-
A novel emulator for fission event generators
Authors:
Karl Daningburg,
A. E. Lovell,
R. O'Shaughnessy
Abstract:
A wide variety of emulators have been developed for nuclear physics, particularly for use in quantifying and propagating parametric uncertainties to observables. Most of these methods have been used to emulate structure observables, such as energies, masses, or separation energies, or reaction observables, such as cross sections. Rarely, if ever, have event generators for theory models been emulat…
▽ More
A wide variety of emulators have been developed for nuclear physics, particularly for use in quantifying and propagating parametric uncertainties to observables. Most of these methods have been used to emulate structure observables, such as energies, masses, or separation energies, or reaction observables, such as cross sections. Rarely, if ever, have event generators for theory models been emulated. Here, we describe one such novel emulator for the fission fragment decay code, $\texttt{CGMF}$, which calculates the emission of prompt neutrons and $γ$ rays from fission fragments. The emulator described in this work uses a combination of a noisy emission model and mixture density network to model the neutron and $γ$-ray multiplicities and energies. In this manuscript, we display the power of this type of emulator for not only modeling average prompt fission observables but also correlations between fission fragment initial conditions and these observables, using both neutron-induced and spontaneous fission reactions.
△ Less
Submitted 24 September, 2024;
originally announced September 2024.
-
Bayesian averaging for ground state masses of atomic nuclei in a Machine Learning approach
Authors:
M. R. Mumpower,
M. Li,
T. M. Sprouse,
B. S. Meyer,
A. E. Lovell,
A. T. Mohan
Abstract:
We present global predictions of the ground state mass of atomic nuclei based on a novel Machine Learning (ML) algorithm. We combine precision nuclear experimental measurements together with theoretical predictions of unmeasured nuclei. This hybrid data set is used to train a probabilistic neural network. In addition to training on this data, a physics-based loss function is employed to help refin…
▽ More
We present global predictions of the ground state mass of atomic nuclei based on a novel Machine Learning (ML) algorithm. We combine precision nuclear experimental measurements together with theoretical predictions of unmeasured nuclei. This hybrid data set is used to train a probabilistic neural network. In addition to training on this data, a physics-based loss function is employed to help refine the solutions. The resultant Bayesian averaged predictions have excellent performance compared to the testing set and come with well-quantified uncertainties which are critical for contemporary scientific applications. We assess extrapolations of the model's predictions and estimate the growth of uncertainties in the region far from measurements.
△ Less
Submitted 17 April, 2023;
originally announced April 2023.
-
Nuclear data activities for medium mass and heavy nuclei at Los Alamos
Authors:
M. R. Mumpower,
T. M Sprouse,
T. Kawano,
M. W. Herman,
A. E. Lovell,
G. W. Misch,
D. Neudecker,
H. Sasaki,
I. Stetcu,
P. Talou
Abstract:
Nuclear data is critical for many modern applications from stockpile stewardship to cutting edge scientific research. Central to these pursuits is a robust pipeline for nuclear modeling as well as data assimilation and dissemination. We summarize a small portion of the ongoing nuclear data efforts at Los Alamos for medium mass to heavy nuclei. We begin with an overview of the NEXUS framework and s…
▽ More
Nuclear data is critical for many modern applications from stockpile stewardship to cutting edge scientific research. Central to these pursuits is a robust pipeline for nuclear modeling as well as data assimilation and dissemination. We summarize a small portion of the ongoing nuclear data efforts at Los Alamos for medium mass to heavy nuclei. We begin with an overview of the NEXUS framework and show how one of its modules can be used for model parameter optimization using Bayesian techniques. The mathematical framework affords the combination of different measured data in determining model parameters and their associated correlations. It also has the advantage of being able to quantify outliers in data. We exemplify the power of this procedure by highlighting the recently evaluated 239-Pu cross section. We further showcase the success of our tools and pipeline by covering the insight gained from incorporating the latest nuclear modeling and data in astrophysical simulations as part of the Fission In R-process Elements (FIRE) collaboration.
△ Less
Submitted 21 October, 2022;
originally announced October 2022.
-
Horizons: Nuclear Astrophysics in the 2020s and Beyond
Authors:
H. Schatz,
A. D. Becerril Reyes,
A. Best,
E. F. Brown,
K. Chatziioannou,
K. A. Chipps,
C. M. Deibel,
R. Ezzeddine,
D. K. Galloway,
C. J. Hansen,
F. Herwig,
A. P. Ji,
M. Lugaro,
Z. Meisel,
D. Norman,
J. S. Read,
L. F. Roberts,
A. Spyrou,
I. Tews,
F. X. Timmes,
C. Travaglio,
N. Vassh,
C. Abia,
P. Adsley,
S. Agarwal
, et al. (140 additional authors not shown)
Abstract:
Nuclear Astrophysics is a field at the intersection of nuclear physics and astrophysics, which seeks to understand the nuclear engines of astronomical objects and the origin of the chemical elements. This white paper summarizes progress and status of the field, the new open questions that have emerged, and the tremendous scientific opportunities that have opened up with major advances in capabilit…
▽ More
Nuclear Astrophysics is a field at the intersection of nuclear physics and astrophysics, which seeks to understand the nuclear engines of astronomical objects and the origin of the chemical elements. This white paper summarizes progress and status of the field, the new open questions that have emerged, and the tremendous scientific opportunities that have opened up with major advances in capabilities across an ever growing number of disciplines and subfields that need to be integrated. We take a holistic view of the field discussing the unique challenges and opportunities in nuclear astrophysics in regards to science, diversity, education, and the interdisciplinarity and breadth of the field. Clearly nuclear astrophysics is a dynamic field with a bright future that is entering a new era of discovery opportunities.
△ Less
Submitted 16 May, 2022;
originally announced May 2022.
-
Physically Interpretable Machine Learning for nuclear masses
Authors:
M. R. Mumpower,
T. M. Sprouse,
A. E. Lovell,
A. T. Mohan
Abstract:
We present a novel approach to modeling the ground state mass of atomic nuclei based directly on a probabilistic neural network constrained by relevant physics. Our Physically Interpretable Machine Learning (PIML) approach incorporates knowledge of physics by using a physically motivated feature space in addition to a soft physics constraint that is implemented as a penalty to the loss function. W…
▽ More
We present a novel approach to modeling the ground state mass of atomic nuclei based directly on a probabilistic neural network constrained by relevant physics. Our Physically Interpretable Machine Learning (PIML) approach incorporates knowledge of physics by using a physically motivated feature space in addition to a soft physics constraint that is implemented as a penalty to the loss function. We train our PIML model on a random set of $\sim$20\% of the Atomic Mass Evaluation (AME) and predict the remaining $\sim$80\%. The success of our methodology is exhibited by the unprecedented $σ_\textrm{RMS}\sim186$ keV match to data for the training set and $σ_\textrm{RMS}\sim316$ keV for the entire AME with $Z \geq 20$. We show that our general methodology can be interpreted using feature importance.
△ Less
Submitted 20 March, 2022;
originally announced March 2022.
