From credible shell model interactions to neutron-capture uncertainties

This paper presents the first uncertainty-quantified neutron-capture cross section for $^{27}$Al, demonstrating that the USDBUQ500 shell model interaction yields nuclear level densities and radiative strength functions with constant uncertainties of 6% and 9%, respectively, which propagate to a 5–25% non-Gaussian uncertainty in the cross section.

The Gist

Imagine you are trying to predict the weather in a city you've never visited. You have a powerful computer model, but you don't know exactly how the wind, temperature, and humidity interact perfectly. If you run the model once, you get one forecast. But what if you ran it 500 times, slightly tweaking the wind speed or humidity each time? You'd get a range of possible outcomes, giving you a much better idea of how confident you can be in your prediction.

This is exactly what the scientists in this paper did, but instead of weather, they were predicting how atomic nuclei behave when hit by neutrons.

Here is the breakdown of their work using simple analogies:

1. The Big Picture: Predicting the Unpredictable

In the world of nuclear physics (which powers stars and nuclear reactors), scientists need to know how likely a specific atom is to "catch" a neutron. This is called a neutron-capture cross-section.

Traditionally, to predict this, scientists use a "statistical recipe" (called Hauser-Feshbach theory). Think of this recipe like a cake mix. To bake the cake, you need ingredients:

• Nuclear Level Densities (NLDs): How crowded the "dance floor" is with energy states.
• Radiative Strength Functions (RSFs): How easily the nucleus can emit light (energy) to calm down.

For decades, scientists estimated these ingredients using smooth, mathematical curves fitted to old data. It was like guessing the weight of an egg by looking at a picture of a chicken. It worked okay, but no one knew how wrong those guesses might be.

2. The New Approach: The "Ensemble" of Interactions

The authors, Oliver and Konstantinos, decided to stop guessing and start calculating from first principles using the Shell Model.

Think of the Shell Model as a massive, detailed Lego set. Instead of guessing the weight of the egg, they build the chicken out of Legos, piece by piece, based on the fundamental laws of physics.

However, there's a catch: We don't know the exact size of every single Lego brick (the interaction between particles). So, instead of using one set of bricks, they created 500 slightly different sets (an "ensemble").

• Some sets have slightly larger bricks.
• Some have slightly smaller bricks.
• Some have slightly different colors.

They ran their simulation 500 times, once for each set of bricks. This gave them 500 different predictions for how the nucleus behaves.

3. The Results: Finding the "Uncertainty Budget"

By comparing these 500 predictions, they could finally put a number on the uncertainty.

• The "Crowded Dance Floor" (NLDs): They found that their calculation of how crowded the energy levels are was very consistent. No matter which of the 500 brick sets they used, the result only varied by about 6%.
• The "Light Emission" (RSFs): The calculation for how the nucleus emits energy was slightly less consistent, varying by about 9%.

When they fed these numbers into their "recipe" to predict the final neutron capture, the final result wasn't a single number. It was a range of 5% to 25% uncertainty.

4. The Surprise: It's Not a Bell Curve

Here is the most interesting part. In science, when you have a bunch of uncertainties, you usually expect them to form a "Bell Curve" (a nice, symmetrical hill where the average is in the middle).

But when they looked at the results for Aluminum-27, the distribution was weird and lopsided.

• Analogy: Imagine you are guessing the price of a house. Most guesses are around \500k. But occasionally, someone guesses \1 million, and the distribution gets pulled hard to one side.
• Why it matters: This "non-Gaussian" shape means that standard statistical tools might fail. You can't just say "we are 95% sure the answer is X." You have to be much more careful because the risks aren't symmetrical.

5. Why Does This Matter?

This paper is a "proof of concept." It's like the first time someone built a weather model that told you not just "it will rain," but "there is a 20% chance of a flood, and here is exactly why."

• For Stars: It helps us understand how elements are forged in the universe (nucleosynthesis).
• For Technology: It helps engineers design safer nuclear reactors and better waste storage by knowing exactly how much "wiggle room" they have in their safety calculations.

The Bottom Line

The authors successfully replaced old, fuzzy guesses with a rigorous, computer-generated "ensemble" of possibilities. They showed that even with our best physics models, there is still a 6% to 25% margin of error in predicting how aluminum reacts to neutrons.

More importantly, they proved that we can now quantify that error. Instead of just saying "we don't know," they can now say, "We know the answer is likely here, but the uncertainty looks like this specific shape." This is a giant leap toward making nuclear physics more reliable and predictable.

Technical Summary

1. Problem Statement

Nuclear astrophysics and nuclear technology rely on accurate predictions of neutron-induced reaction rates, particularly in the fast energy region (hundreds of keV to several MeV) relevant to the $r$-process and reactor design. Currently, the Hauser-Feshbach (HF) statistical theory is the standard tool for these calculations. However, HF theory requires statistical nuclear structure inputs—specifically Nuclear Level Densities (NLDs) and Radiative Strength Functions (RSFs)—which are traditionally modeled using phenomenological analytic functions fitted to experimental data.

