to minimize numerical noise.

3. Key Contributions

1. Exact Analytical Error Bounds: For the Yamaguchi potential, the authors derived closed-form analytical expressions quantifying the systematic errors in kinetic and potential energy expectations caused by finite $p_{cut}$ and $r_{cut}$. This provides a rigorous tool to disentangle discretization errors from truncation effects.
2. High-Precision Benchmark: The study achieves a consistency level of $10^{-10}$ MeV between the 1D PWD and 2D vector-variable formulations.
3. Validation of Vector-Variable Algorithms: The work confirms that the 2D vector formulation correctly recovers the partial-wave limit (pure s-wave) and that higher partial waves ($l > 0$) are suppressed to machine precision ($< 10^{-30}$) when solving the full vector equation for s-wave bound states.
4. Robustness for Non-Separable Potentials: The methodology was successfully extended to local Malfliet–Tjon potentials, demonstrating that the vector-variable approach remains stable and accurate even for interactions with hard cores and slow convergence in momentum space.

4. Key Results

• Convergence:
  • Yamaguchi: Binding energies converged to 12 significant digits with $\sim 32$ quadrature points. The analytical error bounds matched numerical truncation errors perfectly.
  • Malfliet–Tjon: Required significantly higher cut-offs ($p_{cut} \sim 3000\text{--}4000 \text{ fm}^{-1}$) and dense meshes ($N_P \approx 2048$) due to the $1/r$ short-range behavior.
• Equivalence: Table X shows that for both Yamaguchi and MT potentials, the binding energies calculated via the 2D method (with and without explicit partial-wave restriction) match the 1D PWD results to at least 10 significant digits.
• Coordinate Space: The numerical Fourier transform of the momentum-space wave function yielded coordinate-space observables (binding energy, radius) consistent with analytical benchmarks within the expected truncation errors of $r_{cut}$.
• Error Analysis: The study quantified that for MT potentials, the momentum cut-off must be chosen in tandem with mesh density; otherwise, apparent stability in binding energy can be misleading.

5. Significance and Outlook

• Methodological Benchmark: This work provides a "gold standard" reference for validating vector-variable algorithms intended for complex three- and four-body calculations (Faddeev/Yakubovsky equations).
• Error Control: The derived analytical error formulas for the Yamaguchi model offer a template for estimating numerical uncertainties in momentum-space codes, allowing researchers to distinguish between physical truncation effects and numerical discretization errors.
• Future Applications: The validated 2D vector-momentum formulation is structurally similar to the Faddeev equations. The success of this benchmark suggests that the Restarted Arnoldi Method and the vector-variable approach are ready for deployment in high-precision few-body scattering and bound-state problems, particularly at energies where partial-wave expansions fail.

In summary, the paper successfully validates the vector-variable formulation as a numerically equivalent and robust alternative to partial-wave decomposition, providing the necessary tools and benchmarks to advance few-body nuclear physics calculations beyond standard low-energy approximations.