Search

Large language models (LLMs) are typically aligned to be harmless to humans. Unfortunately, recent work has shown that such models are susceptible to automated jailbreak attacks that induce them to generate harmful content. More recent LLMs often incorporate an additional layer of defense, a Guard Model, which is a second LLM that is designed to check and moderate the output response of the primary LLM. Our key contribution is to show a novel attack strategy, PRP, that is successful against several open-source (e.g., Llama 2) and closed-source (e.g., GPT 3.5) implementations of Guard Models. PRP leverages a two step prefix-based attack that operates by (a) constructing a universal adversarial prefix for the Guard Model, and (b) propagating this prefix to the response. We find that this procedure is effective across multiple threat models, including ones in which the adversary has no access to the Guard Model at all. Our work suggests that further advances are required on defenses and Guard Models before they can be considered effective.

After neutrinos decouple from the photon bath, they can populate a thermal dark sector. If this occurs at a temperature above ~100 keV, this can have measurable impacts on light element abundances. We calculate light element abundances in this scenario, studying the impact from rapid cooling of the Standard Model neutrinos, and from an increase in the number of relativistic degrees of freedom $N_{\rm{eff}}$, which can occur in the presence of a mass threshold. We incorporate these changes in the publicly available BBN code PRIMAT, using the reaction networks from PRIMAT and from the BBN code PArthENoPE, to calculate Y$_{\rm{P}}$ and D/H. We provide limits from the two different reaction networks as well as with expanded errors to include both results. If electron neutrinos significantly participate in the cooling, we find limits down to temperatures as low as 100 keV. If electron neutrinos are weakly participating (for instance if only the mass eigenstate $\nu_3$ equilibrates), cooling places no limits. However, if the dark sector undergoes a "step" in $N_{\rm{eff}}$, there can be additional, $\omega_b$-dependent constraints. These limits can vary from strong (for low values of $\omega_b$) to a mild preference for new physics (for high values of $\omega_b$). Future analyses including upcoming CMB data should improve these limits., Comment: 6+6 pages

The structure and composition of simple ices can be modified during stellar evolution by protostellar heating. Key to understanding the involved processes are thermal and chemical tracers that can diagnose the history and environment of the ice. The 15.2 $\mu$m bending mode of $^{12}$CO$_2$ has proven to be a valuable tracer of ice heating events but suffers from grain shape and size effects. A viable alternative tracer is the weaker $^{13}$CO$_2$ isotopologue band at 4.39 $\mu$m which has now become accessible at high S/N with the $\textit{James Webb}$ Space Telescope (JWST). We present JWST NIRSpec observations of $^{13}$CO$_2$ ice in five deeply embedded Class 0 sources spanning a wide range in luminosities (0.2 - 10$^4$ L$_{\odot}$ ) taken as part of the Investigating Protostellar Accretion Across the Mass Spectrum (IPA) program. The band profiles vary significantly, with the most luminous sources showing a distinct narrow peak at 4.38 $\mu$m. We first apply a phenomenological approach and show that a minimum of 3-4 Gaussian profiles are needed to fit the $^{13}$CO$_2$ absorption feature. We then combine these findings with laboratory data and show that a 15.2 $\mu$m $^{12}$CO$_2$ band inspired five-component decomposition can be applied for the isotopologue band where each component is representative of CO$_2$ ice in a specific molecular environment. The final solution consists of cold mixtures of CO$_2$ with CH$_3$OH, H$_2$O and CO as well as segregated heated pure CO$_2$ ice. Our results are in agreement with previous studies of the $^{12}$CO$_2$ ice band, further confirming that $^{13}$CO$_{2}$ is a useful alternative tracer of protostellar heating events. We also propose an alternative solution consisting only of heated CO$_2$:CH$_3$OH and CO$_2$:H$_2$O ices and warm pure CO$_2$ ice for decomposing the ice profiles of the two most luminous sources in our sample.

