More Like this

1. Learning to simulate high energy particle collisions from unlabeled data

   https://doi.org/10.1038/s41598-022-10966-7  

   Howard, Jessica N. ; Mandt, Stephan ; Whiteson, Daniel ; Yang, Yibo ( December 2022 , Scientific Reports)

   Abstract In many scientific fields which rely on statistical inference, simulations are often used to map from theoretical models to experimental data, allowing scientists to test model predictions against experimental results. Experimental data is often reconstructed from indirect measurements causing the aggregate transformation from theoretical models to experimental data to be poorly-described analytically. Instead, numerical simulations are used at great computational cost. We introduce Optimal-Transport-based Unfolding and Simulation (OTUS), a fast simulator based on unsupervised machine-learning that is capable of predicting experimental data from theoretical models. Without the aid of current simulation information, OTUS trains a probabilistic autoencoder to transform directly between theoretical models and experimental data. Identifying the probabilistic autoencoder’s latent space with the space of theoretical models causes the decoder network to become a fast, predictive simulator with the potential to replace current, computationally-costly simulators. Here, we provide proof-of-principle results on two particle physics examples, Z -boson and top-quark decays, but stress that OTUS can be widely applied to other fields. 

   more » « less
2. Statistical (n,$$\gamma $$) cross section model comparison for short-lived nuclei

   https://doi.org/10.1140/epja/s10050-023-00920-0  

   Lewis, R. ; Couture, A. ; Liddick, S. N. ; Spyrou, A. ; Bleuel, D. L. ; Campo, L. Crespo ; Crider, B. P. ; Dombos, A. C. ; Guttormsen, M. ; Kawano, T. ; et al ( March 2023 , The European Physical Journal A)

   Neutron-capture cross sections of neutron-rich nuclei are calculated using a Hauser–Feshbach model when direct experimental cross sections cannot be obtained. A number of codes to perform these calculations exist, and each makes different assumptions about the underlying nuclear physics. We investigated the systematic uncertainty associated with the choice of Hauser-Feshbach code used to calculate the neutron-capture cross section of a short-lived nucleus. The neutron-capture cross section for$$^{73}\hbox {Zn}$$${}^{73}\text{Zn}$(n,$$\gamma $$$\gamma$)$$^{74}\hbox {Zn}$$${}^{74}\text{Zn}$was calculated using three Hauser-Feshbach statistical model codes: TALYS, CoH, and EMPIRE. The calculation was first performed without any changes to the default settings in each code. Then an experimentally obtained nuclear level density (NLD) and$$\gamma $$$\gamma$-ray strength function ($$\gamma \hbox {SF}$$$\gamma \text{SF}$) were included. Finally, the nuclear structure information was made consistent across the codes. The neutron-capture cross sections obtained from the three codes are in good agreement after including the experimentally obtained NLD and$$\gamma \hbox {SF}$$$\gamma \text{SF}$, accounting for differences in the underlying nuclear reaction models, and enforcing consistent approximations for unknown nuclear data. It is possible to use consistent inputs and nuclear physics to reduce the differences in the calculated neutron-capture cross section from different Hauser-Feshbach codes. However, ensuring the treatment of the input of experimental data and other nuclear physics are similar across multiple codes requires a careful investigation. For this reason, more complete documentation of the inputs and physics chosen is important.

    

   more » « less
3. Restitution Payment and Recidivism: An Experimental Analysis

   https://doi.org/10.1111/1745-9133.12401  

   Ruback, R. Barry ; Knoth, Lauren K. ; Gladfelter, Andrew S. ; Lantz, Brendan ( December 2018 , Criminology & Public Policy)

