Search for: All records

Nucleoside triphosphate (NTP)-dependent protein assemblies such as microtubules and actin filaments have inspired the development of diverse chemically fueled molecular machines and active materials but their functional sophistication has yet to be matched by design. Given this challenge, we asked whether it is possible to transform a natural adenosine 5′-triphosphate (ATP)-dependent enzyme into a dissipative self-assembling system, thereby altering the structural and functional mode in which chemical energy is used. Here we report that FtsH (filamentous temperature-sensitive protease H), a hexameric ATPase involved in membrane protein degradation, can be readily engineered to form one-dimensional helical nanotubes. FtsH nanotubes require constant energy input to maintain their integrity and degrade over time with the concomitant hydrolysis of ATP, analogous to natural NTP-dependent cytoskeletal assemblies. Yet, in contrast to natural dissipative systems, ATP hydrolysis is catalyzed by free FtsH protomers and FtsH nanotubes serve to conserve ATP, leading to transient assemblies whose lifetimes can be tuned from days to minutes through the inclusion of external ATPases in solution. 

Despite significant progress in video question answering (VideoQA), existing methods fall short of questions that require causal/temporal reasoning across frames. This can be attributed to imprecise motion representations. We introduce Action Temporality Modeling (ATM) for temporality reasoning via three-fold uniqueness: (1) rethinking the optical flow and realizing that optical flow is effective in capturing the long horizon temporality reasoning; (2) training the visual-text embedding by contrastive learning in an action-centric manner, leading to better action representations in both vision and text modalities; and (3) preventing the model from answering the question given the shuffled video in the fine-tuning stage, to avoid spurious correlation between appearance and motion and hence ensure faithful temporality reasoning. In the experiments, we show that ATM outperforms existing approaches in terms of the accuracy on multiple VideoQAs and exhibits better true temporality reasoning ability. 

Acoustic resonances in open systems, which are usually associated with resonant modes characterized by complex eigenfrequencies, play a fundamental role in manipulating acoustic wave radiation and propagation. Notably, they are accompanied by considerable field enhancement, boosting interactions between waves and matter, and leading to various exciting applications. In the past two decades, acoustic metamaterials have enabled a high degree of control over tailoring acoustic resonances over a range of frequencies. Here, we provide an overview of recent advances in the area of acoustic resonances in non-Hermitian open systems, including Helmholtz resonators, metamaterials and metasurfaces, and discuss their applications in various acoustic devices, including sound absorbers, acoustic sources, vortex beam generation and imaging. We also discuss bound states in the continuum and their applications in boosting acoustic wave–matter interactions, active phononics and non-Hermitian acoustic resonances, including phononic topological insulators and the acoustic skin effect.