We introduce TopoX, a Python software suite that provides reliable and user-friendly building blocks for computing and machine learning on topological domains that extend graphs: hypergraphs, simplicial, cellular, path and combinatorial complexes. TopoX consists of three packages: TopoNetX facilitates constructing and computing on these domains, including working with nodes, edges and higher-order cells; TopoEmbedX provides methods to embed topological domains into vector spaces, akin to popular graph-based embedding algorithms such as node2vec; TopoModelX is built on top of PyTorch and offers a comprehensive toolbox of higher-order message passing functions for neural networks on topological domains. The extensively documented and unit-tested source code of TopoX is available under MIT license at https://pyt-team.github.io/}{https://pyt-team.github.io/.

Molecular hydrogen (H$_2$) formation and dissociation are key processes that drive the gas lifecycle in galaxies. Using the SImulating the LifeCycle of Molecular Clouds (SILCC) zoom-in simulation suite, we explore the utility of future observations of H$_2$ dissociation and formation for tracking the lifecycle of molecular clouds. The simulations used in this work include non-equilibrium H$_2$ formation, stellar radiation, sink particles, and turbulence. We find that, at early times in the cloud evolution, H$_2$ formation rapidly outpaces dissociation and molecular clouds build their mass from the atomic reservoir in their environment. Rapid H$_2$ formation is also associated with a higher early star formation rate. For the clouds studied here, H$_2$ is strongly out of chemical equilibrium during the early stages of cloud formation but settles into a bursty chemical steady-state about 2 Myrs after the first stars form. At the latest stage of cloud evolution, dissociation outweighs formation and the clouds enter a dispersal phase. We discuss how theories for the molecular cloud lifecycle and the star formation efficiency may be distinguished with observational measurements of H$_2$ fluorescence with a space-based high-resolution FUV spectrometer, such as the proposed Hyperion and Eos NASA Explorer missions. Such missions would enable measurements of the H$_2$ dissociation and formation rates, which we demonstrate can be connected to different phases in a molecular cloud's star-forming life, including cloud building, rapidly star-forming, H$_2$ chemical equilibrium, and cloud destruction., Comment: Submitted to ApJ, comments welcome

Magnetic Resonance Imaging (MRI) is a pivotal clinical diagnostic tool, yet its extended scanning times often compromise patient comfort and image quality, especially in volumetric, temporal and quantitative scans. This review elucidates recent advances in MRI acceleration via data and physics-driven models, leveraging techniques from algorithm unrolling models, enhancement-based methods, and plug-and-play models to the emerging full spectrum of generative model-based methods. We also explore the synergistic integration of data models with physics-based insights, encompassing the advancements in multi-coil hardware accelerations like parallel imaging and simultaneous multi-slice imaging, and the optimization of sampling patterns. We then focus on domain-specific challenges and opportunities, including image redundancy exploitation, image integrity, evaluation metrics, data heterogeneity, and model generalization. This work also discusses potential solutions and future research directions, with an emphasis on the role of data harmonization and federated learning for further improving the general applicability and performance of these methods in MRI reconstruction., Comment: Accepted by IEEE Reviews in Biomedical Engineering (RBME)

We conducted an investigation of the Coma cluster of galaxies by running a series of constrained hydrodynamic simulations with GIZMO-SIMBA and GADGET-3, based on initial conditions reconstructed from the SDSS survey volume in the ELUCID project. We compared simulation predictions and observations for galaxies, ICM and IGM in and around the Coma cluster to constrain galaxy formation physics. Our results demonstrate that this type of constrained investigation allows us to probe in more detail the implemented physical processes, because the comparison between simulations and observations is free of cosmic variance and hence can be conducted in a ''one-to-one'' manner. We found that an increase in the earlier star formation rate and the supernova feedback of the original GIZMO-SIMBA model is needed to match observational data on stellar, ISM and ICM metallicity. The simulations without AGN feedback can well reproduce the observational ICM electron density, temperature, and entropy profiles, ICM substructures, and the IGM temperature-density relation, while the ones with AGN feedback usually fail. However, one requires something like AGN feedback to reproduce a sufficiently large population of quiescent galaxies, particularly in low-density regions. The constrained simulations of the Coma cluster thus provide a test bed to understand processes that drive galaxy formation and evolution., Comment: 21 pages, 9 figures

