Self-learning Emulators and Eigenvector Continuation

Avik Sarkar; Dean Lee

doi:10.1103/physrevresearch.4.023214

Home/Archive/arXiv

arXiv · 2107.13449

Self-learning Emulators and Eigenvector Continuation

Avik Sarkar·Dean Lee

View original source ↗Read source document ↗Export metadata ↓

Abstract

Emulators that can bypass computationally expensive scientific calculations with high accuracy and speed can enable new studies of fundamental science as well as more potential applications. In this work we discuss solving a system of constraint equations efficiently using a self-learning emulator. A self-learning emulator is an active learning protocol that can be used with any emulator that faithfully reproduces the exact solution at selected training points. The key ingredient is a fast estimate of the emulator error that becomes progressively more accurate as the emulator is improved, and the accuracy of the error estimate can be corrected using machine learning. We illustrate with three examples. The first uses cubic spline interpolation to find the solution of a transcendental equation with variable coefficients. The second example compares a spline emulator and a reduced basis method emulator to find solutions of a parameterized differential equation. The third example uses eigenvector continuation to find the eigenvectors and eigenvalues of a large Hamiltonian matrix that depends on several control parameters.

Explore related subjects

nucl-th cs.LG cs.NA hep-lat math.NA

Keep this discovery

BibTeX RIS

Avik Sarkar, Dean Lee. 2021-07-28. Self-learning Emulators and Eigenvector Continuation. https://doi.org/10.1103/physrevresearch.4.023214

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Fission Modes and Fragment Shell Structures in $^{258}$Md$^*$ from Six-Dimensional Langevin Calculations

The fission of $^{258}$Md$^*$ is calculated in the excitation energy range of $E^*=6$--36 MeV using a six-dimensional Langevin equation. The calculated events are classified into two symmetric and two asymmetric fission modes based on the fragment mass and the quadrupole deformations of the two fragments at scission. The symmetric modes are separated by their total kinetic energies into the short (high TKE) and superlong (low TKE) modes, whereas the asymmetric modes differ in mass asymmetry. With increasing excitation energy, the yield of the short mode decreases, whereas the combined yield of the two asymmetric modes increases, as observed in the in-beam prompt-fission study of $^{258}$Md$^*$. From an analysis of the fragment shapes and associated single-particle levels, the short mode and the dominant asymmetric mode with the smaller mass asymmetry are found to involve a compact fragment characterized by deformed shell gaps at $Z=52$ and $N=84$, while the complementary fragments have different quadrupole deformations in the two modes.

nucl-th↗

arXiv2026

Classification of fission modes in $^{236}$U using a six-dimensional Langevin approach

Thermal neutron-induced fission of $^{235}$U is studied using a six-dimensional Langevin approach based on the Cassini shape parametrization. Scission events are classified into Asymmetric 1 (AS1), Asymmetric 2 (AS2), and Superlong (SL) fission modes by applying the $k$-means algorithm to the fragment mass and the quadrupole deformations of both fragments. For each mode, proton and neutron single-particle levels are calculated for representative fragments to examine their shell structures. The AS1 heavy fragment exhibits proton gaps at $Z=50$ and 52 and neutron gaps at $N=82$ and 84, whereas well-developed gaps appear at $Z=56$ and $N=88$ in the AS2 heavy fragment. The mass splits of AS1 and AS2 are close to those of the conventional Standard I and Standard II modes, respectively. However, the average total kinetic energy is lower for AS1 than for AS2, opposite to the conventional ordering of Standard I and Standard II. This reversal reflects the more elongated shape of the AS1 light fragment. The SL mode is conventionally interpreted in terms of macroscopic liquid-drop effects, whereas the pronounced proton shell gap at $Z=46$ suggests that proton shell effects also contribute to the elongated symmetric configuration. The classification based on fragment mass and the quadrupole deformations of both fragments provides a basis for distinguishing fission modes and examining the corresponding fragment shell structures at scission.

nucl-th↗

arXiv2026

Gogny interaction from beginnings to current challenges

The main goal of the present review article is to gather for the first time various facets of the phenomenological effective Gogny interaction which was originally proposed in the 70's. This involves both nuclear phenomena of interest that led to its creation and evolution as well as highly technical aspects that led the objectives to be achieved. With this in mind, we propose a discussion structured around four points. After a general introduction, the history and philosophy of the Gogny interaction is exposed. In particular, one highlights an intuitive way of guiding the determination of the parameters of the phenomenological interaction with the results obtained from a realistic interaction using Hartree-Fock calculations and second order corrections and a G-matrix. One also shows that physical phenomena such as pairing or fission were essential to improve the parameterization. The evolution of the original analytical form over the years is also discussed. The second point concern the emulator that was used for the generation of parameterizations. Its modifications, consistent with the evolution of the analytical form, are given. Other fitting procedures, more recent, are also evoked. The third key point is dedicated to the role of the nuclear matter in the fitting process and the acceptance of a parameterization. The objective of the last key point is to highlight some results obtained with the Gogny interaction in nuclear structure, fission and reactions that have allowed to interpret experimental data.

nucl-th↗