62588 research outputs found
Sort by
MetaNO: How to Transfer Your Knowledge on Learning Hidden Physics
No full textGradient-based meta-learning methods have primarily been applied to classical
machine learning tasks such as image classification. Recently, PDE-solving deep
learning methods, such as neural operators, are starting to make an important
impact on learning and predicting the response of a complex physical system
directly from observational data. Since the data acquisition in this context is
commonly challenging and costly, the call of utilization and transfer of
existing knowledge to new and unseen physical systems is even more acute.
Herein, we propose a novel meta-learning approach for neural operators, which
can be seen as transferring the knowledge of solution operators between
governing (unknown) PDEs with varying parameter fields. Our approach is a
provably universal solution operator for multiple PDE solving tasks, with a key
theoretical observation that underlying parameter fields can be captured in the
first layer of neural operator models, in contrast to typical final-layer
transfer in existing meta-learning methods. As applications, we demonstrate the
efficacy of our proposed approach on PDE-based datasets and a real-world
material modeling problem, illustrating that our method can handle complex and
nonlinear physical response learning tasks while greatly improving the sampling
efficiency in unseen tasks
Enhancing Supercapacitive Swing Adsorption of CO <sub>2</sub> with Advanced Activated Carbon Electrodes
No full textGlobal warming due to anthropogenic CO 2 emissions argues for the rapid development of efficient carbon capture technologies. Supercapacitive swing adsorption (SSA) is a gas separation technology that relies on the reversible charge and discharge of supercapacitor electrodes to absorb and desorb CO 2 highly selectively and reversibly. However, thus far SSA only showed low sorption capacity, and slow sorption kinetics. Herein, it is shown that the sorption capacity can be substantially increased via the use of carbons with a higher specific capacitance. The highest gravimetric sorption capacity is measured with electrodes made from garlic‐roots derived activated carbon valuing 273 mmol kg −1 having a specific capacitance of 257 F g −1 . In addition, the overall adsorption rate and productivity are improved. Cycling the electrodes for over 100 h showed highly reproducible, reversible CO 2 adsorption and desorption behavior. A preliminary technoeconomic and sensitivity analysis is provided to demonstrate the potential of SSA for commercial applications
The formation and functions of school‐based trauma‐leadership teams
No full textStudies suggest that among children, adverse childhood experiences increase the risk of developing behavioral challenges in and out of the school environment. Rooted in distributed leadership, trauma‐leadership teams (TLTs) are a novel systems‐based intervention in which a team of educators deepens knowledge and works to implement trauma‐responsive policies and practices within the school community. The current study used a consensual qualitative research design to (a) understand the ways TLTs are created, (b) describe educators\u27 perceived benefits and outcomes of these teams, and (c) describe how TLTs are applied in schools to improve trauma‐informed care practices. Additionally, the study highlights growth areas for TLT implementation. Domains from interviews include (1) Formation of TLTs ; (2) Benefits and Outcomes of TLTs ; (3) Trauma‐Responsive Competency ; and (4) Growth Areas of TLT . Results suggest that TLT members view TLTs positively, offer insight into how TLTs are formed, and see benefits from TLTs within their school communities. Results also suggest areas for growth for TLTs. , Practitioner points School‐based trauma‐leadership teams (TLTs) are perceived to be a useful mechanism for planning and implementing strategies and policies that are trauma‐informed. Educators engaged in TLTs that employ distributed leadership or cooperative models report feeling greater levels of trust and stronger relationships within the team and with peers across the school. Both are core components of a trauma‐informed organization. TLTs may be more effective in addressing mental health problems and promoting a positive school culture with even more structure and a greater frequency of meetings
Substrate Effects on Growth Dynamics of WTe <sub>2</sub> Thin films
No full textSynthesis of transition metal dichalcogenides (TMDCs) has been achieved through the direct conversion of metal and metal‐oxide films, demonstrating the ability to grow large area thin films with uniform thickness on a variety of substrates and direct control over the growth orientation (horizontal vs vertical) of the TMDC layers. However, the synthesized TMDC films often exhibit small grains and are more defective than their bulk counterparts. This is especially true for 2D telluride films due to the low reactivity between tellurium and transition metals such as W and Mo. In this work, the substrate interactions is examined for WTe 2 converted from amorphous WO x thin films grown by atomic layer deposition, on c‐plane sapphire and SiO 2 through tellurization at high temperatures. Similar to TMDC telluride MoTe 2 , the formation of monolayer WTe 2 on sapphire is observed, but not on SiO 2 . However, due to decreased diffusion of W on sapphire compared to Mo, the formation of WTe 2 flakes instead of continuous films is observed, providing insight into the role of the specific transition metal during the direct synthesis of TMDC telluride films
Exploring the co‐occurrence of students\u27 learning behaviours and reasoning processes in an intelligent tutoring system
