Characterizing Probabilistic Bits with Quantum Spin Defects

Many complex problems related to natural phenomena as well as to artificial intelligence are inherently probabilistic. Traditional digital computers with deterministic hardware are slow and energy inefficient at solving such large-scale stochastic problems. Probabilistic computers, built with thermally unstable low-barrier nanomagnets, promise orders of magnitude improvement in speed and energy efficiency over deterministic computers. However, electrically characterizing such stochastic nanomagnets is challenging and slow as it would require the fabrication of a full magnetic tunnel junction stack for each nanomagnet. In this work we propose quantum spin defects such as nitrogen vacancy centers (NV centers) as novel probes for rapidly characterizing magnetic probabilistic bits (p-bits). Through micromagnetic simulations and analytical calculations, we show that NV relaxometry techniques can be used to extract key parameters such as energy barrier and the so-called attempt time of a stochastic nanomagnet over a wide range of fluctuation frequencies. We further show that large angle nanomagnet dynamics show distinct spatial and thermal relaxometry (T1 and T2) signatures compared to small angle fluctuations such as thermal magnons. Moreover, thanks to the atomic size of the NV probes and the weak magnetic fields they generate on the sample, this technique has nanoscale resolution and is noninvasive. Our work paves the way for optimizing magnetic materials for probabilistic computing applications.