Scott Sommerfeldt Elected ASA President

When a sonic boom propagates through the atmospheric boundary layer, atmospheric turbulence distorts the waveform, producing stochastic variations over short distances. Accurate prediction of this variability is essential because turbulence affects both waveform characteristics and human-perception metrics used to evaluate supersonic overflights. During NASA's 2019 Carpet Determination in Entirety Measurements I (CarpetDIEM I) flight test campaign, Brigham Young University deployed a 120 m (400 ft) linear array with seven microphones to quantify turbulence-induced variability of sonic boom waveforms, spectra, and metrics. In this study, state-of-the-art models PCBoom and KZKFourier are combined with two weather models, CFSv2 and ERA5, to predict metric variability and compare results with CarpetDIEM I measurements. Depending on the metric and confidence interval considered, prediction success ranges between about 30% and 80%. For the perceived level metric, the prediction successes for the mean and standard deviation were 30%–48% and 52%–65% respectively, depending on the weather model. This compares favorably with the Sonic Booms in Atmospheric Turbulence prediction success results of 45% and 71% for the perceived level mean and standard deviation respectively, using data reported by Stout, Sparrow, and Blanc-Benon [(2021). J. Acoust. Soc. Am. 149, 3250–3260].

Machine learning interatomic potentials (MLIPs) enable atomistic simulations with near first-principles accuracy at substantially reduced computational cost, making them powerful tools for large-scale materials modeling. The accuracy of MLIPs is typically validated on a held-out dataset of ab initio energies and atomic forces. However, accuracy on these small-scale properties does not guarantee reliability for emergent, system-level behavior—precisely the regime where atomistic simulations are most needed, but for which direct validation is often computationally prohibitive. As a practical heuristic, predictive precision—quantified as inverse uncertainty—is commonly used as a proxy for accuracy, but its reliability remains poorly understood, particularly for system-level predictions. In this work, we systematically assess the relationship between predictive precision and accuracy in both in-distribution (ID) and out-of-distribution (OOD) regimes, focusing on ensemble-based uncertainty quantification (UQ) methods for neural network potentials, including bootstrap, dropout, random initialization, and snapshot ensembles. We use held-out cross-validation for ID assessment and calculate cold curve energies and phonon dispersion relations for OOD testing. These evaluations are performed across various carbon allotropes as representative test systems. We find that uncertainty estimates can behave counterintuitively in OOD settings, often plateauing or even decreasing as predictive errors grow. These results highlight fundamental limitations of current UQ approaches and underscore the need for caution when using predictive precision as a stand-in for accuracy in large-scale, extrapolative applications.

Machine-learning models that estimate crowd engagement at college football games could improve the home-field advantage, help stadiums serve fans better, and become a tool for other scientific studies. To prepare to create these models, college football games were acoustically characterized with feature identification and selection in mind. Data were collected from six Brigham Young University (BYU) football games and analyzed for patterns in level, spectra, and principal components. The probability distribution of sound pressure levels from an entire game was found to contain three distinct peaks that were found to be associated with (1) quiet during offense, (2) music being played over the stadium loudspeakers, and (3) distraction noise when BYU was on defense. Crowd-generated sounds were found to have the most energy in the 630 and 800 Hz bands, consistent with other crowd noise previously studied. Furthermore, a principal component analysis on the football games’ one-third octave spectra revealed that the principal components were similar to those from BYU basketball and volleyball games. Based on these acoustic characteristics, the levels of the 100 Hz, 800 Hz, and 2.5 kHz bands are proposed as features for estimating crowd engagement at college football games.

We provide experimental evidence for the absence of a magnetic moment in bulk RuO₂, a candidate altermagnetic material, by using a combination of Mössbauer spectroscopy, nuclear forward scattering, inelastic X-ray and neutron scattering, and density functional theory calculations. Using complementary Mössbauer and nuclear forward scattering, we determine the Ru magnetic hyperfine splitting to be negligible. Inelastic X-ray and neutron scattering-derived lattice dynamics of RuO₂ are compared to density functional theory calculations of varying flavors. Comparisons among theory with experiments indicate that electronic correlations, rather than magnetic order, are key in describing the lattice dynamics.

Transfer learning (TL) is used to predict source-receiver range in a laboratory tank with varying water temperature. The input data are single-hydrophone spectral levels from linear chirps over the 50–100 kHz band recorded at different ranges. Data measured in room temperature water are used to train one-dimensional convolutional neural networks. When the trained models are applied to data measured in warmer water, a bias is introduced. TL with a small dataset improves the generalization results at the new temperature, demonstrating the potential of TL to improve performance under variable environmental conditions.

Crowds at collegiate basketball games react acoustically to events on the court in many ways, including applauding, chanting, cheering, and making distracting noises. Acoustic features can be extracted from recordings of crowds at basketball games to train machine learning models to classify crowd reactions. Such models may help identify crowd mood, which could help players secure fair contracts, venues refine fan experience, and safety personnel improve emergency response services or to minimize conflict in policing. By exposing the key features in these models, feature selection highlights physical insights about crowd noise, reduces computational costs, and often improves model performance. Feature selection is performed using random forests and least absolute shrinkage and selection operator logistic regression to identify the most useful acoustic features for identifying and classifying crowd reactions. The importance of including short-term feature temporal histories in the feature vector is also evaluated. Features related to specific 1/3-octave band shapes, sound level, and tonality are highly relevant for classifying crowd reactions. Additionally, the inclusion of feature temporal histories can increase classifier accuracies by up to 12%. Interestingly, some features are better predictors of future crowd reactions than current reactions. Reduced feature sets are human-interpretable on a case-by-case basis for the crowd reactions they predict.