TechLab (United States)

This study investigates the performance of different OpenAI GPT model versions (GPT-3.5 Turbo, GPT-4 Turbo, GPT-4o, o1-preview and GPT-5 at different reasoning effort values) in solving chemistry problems across various educational and conceptual difficulty levels. The evaluation, aimed at assessing both accuracy and reasoning quality, was based on a data set comprising 150 multiple-choice questions and 10 open-ended questions at the high school level, as well as 75 multiple-choice questions, 10 open-ended questions, and 100 stoichiometry exercises at the university level. The results reveal a clear trend of improvement in both accuracy and consistency with successive GPT model versions, with o1-preview and GPT-5 demonstrating the highest overall performance due to their reasoning capabilities. Error analysis shows that, while conceptual understanding is generally strong, computational mistakes remain frequent, particularly in tasks related to chemical equilibrium exercises and redox reaction balancing, though GPT-5 markedly reduced these errors compared to earlier models. Additionally, misinterpretations of questions requiring judgment or historical context have emerged as a recurring issue. While prompt formulation influences performance in specific contexts, such as redox balancing, the overall sophistication of the model appears to be the primary determinant of performance. These findings suggest that recent advancements in large language models have significantly enhanced their potential for chemistry education, although careful oversight remains necessary to address numerical inaccuracies and interpretative limitations.

Yilan Crater is a newly discovered impact structure in Northeast China. However, the impact process and its subsurface properties have not been properly investigated yet. Here we employed multiple seismic methods to gain insight on its subsurface structures, based on an ultra-dense observation with 220 seismic nodes. We observed a clear site amplification within the circular region of the crater generated by loose sediment and impact-fractured rocks. The resulting bowl-shaped structure beneath the Yilan crater was further revealed by the ambient noise tomography and Horizontal-to-Vertical Spectral Ratio analysis. Multiple solutions for the impact velocity and diameter were found by a parametric investigation, while the impact energy is around 1 E17 Joules, likely representing one of the most significant impact events in the last 80,000 years. These findings offer new insights into the Yilan impact event and its potential effects on the surrounding environment, highlighting the need for further multidisciplinary investigations.

We report a comprehensive evaluation of the toxin B (TcdB) vaccine adjuvanted with a dual Toll-like receptor ligand liposome adjuvant for Clostridioides difficile infection (CDI). The vaccine completely protected mice from a lethal infection. Compared to alum adjuvanted TcdB, it generated functionally superior systemic antibodies and supported strong memory B cell and gut IgA responses. This pharmaceutically acceptable adjuvant platform holds promise for developing a next-generation CDI vaccine.

Simultaneous light information and power transfer (SLIPT) enabled non-orthogonal multiple access (NOMA) based radio frequency (RF)/visible light communication (VLC) systems are considered a viable option for indoor wireless systems as they provide a higher data rate with higher energy efficiency. In this work, a grouping algorithm for a dual-hop cooperative NOMA-based multi-user RF/VLC system is introduced. The system consists of a light-emitting diode (LED) source, multiple VLC users randomly distributed inside the coverage area of the LED, and RF users located outside the coverage area of the LED. In grouping, all the VLC users are assigned as near users (NUs) and all the RF users are assigned as far users (FUs). In the first phase of the communication, the NU of a group decodes both its own information and the information of the group's FUs from the alternating current (AC) component of the received NOMA-coded optical signal. Also, it harvests the energy from the direct current (DC) component of the received NOMA-coded optical signal. In the second phase, the NU forward signal of FUs of the same group as NOMA coded RF signal by solely using harvested energy in the first phase. We propose a grouping algorithm by considering either the users' channel gain or signal-to-interference-plus-noise ratio (SINR) as a grouping parameter. Then, we obtained the achievable rate of the NUs and FUs. The presented numerical results demonstrate the effectiveness of the proposed grouping algorithm for the cooperative NOMA-based RF/VLC system with SLIPT capabilities.

