Optoelectronics/Quantum Devices

Neuromorphic computing using quantum dot networks

NeuroQNet II has the overall goals of (i) establishing a universal platform for nonlinear optical networks and (ii) a revolutionary 3D integration concept.

With regard to (i), we will decisively further develop the great potential of the nanophotonic hardware implemented in the previous project NeuroQNet I. While the present quantum dot micropillar arrays (QDMPA) are already very well suited for photonic reservoir computing (RC), their spectral homogeneity and optical amplification still need to be significantly increased for applications beyond this in coherent annealing computing and coherent beam combining. Interestingly, these extremely attractive concepts are based on the same architecture as RC: the coupling and/or optical injection of individual lasers in an array. In the second phase of the project, we will optimize our hardware platform for this purpose in order to demonstrate and investigate coherent annealing and coherent beam combination in a coupled microlaser system for the first time.

Furthermore, with respect to (ii), we will develop the integration of photonic circuits into three-dimensional waveguide structures via the extremely flexible Nanoscribe 3D printing technique. Such an architecture will have an extremely far-reaching impact for optical neuromorphic networks. Currently, corresponding 2D solutions for hardware integration are not scalable. In NeuroQNet II, we will effectively remove this limitation by transitioning from 2D arrays to 3D networks in order to decisively increase scaling and thus computing power.

Quick Info

Project startJuly 2019
Funding sourceVolkswagenStiftung
Funding ID