Neural Wavefunction Calculations of μ \upmu SR Spectra with Quantum Muons and Protons
Jamie Carr, Mathias Volkai, W. M. C. Foulkes, Andres Perez Fadon · 2026
Read Neural Wavefunction Calculations of μ \upmu SR Spectra with Quantum Muons and Protons by Jamie Carr, Mathias Volkai, W. M. C. Foulkes, Andres Perez Fadon (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-05453.txt.
Introduction
Carr and colleagues tackle muon hyperfine constants, needed to interpret muon spin spectroscopy, where density functional theory treats the muon as a fixed classical particle. They use variational quantum Monte Carlo with neural-network trial wavefunctions for muoniated methyl and ethyl radicals, treating the muon as fixed and as fully quantum. Fixed-muon values differ sharply from density functional results, while quantum-muon values sit closer to experiment after environmental effects are accounted for.
Catalog record
| CREATOR | Jamie Carr, Mathias Volkai, W. M. C. Foulkes, Andres Perez Fadon |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44750 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.05453
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Neural Wavefunction Calculations of μ \upmu SR Spectra with Quantum Muons and Protons Jamie Carr, Mathias Volkai, W. M. C. Foulkes, Andres Perez Fadon Abstract Accurate prediction of muon hyperfine constants is useful for interpreting muon spin spectroscopy data, yet standard methods such as density functional theory (DFT) compute muon-electron pair density functions, and thus hyperfine constants, by treating the muon as a fixed classical particle. This work uses the variational quantum Monte Carlo method with neural-network trial wavefunctions, a highly accurate and flexible approach recently applied to other quantum chemical problems. The muon can be treated classically or included in the many-particle electron-muon wavefunction, in which case the fully quantum mechanical pair density is obtained directly. We calculate muon hyperfine constants in muoniated methyl and ethyl radicals for both quantum mechanical and fixed classical muons. The hyperfine constants obtained from our fixed-muon calculations in the methyl and ethyl radicals differ from the corresponding DFT results significantly, highlighting the limitations of DFT even when the muon is treated classically. The results with quantum muons are closer to experiment after accounting for environmental effects. These findings suggest that explicitly calculating the quantum mechanical muon-electron pair density improves the accuracy of muon hyperfine constant predictions. I Introduction In a typical \upmu SR experiment, low-energy spin-polarized anti-muons (referred to as “muons” in most of the literature and throughout the rest of this paper) of charge +e are implanted into solids and molecules, where they act as sensitive probes of the local magnetic environment. 1 ; 36 ; 3 The angular distribution of the positrons produced by the resulting muon decay reveals details about the local spin density of materials. 1 ; 3 Frequently, the implanted muon interacts with an electron, causing the many-particle wavefunction to develop a muonium-like component: the muon binds to an electron to produce a Muonium (Mu) atom, chemically analogous to a light hydrogen atom. 19 ; 1 When a muonium atom binds to an unsaturated closed-shell molecule, the result is a muoniated radical with an unpaired electron and non-zero electronic and muonic spin densities. 19 ; 27 ; 36 For such radicals, key measurable quantities are the muon hyperfine constants, which represent the strength of the hyperfine interaction between the muon and electron spins. 19 ; 1 ; 36 ; 3 Computational predictions of these muon hyperfine constants can then be compared with experimental \upmu SR spectra to identify which chemical species are present in the sample. 41 Hyperfine constants can be measured using transverse field \upmu SR (TF- \upmu SR). 37 ; 38 ; 36 ; 19 In an applied transverse magnetic field, the muon and unpaired electron in a muoniated radical form a superposition of four spin eigenstates: |\uparrow_{\text{e}}\uparrow_{\upmu}\rangle , |\uparrow_{\text{e}}\downarrow_{\upmu}\rangle , |\downarrow_{\text{e}}\uparrow_{\upmu}\rangle , and |\downarrow_{\text{e}}\downarrow_{\upmu}\rangle . 