We present a comprehensive study of the nonproportionality of NaI(Tl) scintillation detectors within the context of dark matter search experiments. Our investigation, which integrates COSINE-100 data with supplementary $\gamma$ spectroscopy, measures light yields across diverse energy levels from full-energy $\gamma$ peaks produced by the decays of various isotopes. These $\gamma$ peaks of interest were produced by decays supported by both long and short-lived isotopes. Analyzing peaks from decays supported only by short-lived isotopes presented a unique challenge due to their limited statistics and overlapping energies, which was overcome by long-term data collection and a time-dependent analysis. A key achievement is the direct measurement of the 0.87 keV light yield, resulting from the cascade following electron capture decay of $^{22}$Na from internal contamination. This measurement, previously accessible only indirectly, deepens our understanding of NaI(Tl) scintillator behavior in the region of interest for dark matter searches. This study holds substantial implications for background modeling and the interpretation of dark matter signals in NaI(Tl) experiments., Comment: 12 pages, 7 figures

We prove that a subspace of a real JBW$^*$-triple is an $M$-summand if and only if it is a weak$^*$-closed triple ideal. As a consequence, $M$-ideals of real JB$^*$-triples correspond to norm-closed triple ideals. As in the setting of complex JB$^*$-triples, a geometric property is characterized in purely algebraic terms. This is a newfangled treatment of the classical notion of $M$-ideal in the real setting by a fully new approach due to the unfeasibility of the known arguments in the setting of complex C$^*$-algebras and JB$^*$-triples. The results in this note also provide a full characterization of all $M$-ideals in real C$^*$-algebras, real JB$^*$-algebras and real TROs.

We present the design of a mixed reality (MR) telehealth training system that aims to close the gap between in-person and distance training and re-training for medical procedures. Our system uses real-time volumetric capture as a means for communicating and relating spatial information between the non-colocated trainee and instructor. The system's design is based on a requirements elicitation study performed in situ, at a medical school simulation training center. The focus is on the lightweight real-time transmission of volumetric data - meaning the use of consumer hardware, easy and quick deployment, and low-demand computations. We evaluate the MR system design by analyzing the workload for the users during medical training. We compare in-person, video, and MR training workloads. The results indicate that the overall workload for central line placement training with MR does not increase significantly compared to video communication. Our work shows that, when designed strategically together with domain experts, an MR communication system can be used effectively for complex medical procedural training without increasing the overall workload for users significantly. Moreover, MR systems offer new opportunities for teaching due to spatial information, hand tracking, and augmented communication., Comment: https://hdl.handle.net/10125/102680

Augmented reality (AR) has great potential for use in healthcare applications, especially remote medical training and supervision. In this paper, we analyze the usage of an AR communication system to teach a medical procedure, the placement of a central venous catheter (CVC) under ultrasound guidance. We examine various AR communication and collaboration components, including gestural communication, volumetric information, annotations, augmented objects, and augmented screens. We compare how teaching in AR differs from teaching through videoconferencing-based communication. Our results include a detailed medical training steps analysis in which we compare how verbal and visual communication differs between video and AR training. We identify procedural steps in which medical experts give visual instructions utilizing AR components. We examine the change in AR usage and interaction over time and recognize patterns between users. Moreover, AR design recommendations are given based on post-training interviews., Comment: https://doi.org/10.1145/3611659.3615685

We design and evaluate a mixed reality real-time communication system for remote assistance during CPR emergencies. Our system allows an expert to guide a first responder, remotely, on how to give first aid. RGBD cameras capture a volumetric view of the local scene including the patient, the first responder, and the environment. The volumetric capture is augmented onto the remote expert's view to spatially guide the first responder using visual and verbal instructions. We evaluate the mixed reality communication system in a research study in which participants face a simulated emergency. The first responder moves the patient to the recovery position and performs chest compressions as well as mouth-to-mask ventilation. Our study compares mixed reality against videoconferencing-based assistance using CPR performance measures, cognitive workload surveys, and semi-structured interviews. We find that more visual communication including gestures and objects is used by the remote expert when assisting in mixed reality compared to videoconferencing. Moreover, the performance and the workload of the first responder during simulation do not differ significantly between the two technologies., Comment: https://doi.org/10.1007/978-3-031-32883-1_38