Abstract The aim of this study was to evaluate the antimicrobial efficacy of an air gas soft jet CAP for its potential use in removing oral biofilms, given that plasma-based technologies have emerged as promising methods in periodontology. Two types of biofilms were developed, one by Streptococcus mutans UA 159 bacterial strain and the other by a complex mixture of saliva microorganisms isolated from a patient with periodontitis. This latter biofilm was characterized via Next Generation Sequencing to determine the main bacterial phyla. The CAP source was applied at a distance of 6mm for different time points. A statistically significant reduction of both CFU count and XTT was detected after 60s of CAP treatment, while the treatment for 120s resulted in both biofilms eradication. CLSM analysis supported CAP effectiveness in killing the microorganisms inside the biofilm and in reducing the thickness of the biofilm matrix. Cytotoxicity tests demonstrated the possible use of CAP without important side effects towards human gingival fibroblasts cell line. The current study showed that CAP treatment was able to eradicate preformed biofilms developed by both S. mutans and the complex mixture of saliva microorganisms, representing a potential innovative strategy to counteract oral pathogens responsible for periodontal diseases.

In the course of studying the biomechanical characteristics of hydrobionts there appears a need to apply the theory of similarity of living objects. Along with the theory of similarity, the theory of dimensions is observed. There is considered the multiphysical similarity of hydrobionts of the same species, based on the theory of physical similarity of A. A. Nedostup. One of the tasks of studying the mechanics and hydrodynamics of living systems (hydrobionts) in the aquatic environment is to automate the collection of statistics with subsequent visualization of the characteristics of hydrobionts using augmented reality technologies. Based on the similarity theory, the rules of physical similarity have been developed, similarity criteria have been obtained with respect to geometric, mechanical, hydroacoustic, hydrodynamic, thermodynamic, and optical characteristics of hydrobionts. An important component of similarity is to preserve the physical nature of the original phenomena in the model. Based on the scale of the multiphysical similarity of hydrobionts, the criteria for the similarity of hydrobionts are determined. There is considered the solution of the problems of describing the biomechanics of hydrobionts in recirculation aquaculture systems, where only one species of hydrobionts is grown at the same time. The importance of studies of biomechanics of aquatic organisms in certain habitat conditions, in our case in an artificial enclosed space, with artificial lighting, feeding, cleaning system, etc., is explained by the need to increase fish productivity and environmental friendliness of fish farming, which requires automation of the growing process. Automation tasks require a preliminary mathematical description of processes in order to build simulation models. In turn, for a mathematical description of the growth process of a growing object, it is necessary to derive mathematical dependencies linking the target characteristics of the hydrobionts with the initial data of breeding conditions and a set of characterizing parameters of the object itself, including behavioral ones.

The higher beam intensity available for Mu2e-II will require a substantially different target design. This paper discusses our recent advances in conceptual R&D for a Mu2e-II target station. The design is based on energy deposition and radiation damage simulations, as well as thermal and mechanical analyses, to estimate the survivability of the system. We considered rotated targets, fixed granular targets and a novel conveyor target with tungsten or carbon spherical elements that are circulated through the beam path. The motion of the spheres can be generated either mechanically or both mechanically and by a He gas flow. The simulations identified the conveyor target as the preferred approach, and that approach has been developed into a prototype. We describe this first prototype for the Mu2e-II target and report on its mechanical tests performed at Fermilab, which indicate the feasibility of the design, and discuss its challenges as well as suggest directions for further improvement.

We investigate a novel method to mitigate space charge effects of high intensity proton beams propagating in circular accelerators by means of trapping and controlling electrons generated from beam-induced residual gas ionization. This compensation method uses Coulomb repulsion force between a proton beam and electrons to mitigate self-space charge effects of the beam if it passes through a plasma column. The transverse electron-proton (e-p) instability in the plasma column is well controlled by the longitudinal magnetic field of a solenoid magnet and the bias voltages on electrodes. In this report, we will show simulation results how to control distributions of electrons and ions as well as that of the proton beam inside the column.