No full textBackground Medical students use a variety of self‐regulated learning (SRL) strategies in different medical reasoning (MR) processes to solve patient cases of varying complexity. However, the interplay between SRL and MR processes is still unclear. Objectives This study investigates how self‐regulated learning (SRL) and medical reasoning (MR) occurred concurrently in medical students while completing a diagnostic task in an intelligent tutoring system. This study aims to provide new insights into performance differences between high‐ and low‐achieving students in tasks of varying complexity. Methods Thirty‐one medical students (67.6% female) from a large North American university were tasked with solving two virtual patient cases in an intelligent tutoring system, BioWorld. BioWorld was designed for medical students to practice clinical reasoning skills deliberately. We collected students\u27 think‐aloud protocols, based on which we coded their use of SRL behaviours and medical reasoning activities. We analysed the co‐occurrences of SRL behaviours and medical reasoning activities using the epistemic network analysis (ENA) method. Results The SRL behaviour self‐reflection and MR activity lines of reasoning co‐occurred more frequently in a difficult task than in an easy task. In both tasks, high performers demonstrated more co‐occurrences of self‐reflection and lines of reasoning than low performers. Moreover, the MR activity conceptual operations co‐occurred more frequently with the SRL activities of monitoring and evaluation among high performers compared to low performers in an easy task. Implications The co‐occurrences of SRL behaviours and MR processes account for students\u27 performance differences. The design of computer‐based learning environments for clinical reasoning should promote the acquisition of both SRL and medical reasoning abilities. Moreover, medical educators should consider task complexity when scaffolding. , Lay Description What is already known about this topic Self‐regulated learning (SRL) and medical reasoning skills are both crucial for diagnosing patients. Medical students can practice clinical reasoning with computer simulations. What this paper adds Students solved virtual patients of varying complexity in an intelligent tutoring system We examined the co‐occurrences of SRL behaviours and medical reasoning process. Epistemic network analysis was used to analyse the interplay of SRL and medical reasoning. High performers show more co‐occurrences of reflection and higher‐order reasoning. Implications for practice and/or policy Task complexity has impact on students\u27 learning and reasoning co‐occurrences. Intelligent tutoring systems should foster regulation and reasoning acquisition
You\u27re driving me crazy! How emotions elicited by negative driver behaviors impact customer outcomes in last mile delivery
No full textWith the growth of e‐commerce and associated home deliveries, understanding the role of drivers in shaping the customer experience in last‐mile delivery is now more crucial than ever. Delivery drivers increasingly act as retailers\u27 frontline employees and are thus instrumental in developing pseudorelationships between customers and retailers. Industry surveys, however, reveal that drivers admit to engaging in unprofessional behaviors with customers and often refuse to address customers\u27 requests beyond package delivery. Following a middle‐range theorizing approach and leveraging Cognitive Appraisal Theory, we investigate how two negative driver behaviors, inappropriate behavior and inflexibility, impact customer satisfaction and repurchase intentions. We also examine the moderating effect of driver affiliation, private versus outsourced, in altering the magnitude of customer responses. Results from a scenario‐based experiment indicate that while the negative effects of driver inappropriate behavior on customer outcomes are mediated by anger, the effects of driver inflexibility are mediated by sadness. Moreover, the negative effect of driver inflexibility on customer outcomes is weaker for outsourced logistics than for private fleet drivers. In turn, driver inappropriate behavior exhibits similar negative effects on customer outcomes for both driver affiliations. These findings offer important insights for last‐mile delivery strategy and operations research and practice
Dark dimension gravitons as dark matter
No full textAbstract
We consider cosmological aspects of the Dark Dimension (a mesoscopic dimension of micron scale), which has recently been proposed as the unique corner of the quantum gravity landscape consistent with both the Swampland criteria and observations. In particular we show how this leads, by the universal coupling of the Standard Model sector to bulk gravitons, to massive spin 2 KK excitations of the graviton in the dark dimension (the “dark gravitons”) as an unavoidable dark matter candidate. Assuming a lifetime for the current de Sitter phase of our universe of order Hubble, which follows from both the dS Swampland Conjecture and TCC, we show that generic features of the dark dimension cosmology can naturally lead to the correct dark matter density and a resolution of the cosmological coincidence problem, where the matter/radiation equality temperature (T ~ 1 eV) coincides with the temperature where the dark energy begins to dominate. Thus one does not need to appeal to Weinberg’s anthropic argument to explain this coincidence. The dark gravitons are produced at T ~ 4 GeV, and their composition changes as they mainly decay to lighter gravitons, without losing much total mass density. The mass of dark gravitons is mDM ∼ 1 − 100 keV today.</jats:p
Breaking rotations without violating the KSS viscosity bound
No full textAbstract
We revisit the computation of the shear viscosity to entropy ratio in a holographic p-wave superfluid model, focusing on the role of rotational symmetry breaking. We study the interplay between explicit and spontaneous symmetry breaking and derive a simple horizon formula for η/s, which is valid also in the presence of explicit breaking of rotations and is in perfect agreement with the numerical data. We observe that a source which explicitly breaks rotational invariance suppresses the value of η/s in the broken phase, competing against the effects of spontaneous symmetry breaking. However, η/s always reaches a constant value in the limit of zero temperature, which is never smaller than the Kovtun-Son-Starinets (KSS) bound, 1/4π. This behavior appears to be in contrast with previous holographic anisotropic models which found a power-law vanishing of η/s at small temperature. This difference is shown to arise from the properties of the near-horizon geometry in the extremal limit. Thus, our construction shows that the breaking of rotations itself does not necessarily imply a violation of the KSS bound.</jats:p
