Convolutional Neural Networks (CNNs) are widely adopted in object recognition, speech processing and machine translation, due to their extremely high inference accuracy. However, it is challenging to compute massive computationally expensive convolutions of deep CNNs on traditional CPUs and GPUs. Emerging Nanophotonic technology has been employed for on-chip data communication, because of its CMOS compatibility, high bandwidth and low power consumption. In this paper, we propose a nanophotonic accelerator, HolyLight, to boost the CNN inference throughput in datacenters. Instead of an all-photonic design, HolyLight performs convolutions by photonic integrated circuits, and process the other operations in CNNs by CMOS circuits for high inference accuracy. We first build HolyLight-M by microdisk-based matrix-vector multipliers. We find analog-todigital converters (ADCs) seriously limit its inference throughput per Watt. We further use microdisk-based adders and shifters to architect HolyLight-A without ADCs. Compared to the stateof-the-art ReRAM-based accelerator, HolyLight-A improves the CNN inference throughput per Watt by 13× with trivial accuracy degradation.
Nanoparticles (NPs) grafted with highly dense DNA strands are termed as spherical nucleic acids (SNAs), which have important applications benefiting from various unique properties unpossessed by naturally occurring nucleic acids. To overcome existing challenges toward an ideal SNA synthesis, herein, a very simple, while highly effective evaporative drying strategy featuring various long‐desired advantages, is reported. This includes record‐high DNA loading, generality for more NP materials, fully and quantitatively tunable DNA density, and readiness toward bulk production. The process requires almost zero care and the solid products are especially suitable for a long‐time storage without quality degradation. The research reveals a quick and highly efficient packing of thiol‐tagged DNA on the NP surface at the critical moment of drying, which refreshes previous knowledge on DNA conjugation chemistry. Based on this advancement, practical applications of SNAs in various fields may become possible.
In this paper, the nonlinear dynamics of a novel model based on optically pumped spin-polarized vertical-cavity surface-emitting lasers (spin-VCSELs) with optical feedback is investigated numerically. Due to optical feedback being the external disturbance component, the complex nonlinear dynamical behaviors can be enhanced and the regions of different nonlinear dynamics in size can be extended with appropriate parameters of spin-VCSELs. According to the equations of the modified spin-flip model (SFM), the comparison of bifurcation diagrams is first presented for the clear presentation of different routes to chaos. Meanwhile, numerous bifurcation diagrams in color are illustrated to demonstrate the rich dynamical regimes intuitively, and the crucial effects of optical feedback strength, feedback delay, linewidth enhancement factor, and spin-flip relaxation rate on the region evolvement of complex dynamics of the proposed model are revealed to investigate the dependence of dynamical behaviors on external and internal parameters when the optical feedback scheme is introduced. These parameters play a remarkable role in enhancing the mechanism of complex dynamic oscillations. Furthermore, utilizing combination with time series, power spectra, and phase portraits, the various dynamical behaviors observed in the bifurcation diagram are simulated numerically. Correspondingly, the powerful measure 0–1 test is employed to distinguish between chaos and non-chaos.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.