ترغب بنشر مسار تعليمي؟ اضغط هنا

Supervised learning of an opto-magnetic neural network with ultrashort laser pulses

67   0   0.0 ( 0 )
 نشر من قبل Johan Mentink
 تاريخ النشر 2018
والبحث باللغة English




اسأل ChatGPT حول البحث

The explosive growth of data and its related energy consumption is pushing the need to develop energy-efficient brain-inspired schemes and materials for data processing and storage. Here, we demonstrate experimentally that Co/Pt films can be used as artificial synapses by manipulating their magnetization state using circularly-polarized ultrashort optical pulses at room temperature. We also show an efficient implementation of supervised perceptron learning on an opto-magnetic neural network, built from such magnetic synapses. Importantly, we demonstrate that the optimization of synaptic weights can be achieved using a global feedback mechanism, such that the learning does not rely on external storage or additional optimization schemes. These results suggest there is high potential for realizing artificial neural networks using optically-controlled magnetization in technologically relevant materials, that can learn not only fast but also energy-efficient.



قيم البحث

اقرأ أيضاً

Quantum cascade lasers (QCL) have revolutionized the generation of mid-infrared light. Yet, the ultrafast carrier transport in mid-infrared QCLs has so far constituted a seemingly insurmountable obstacle for the formation of ultrashort light pulses. Here, we demonstrate that careful quantum design of the gain medium and control over the intermode beat synchronization enable transform-limited picosecond pulses from QCL frequency combs. Both an interferometric radio-frequency technique and second-order autocorrelation shed light on the pulse dynamics and confirm that mode-locked operation is achieved from threshold to rollover current. Being electrically pumped and compact, mode-locked QCLs pave the way towards monolithically integrated non-linear photonics in the molecular fingerprint region beyond 6 $mu$m wavelength.
Lead-magnesium niobate lead-titanate (PMN-PT) has been proven as an excellent material for sensing and actuating applications. The fabrication of advanced ultra-small PMN-PT-based devices relies on the availability of sophisticated procedures for the micro-machining of PMN-PT thin films or bulk substrates. Approaches reported up to date include chemical etching, excimer laser ablation and ion milling. To ensure an excellent device performance, a key mandatory feature for a micro-machining process is to preserve as far as possible the crystalline quality of the substrates; in other words, the fabrication method must induce a low density of cracks and other kind of defects. In this work, we demonstrate a relatively fast procedure for the fabrication of high-quality PMN-PT micro-machined actuators employing green femtosecond laser pulses. The fabricated devices feature absence of extended cracks and well defined edges with relatively low roughness, which is advantageous for the further integration of nanomaterials onto the piezoelectric actuators.
The dispersion scan (d-scan) technique has emerged as a simple-to-implement characterization method for ultrashort laser pulses. D-scan traces are intuitive to interpret and retrieval algorithms that are both fast and robust have been developed to ob tain the spectral phase and the temporal pulse profile. Here, we give a review of the d-scan technique based on second harmonic generation. We describe and compare recent implementations for the characterization of few- and multi-cycle pulses as well as two different approaches for recording d-scan traces in single-shot, thus showing the versatility of the technique.
111 - Z. Samsonova 2018
We report on experimental results in a new regime of a relativistic light-matter interaction employing mid-infrared (3.9-micrometer wavelength) high-intensity femtosecond laser pulses. In the laser generated plasma, the electrons reach relativistic e nergies already at rather low intensities due to the fortunate lambda^2-scaling of the kinetic energy with the laser wavelength. The lower intensity suppresses optical field ionization and creation of the pre-plasma at the rising edge of the laser pulse efficiently, enabling an enhanced efficient vacuum heating of the plasma. The lower critical plasma density for long-wavelength radiation can be surmounted by using nanowires instead of flat targets. In our experiments about 80% of the incident laser energy has been absorbed resulting in a long living, keV-temperature, high-charge state plasma with a density of more than three orders of magnitude above the critical value. Our results pave the way to laser-driven experiments on laboratory astrophysics and nuclear physics at a high repetition rate.
We study a model for frustrated tunneling ionization using ultrashort laser pulses. The model is based on the strong field approximation and it employs the saddle point approximation to predict quasiclassical trajectories that are captured on Rydberg states. We present a classification of the saddle-point solutions and explore their behavior as functions of angular momentum of the final state, as well as the carrier--envelope phase (CEP) of the laser pulse. We compare the final state population computed by the model to results obtained by numerical propagation of the time-dependent Schrodinger equation (TDSE) for the hydrogen atom. While we find qualitative agreement in the CEP dependence of the populations in principal quantum numbers, $n$, the populations to individual angular momentum channels, $ell$, are found to be inconsistent between model and TDSE. Thus, our results show that improvements of the quasiclassical trajectories are in order for a quantitative model of frustrated tunneling ionizaiton.
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