   Restitution is court‐ordered payment by an offender to compensate a crime victim for direct tangible losses resulting from the crime. Making regular restitution payments, it has been argued, may serve as a continual reminder to offenders of the harm they caused and their need to restore justice. In this study, we examined the recidivism of offenders who were delinquent in paying restitution and who had participated in a prior experiment aimed at testing whether receiving monthly letters would increase their payment of restitution. Findings from that experiment revealed that offenders who had received letters containing information about the economic sanctions they had paid and what they still owed paid a significantly larger proportion of the restitution owed and made significantly more monthly payments than did offenders in the other three experimental conditions. Results of the present follow‐up study revealed that even though recidivists (as measured by a new arrest) and nonrecidivists did not differ significantly in the amount of restitution they owed, nonrecidivists were more likely to have paid something toward the restitution they owed and more likely to have made more monthly payments. Statistical tests indicated that payment mediated the relationship between the experimental manipulation of information and observed recidivism, suggesting that the experimental manipulation of information led to more payments that, in turn, led to less recidivism.

   The findings from the prior experiment indicated that offenders who had received information about the economic sanctions they had paid and what they still owed were more likely to make payments. In that study, scholars concluded that giving offenders this information was cost‐effective, in that for every dollar spent, more than $6 in restitution was received. The results of this follow‐up analysis of recidivism demonstrate that the cost‐effectiveness of providing information is greater than previously indicated. At a cost of $64.19 per day for jail in Pennsylvania, the difference of 46.4 days between offenders who received information and those in the other experimental conditions could be as much as $2,978.42 per offender, less the costs of monitoring payments.

    

   more » « less
4. Scientific machine learning framework to understand flash graphene synthesis

   Sattari, K. ; Beckham, J. L. ; Eddy, L. ; Wyss, K. M. ; Byfield, R. ; Tour, J. M. ; Lin, J. ( January 2023 , arXivorg)

   Flash Joule heating (FJH) is a far-from-equilibrium (FFE) processing method for converting low-value carbon-based materials to flash graphene (FG). Despite its promise in scalability and performance, attempts to explore the reaction mechanism have been limited due to complexity involved in the FFE process. Data-driven machine learning (ML) models effectively account for this complexity, but the model training requires considerable amount of experimental data. To tackle this challenge, we constructed a scientific ML (SML) framework trained by using both direct processing variables and indirect, physics-informed variables to predict the FG yield. The indirect variables include current-derived features (final current, maximum current, and charge density) predicted from the proxy ML models and reaction temperatures simulated from multi-physics modeling. With the combined indirect features, the final ML model achieves an average R2 score of 0.81 ± 0.05 and an average RMSE of 12.1% ± 2.0% in predicting the FG yield, which is significantly higher than the model trained without them (R2 of 0.73 ± 0.05 and an RMSE of 14.3% ± 2.0%). Feature importance analysis validates the key roles of these indirect features in determining the reaction outcome. These results illustrate the promise of this SML to elucidate FFE material synthesis outcomes, thus paving a new avenue to processing other datasets from the materials systems involving the same or different FFE processes. 

   more » « less
5. How Augmented Reality Affects Collaborative Learning of Physics: a Qualitative Analysis

   Unahalekhaka, A. ; Radu, I. ; Schneider, B. ( July 2019 , Computer-supported collaborative learning)

   Augmented reality (AR) is a powerful visualization tool to support learning of scientific concepts across learners of various ages. AR can make information otherwise invisible visible in the physical world in real-time. In this study, we are looking at a subset of data from a larger study (N=120), in which participant pairs interacted with an augmented sound producing speaker. We explored the learning behaviors in eight pairs of learners (N=16) who participated in an unstructured physics activity under two conditions: with or without AR. Comparing behaviors between the two experimental conditions, we found that AR affected learning in four different ways: participants in the AR condition (1) learned more about visual concepts (ex: magnetic field structures) but learned less about nonvisual content (ex: relationship between electricity and physical movement); (2) stopped exploring the system faster than NonAR participants; (3) used less aids in exploration and teaching; and (4) spent less time in teaching their collaborators. We discuss implications of those results for designing collaborative learning activities with augmented reality. 

   more » « less