The critical gap identified by the authors is the lack of credible uncertainty estimates for these inputs, especially in regions where experimental data is scarce. While microscopic models (like the nuclear shell model) offer a path to replace phenomenological models, they have not yet been fully integrated into reaction codes with rigorous uncertainty quantification (UQ).

2. Methodology

The authors propose a framework to bridge microscopic nuclear structure theory with macroscopic reaction theory by propagating uncertainties from shell model interactions to reaction cross sections.

• Uncertainty-Quantified Shell Model (SM):

  • Instead of using a single "best-fit" interaction, the authors utilize the USDBUQ500 ensemble. This is a set of 560 realizations of the effective shell model interaction matrix elements (single-particle energies and two-body matrix elements) derived from a Bayesian analysis of fitting uncertainties [Ref. 4].
  • By sampling this ensemble, they generate a distribution of nuclear wave functions and energies rather than a single deterministic solution.

• Calculation of Statistical Properties:

  • Nuclear Level Densities (NLDs): For each of the 560 interaction realizations, the authors compute the lowest $M$ energy levels for $^{27}\text{Al}$ and $^{28}\text{Al}$ using the Large Scale Shell Model (LSSM) within the $sd$-valence space. They explicitly bin these levels to generate NLDs. Since the $sd$-space only yields positive parity states, they approximate the total NLD by adding a shifted positive-parity density to account for negative parity states (using experimental energy shifts).
  • Radiative Strength Functions (RSFs): They compute the M1 (magnetic dipole) RSF directly from the shell model wave functions using the Bartholomew definition. For the E1 (electric dipole) component, which cannot be fully captured in a single-parity model space, they use a standard Systematic Modified Lorentzian (SMLO) model.

• Reaction Calculation:

  • The resulting distributions of NLDs and RSFs are fed as inputs into the YAHFC Hauser-Feshbach reaction code.
  • This generates a distribution of neutron-capture cross sections for $^{27}\text{Al}$, allowing for the statistical analysis of the final reaction uncertainty.

3. Key Contributions

• First UQ Neutron-Capture Cross Section: This work presents the first neutron-capture cross section for $^{27}\text{Al}$ with uncertainties derived directly from a microscopic shell model interaction ensemble.
• Integration of Micro and Macro: It successfully demonstrates a workflow connecting the uncertainty in shell model Hamiltonian parameters to the final reaction cross section, moving beyond phenomenological error estimates.
• Discovery of Non-Gaussian Behavior: The study reveals that the resulting distribution of cross sections is non-Gaussian and asymmetric, challenging the common assumption in many uncertainty propagation studies that errors follow a normal distribution.
• Correlation Analysis: The authors provide covariance information between isotopes ($^{27}\text{Al}$ and $^{28}\text{Al}$), showing a moderate correlation (Pearson $R \approx 0.6$) in level spacings at neutron separation energies.

4. Key Results

• Input Uncertainties:

  • The USDBUQ500 interaction predicts NLDs with a constant uncertainty of 6% across the relevant energy range.
  • The M1 component of the RSF exhibits a constant uncertainty of 9%.
  • Note: Uncertainties appear to increase at low energies (<3 MeV) for NLDs and high energies (>10 MeV) for RSFs, but the authors attribute these to artifacts of binning and limited transition counts rather than fundamental interaction uncertainties.

• Output Uncertainties (Neutron Capture):

  • The propagated uncertainties result in a 5% to 25% uncertainty band for the neutron-capture cross section of $^{27}\text{Al}$.
  • The distribution of cross sections is asymmetric. The histogram of cross sections between 2.7 and 3.2 MeV shows a marked deviation from a Gaussian shape.

• Comparison with Data:

  • The calculated cross sections and their uncertainty bands are consistent with recent experimental measurements [19, 20] and the ENDF/B-VIII.0 evaluation, validating the approach while highlighting the necessity of including microscopic uncertainties.

5. Significance and Future Outlook

• Astrophysical Impact: Providing robust, uncertainty-quantified reaction rates is crucial for improving the accuracy of stellar nucleosynthesis simulations (e.g., $r$-process). This work moves the field toward "self-consistent" descriptions where the nuclear structure theory and reaction theory share a common uncertainty budget.
• Limitations: The authors acknowledge that the current model space ($sd$-shell) is finite. It misses configurations outside the valence space (e.g., $1p-1h$ or $2p-2h$ excitations breaking the $^{16}\text{O}$ core) and cannot fully model E1 strength. These "model defects" represent a source of systematic error not captured by the parametric uncertainty of the interaction.
• Future Work: The methodology paves the way for large-scale studies of nuclear physics impacts on nucleosynthesis. Future efforts must address model space limitations and incorporate uncertainty quantification for optical model potentials (currently treated as fixed in this study) to achieve a complete uncertainty budget.

In summary, this paper establishes a proof-of-concept for using uncertainty-quantified shell model interactions to generate credible, microscopic-based uncertainties for nuclear reaction rates, revealing complex statistical behaviors (non-Gaussianity) that were previously unaccounted for in standard evaluations.