Medical procedures are an essential part of healthcare delivery, and the acquisition of procedural skills is a critical component of medical education. Unfortunately, procedural skill is not evenly distributed among medical providers. Skills may vary within departments or institutions, and across geographic regions, depending on the provider's training and ongoing experience. We present a mixed reality real-time communication system to increase access to procedural skill training and to improve remote emergency assistance. Our system allows a remote expert to guide a local operator through a medical procedure. RGBD cameras capture a volumetric view of the local scene including the patient, the operator, and the medical equipment. The volumetric capture is augmented onto the remote expert's view to allow the expert to spatially guide the local operator using visual and verbal instructions. We evaluated our mixed reality communication system in a study in which experts teach the ultrasound-guided placement of a central venous catheter (CVC) to students in a simulation setting. The study compares state-of-the-art video communication against our system. The results indicate that our system enhances and offers new possibilities for visual communication compared to video teleconference-based training.

We investigate the bright CO fundamental emission in the central regions of five protostars in their primary mass assembly phase using new observations from JWST's Near-Infrared Spectrograph (NIRSpec) and Mid-Infrared Instrument (MIRI). CO line emission images and fluxes are extracted for a forest of $\sim$150 ro-vibrational transitions from two vibrational bands, $v=1-0$ and $v=2-1$. However, ${}^{13}$CO is undetected, indicating that ${}^{12}$CO emission is optically thin. We use H$_2$ emission lines to correct fluxes for extinction and then construct rotation diagrams for the CO lines with the highest spectral resolution and sensitivity to estimate rotational temperatures and numbers of CO molecules. Two distinct rotational temperature components are required for $v=1$ ($\sim600$ to 1000 K and 2000 to $\sim 10^4$ K), while one hotter component is required for $v=2$ ($\gtrsim 3500$ K). ${}^{13}$CO is depleted compared to the abundances found in the ISM, indicating selective UV photodissociation of ${}^{13}$CO; therefore, UV radiative pumping may explain the higher rotational temperatures in $v=2$. The average vibrational temperature is $\sim 1000$ K for our sources and is similar to the lowest rotational temperature components. Using the measured rotational and vibrational temperatures to infer a total number of CO molecules, we find that the total gas masses range from lower limits of $\sim10^{22}$ g for the lowest mass protostars to $\sim 10^{26}$ g for the highest mass protostars. Our gas mass lower limits are compatible with those in more evolved systems, which suggest the lowest rotational temperature component comes from the inner disk, scattered into our line of sight, but we also cannot exclude the contribution to the CO emission from disk winds for higher mass targets., Comment: 31 pages, 7 figures, 4 tables, received to ApJ December 10 2023, accepted to ApJ August 4 2024

Protostellar outflows often present a knotty appearance, providing evidence of sporadic accretion in stellar mass growth. To understand the direct relation between mass accretion and ejection, we analyze the contemporaneous accretion activity and associated ejection components in B335. B335 has brightened in the mid-IR by 2.5 mag since 2010, indicating increased luminosity, presumably due to increased mass accretion rate onto the protostar. ALMA observations of 12CO emission in the outflow reveal high-velocity emission, estimated to have been ejected 4.6 - 2 years before the ALMA observation and consistent with the jump in mid-IR brightness. The consistency in timing suggests that the detected high-velocity ejection components are directly linked to the most recent accretion activity. We calculated the kinetic energy, momentum, and force for the ejection component associated with the most recent accretion activity and found that at least, about 1.0% of accreted mass has been ejected. More accurate information on the jet inclination and the temperature of the ejected gas components will better constrain the ejected mass induced by the recently enhanced accretion event., Comment: 11 pages, 5 figures, and 1 table