19 ; 1 As the magnetic moment of the electron is much larger than that of the muon, the Zeeman interaction separates these into electron-up and electron-down pairs, while the muon Zeeman and hyperfine interactions determine the energy splitting between the muon spin states within each pair. 19 ; 1 These splittings manifest themselves in the Larmor precession of the muon spin as two distinct frequencies, which can be measured from the angular distribution of the decay positrons, emitted preferentially along the muon spin direction. 19 ; 1 As the hyperfine energy shift arises from the muon and electron spins being aligned or anti-aligned with the external magnetic field, the difference between the two precession frequencies cancels the muon Zeeman term and yields the isotropic muon hyperfine constant, A_{\upmu} . 1 ; 19 Computational quantum chemistry methods are often used to calculate muon hyperfine constants. 41 ; 40 ; 42 ; 1 ; 31 While high-accuracy methods such as coupled cluster can describe the ground state reliably, their computational cost is often prohibitive for large systems. 34 Density functional theory (DFT) provides a much faster alternative and, although generally less accurate, yields surprisingly good results. 42 ; 2 The isotropic muon hyperfine constant depends on the electron spin density at the location of the muon, 1 obtained by taking the zero-separation limit of the muon–electron pair density function (PDF). This dependence exposes a key limitation of DFT methods. Since DFT is a mean-field-like theory that produces one-electron wavefunctions, two-particle correlation functions such as the pair density are not directly available. 33 One solution to this problem is to treat the muon as a fixed classical particle in the Born-Oppenheimer approximation, in which case the muon-electron pair density is replaced by the electron density near the fixed muon. 40 ; 42 The results of DFT simulations with fixed muons are often good, but the mass of a muon is only 207 times that of an electron (about one ninth that of a proton) and their accuracy is far from guaranteed. In some systems, quantum effects such as zero-point motion are known to influence computed hyperfine constants significantly. 30 Approaches such as path-integral molecular dynamics are often used to capture nuclear quantum effects in a DFT framework and have also been used to sample distributions of muon positions in \upmu SR simulations, 42 but the muon-electron pair density is still obtained by treating the muon as fixed at each position. Two-component DFT, often used to study electron-positron systems, is able to treat both electrons and muons quantum mechanically and can compute the electron and muon densities, but remains unable to describe electron-muon correlations. 5 The zero-separation limit of the electron-muon pair density is written as the product of the electron and muon densities multiplied by a correlation factor that must be approximated. 16 Local-density-like approximations, in which the correlation factor is assumed to depend on the electron and muon densities (and perhaps their gradients) at the coincidence point, have been used to compute \upmu SR frequencies in some cases. 11 Over the past few years, the neural wavefunction approach shown itself able to describe molecules without muons at a level of accuracy comparable to high-level quantum chemistry methods such as coupled cluster with singles, doubles, and perturbative triples (CCSD(T)), whilst scaling more favorably to large system sizes. 15 ; 34 ; 18 ; 10 ; 17 Further, neural-network-based wavefunctions have been used to study quantum systems with multiple species of particles, such as positrons in molecules, 8 and dense hydrogen with quantum protons. 23 In this paper, we treat electron-muon correlations quantum mechanically by constructing a neural-network-based many-particle wavefunction ansatz that depends on the positions of electrons and muons, treating both on exactly the same footing. This yields more accurate estimates of muon hyperfine constants than approaches that treat the muon as a classical point particle. We compute the isotropic muon hyperfine constant in the muoniated methyl and ethyl radicals, comparing results obtained when the muon is treated either as a classical point particle or as a quantum particle within the many - body wavefunction. By comparing the predicted muon hyperfine constants with low-temperature experimental data, we show that a fully quantum treatment of the muon-electron system yields values closer to experiment. Furthermore, when accounting for systematic environmental effects, our approach produces results that are more accurate than standard DFT calculations. We hope that neural wavefunction methods will in future become an