Bayesian networks are powerful statistical models to understand causal relationships in real-world probabilistic problems such as diagnosis, forecasting, computer vision, etc. For systems that involve complex causal dependencies among many variables, the complexity of the associated Bayesian networks become computationally intractable. As a result, direct hardware implementation of these networks is one promising approach to reducing power consumption and execution time. However, the few hardware implementations of Bayesian networks presented in literature rely on deterministic CMOS devices that are not efficient in representing the stochastic variables in a Bayesian network that encode the probability of occurrence of the associated event. This work presents an experimental demonstration of a Bayesian network building block implemented with inherently stochastic spintronic devices based on the natural physics of nanomagnets. These devices are based on nanomagnets with perpendicular magnetic anisotropy, initialized to their hard axes by the spin orbit torque from a heavy metal under-layer utilizing the giant spin Hall effect, enabling stochastic behavior. We construct an electrically interconnected network of two stochastic devices and manipulate the correlations between their states by changing connection weights and biases. By mapping given conditional probability tables to the circuit hardware, we demonstrate that any two node Bayesianmore »
Employing the probabilistic nature of unstable nano-magnet switching has recently emerged as a path towards unconventional computational systems such as neuromorphic or Bayesian networks. In this letter, we demonstrate proof-of-concept stochastic binary operation using hard axis initialization of nano-magnets and control of their output state probability (activation function) by means of input currents. Our method provides a natural path towards addition of weighted inputs from various sources, mimicking the integration function of neurons. In our experiment, spin orbit torque (SOT) is employed to “drive” nano-magnets with perpendicular magnetic anisotropy (PMA) -to their metastable state, i.e. in-plane hard axis. Next, the probability of relaxing into one magnetization state (+mi) or the other (−mi) is controlled using an Oersted field generated by an electrically isolated current loop, which acts as a “charge” input to the device. The final state of the magnet is read out by the anomalous Hall effect (AHE), demonstrating that the magnetization can be probabilistically manipulated and output through charge currents, closing the loop from charge-to-spin and spin-to-charge conversion. Based on these building blocks, a two-node directed network is successfully demonstrated where the status of the second node is determined by the probabilistic output of the previous node more »
- Award ID(s):
- Publication Date:
- NSF-PAR ID:
- Journal Name:
- Scientific Reports
- Nature Publishing Group
- Sponsoring Org:
- National Science Foundation
More Like this
A low-energy hardware implementation of deep belief network (DBN) architecture is developed using near-zero energy barrier probabilistic spin logic devices (p-bits), which are modeled to real- ize an intrinsic sigmoidal activation function. A CMOS/spin based weighted array structure is designed to implement a restricted Boltzmann machine (RBM). Device-level simulations based on precise physics relations are used to validate the sigmoidal relation between the output probability of a p-bit and its input currents. Characteristics of the resistive networks and p-bits are modeled in SPICE to perform a circuit-level simulation investigating the performance, area, and power consumption tradeoffs of the weighted array. In the application-level simulation, a DBN is implemented in MATLAB for digit recognition using the extracted device and circuit behavioral models. The MNIST data set is used to assess the accuracy of the DBN using 5,000 training images for five distinct network topologies. The results indicate that a baseline error rate of 36.8% for a 784x10 DBN trained by 100 samples can be reduced to only 3.7% using a 784x800x800x10 DBN trained by 5,000 input samples. Finally, Power dissipation and accuracy tradeoffs for probabilistic computing mechanisms using resistive devices are identified.
Abstract Spin-orbit coupled honeycomb magnets with the Kitaev interaction have received a lot of attention due to their potential of hosting exotic quantum states including quantum spin liquids. Thus far, the most studied Kitaev systems are 4 d /5 d -based honeycomb magnets. Recent theoretical studies predicted that 3 d -based honeycomb magnets, including Na 2 Co 2 TeO 6 (NCTO), could also be a potential Kitaev system. Here, we have used a combination of heat capacity, magnetization, electron spin resonance measurements alongside inelastic neutron scattering (INS) to study NCTO’s quantum magnetism, and we have found a field-induced spin disordered state in an applied magnetic field range of 7.5 T < B (⊥ b -axis) < 10.5 T. The INS spectra were also simulated to tentatively extract the exchange interactions. As a 3 d -magnet with a field-induced disordered state on an effective spin-1/2 honeycomb lattice, NCTO expands the Kitaev model to 3 d compounds, promoting further interests on the spin-orbital effect in quantum magnets.
Spin currents are used to write information in magnetic random access memory (MRAM) devices by switching the magnetization direction of one of the ferromagnetic electrodes of a magnetic tunnel junction (MTJ) nanopillar. Different physical mechanisms of conversion of charge current to spin current can be used in two-terminal and three-terminal device geometries. In two-terminal devices, charge-to-spin conversion occurs by spin filtering in the MTJ's ferromagnetic electrodes and present day MRAM devices operate near the theoretically expected maximum charge-to-spin conversion efficiency. In three-terminal devices, spin–orbit interactions in a channel material can also be used to generate large spin currents. In this Perspective article, we discuss charge-to-spin conversion processes that can satisfy the requirements of MRAM technology. We emphasize the need to develop channel materials with larger charge-to-spin conversion efficiency—that can equal or exceed that produced by spin filtering—and spin currents with a spin polarization component perpendicular to the channel interface. This would enable high-performance devices based on sub-20 nm diameter perpendicularly magnetized MTJ nanopillars without need of a symmetry breaking field. We also discuss MRAM characteristics essential for CMOS integration. Finally, we identify critical research needs for charge-to-spin conversion measurements and metrics that can be used to optimize device channel materials andmore »
Spin Wave Excitation, Detection, and Utilization in the Organic‐Based Magnet, V(TCNE)x (TCNE = Tetracyanoethylene)Spin waves, quantized as magnons, have low energy loss and magnetic damping, which are critical for devices based on spin‐wave propagation needed for information processing devices. The organic‐based magnet [V(TCNE)x; TCNE = tetracyanoethylene; x ≈ 2] has shown an extremely low magnetic damping comparable to, for example, yttrium iron garnet (YIG). The excitation, detection, and utilization of coherent and non‐coherent spin waves on various modes in V(TCNE)x is demonstrated and show that the angular momentum carried by microwave‐excited coherent spin waves in a V(TCNE)x film can be transferred into an adjacent Pt layer via spin pumping and detected using the inverse spin Hall effect. The spin pumping efficiency can be tuned by choosing different excited spin wave modes in the V(TCNE)x film. In addition, it is shown that non‐coherent spin waves in a V(TCNE)x film, excited thermally via the spin Seebeck effect, can also be used as spin pumping source that generates an electrical signal in Pt with a sign change in accordance with the magnetization switching of the V(TCNE)x. Combining coherent and non‐coherent spin wave detection, the spin pumping efficiency can be thermally controlled, and new insight is gained for the spintronic applications of spin wave modes in organic‐basedmore »