ترغب بنشر مسار تعليمي؟ اضغط هنا

Adiabatic Superconducting Artificial Neural Network: Basic Cells

209   0   0.0 ( 0 )
 نشر من قبل Igor Soloviev I
 تاريخ النشر 2018
  مجال البحث فيزياء
والبحث باللغة English




اسأل ChatGPT حول البحث

We consider adiabatic superconducting cells operating as an artificial neuron and synapse of a multilayer perceptron (MLP). Their compact circuits contain just one and two Josephson junctions, respectively. While the signal is represented as magnetic flux, the proposed cells are inherently nonlinear and close-to-linear magnetic flux transformers. The neuron is capable of providing a one-shot calculation of sigmoid and hyperbolic tangent activation functions most commonly used in MLP. The synapse features by both positive and negative signal transfer coefficients in the range ~ (-0.5,0.5). We briefly discuss implementation issues and further steps toward multilayer adiabatic superconducting artificial neural network which promises to be a compact and the most energy-efficient implementation of MLP.



قيم البحث

اقرأ أيضاً

We perform infrared conductivity measurements on a series of CaCuO$_2$/SrTiO$_3$ heterostructures made by the insulating cuprate CaCuO$_2$ (CCO) and the insulating perovkite SrTiO$_3$ (STO). We estimate the carrier density of various heterostructures , with different level of hole doping from the integral of the optical conductivity and we measure the corresponding degree of correlation by estimating the ratio between the Drude weight and the integral of the infrared spectrum. The analysis demonstrates a large degree of correlation, which increases as the doping is reduced. The experimental results can be reproduced by Dynamical Mean-Field Theory calculations which strongly support the role of correlations in the CCO/STO heterostructures and their similarities with the most common cuprate superconductors. Our results suggest that cuprate superconductors can be looked at as natural superlattices, where the properties of the CuO$_2$ conducting planes and charge reservoir blocks can be completely disentangled.
195 - Z.H. Peng , J.H. Ding , Y. Zhou 2017
We experimentally study a vacuum-induced Autler-Townes doublet in a superconducting three-level artificial atom strongly coupled to a coplanar waveguide resonator and simultaneously to a transmission line. The Autler-Townes splitting is observed in t he reflection spectrum from the three-level atom in a transition between the ground state and the second excited state when the transition between the two excited states is resonant with a resonator. By applying a driving field to the resonator, we observe a change in the regime of the Autler-Townes splitting from quantum (vacuum-induced) to classical (with many resonator photons). Furthermore, we show that the reflection of propagating microwaves in a transmission line could be controlled by different frequency single photons in a resonator.
Based on recent studies regarding high-temperature (high-$T_c$) La-Y ternary hydrides (e.g., $P{bar{1}}$-La$_2$YH$_{12}$, $Pm{bar{3}}m$-LaYH$_{12}$, and $Pm{bar{3}}m$-(La,Y)H$_{10}$ with a maximum $T_c sim 253$ K), we examined the phase and structura l stabilities of the (LaH$_6$)(YH$_6$)$_y$ series as high-$T_c$ ternary hydride compositions using a genetic algorithm and $it ab$ $it initio$ calculations. Our evaluation showed that the $Pmbar{3}m$-LaYH$_{12}$ reported in the previous study was unstable during decomposition into $Rbar{3}c$-LaH$_{6}$ + $Imbar{3}m$-YH$_{6}$. We also discovered new crystal structures, namely $Cmmm$-LaYH$_{12}$ ($y=1$), $Rbar{3}c$-LaYH$_{12}$ ($y=1$), $Cmmm$-LaY$_3$H$_{24}$ ($y=3$), and $Rbar{3}$-LaY$_3$H$_{24}$ ($y=3$), showing stability against such decomposition. While $Rbar{3}c$ ($y=1$) and $Rbar{3}$ ($y=3$) did not exhibit superconductivity owing to the extremely low density of states at the Fermi level, $Cmmm$ phases exhibited a $T_{c}$ of approximately 140~K at around 200~GPa owing to the extremely high electron--phonon coupling constant ($lambda$ = 1.876 for LaYH$_{12}$). By the twice longer stacking for $Cmmm$-LaY$_3$H$_{24}$, the coupling constant increased owing to the chemical pressure of Y, leading to a slightly increased $T_{c}$.
We study a superconducting artificial atom which is represented by a single Josephson junction or a Josephson junction chain, capacitively coupled to a coherently driven transmission line, and which contains exactly one residual quasiparticle (or up to one quasiparticle per island in a chain). We study the dissipation in the atom induced by the quasiparticle tunneling, taking into account the quasiparticle heating by the drive. We calculate the transmission coefficient in the transmission line for drive frequencies near resonance and show that, when the artificial atom spectrum is nearly harmonic, the intrinsic quality factor of the resonance increases with the drive power. This counterintuitive behavior is due to the energy dependence of the quasiparticle density of states.
Prediction of material properties from first principles is often a computationally expensive task. Recently, artificial neural networks and other machine learning approaches have been successfully employed to obtain accurate models at a low computati onal cost by leveraging existing example data. Here, we present a software package Properties from Artificial Neural Network Architectures (PANNA) that provides a comprehensive toolkit for creating neural network models for atomistic systems. Besides the core routines for neural network training, it includes data parser, descriptor builder and force-field generator suitable for integration within molecular dynamics packages. PANNA offers a variety of activation and cost functions, regularization methods, as well as the possibility of using fully-connected networks with custom size for each atomic species. PANNA benefits from the optimization and hardware-flexibility of the underlying TensorFlow engine which allows it to be used on multiple CPU/GPU/TPU systems, making it possible to develop and optimize neural network models based on large datasets.
التعليقات
جاري جلب التعليقات جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
mircosoft-partner

هل ترغب بارسال اشعارات عن اخر التحديثات في شمرا-اكاديميا