ChipAI Neuromorphic System Simulator Free-software
The ChipAI app: a brief overview
In the framework of the ChipAI project, a collection of three apps that simulate the dynamics of neuromorphic nanophotonic devices was developed. The app is intended for users interested in simulating the generation and transmission of excitable optical pulses in artificial nano-optoelectronic spiking neurons. The apps are available here on the ChipAI project website and are to be used by other researchers/stakeholders for them to have an understanding of how optical pulses (spikes) are generated in these systems, and how these are affected by the systems’ parameters and hardware specifications.
The apps contribute to disseminating the project’s research concepts and outputs in a practical, interactive way, as well as serve as a tool for other scientists to use in their own research.
The simulator is based on the physical model of the hardware, which consists of nanoscale resonant tunnelling diode (RTD) elements coupled to either nanolaser diodes (LD) and nanophotodetector (PD) components forming respectively spiking transmitter (TX) and receiver (RX) modules, as illustrated in Fig. 1. A comprehensive tutorial ChipAI Neuromorphic System Simulator Free-software app can be downloaded here.
The three apps account for:
App 2: A two-node network, consisting of two RTD-LD nodes connected via a photodetector;
App 3: A two-node network with an additional (recurrent) connection in the reverse direction via a delay line.
Figure 1. Illustration of the proposed solution for a spike-based neuromorphic system based on two types of RTD-powered optoelectronic nodes: RTD-LD (transmitter) and RTD-PD (receiver) nodes [1].
About the app
Three apps with graphical user interfaces (GUI) have been programmed intending to reproduce the dynamics of a single optoelectronic node, a two-node integrated network and a two-node integrated network with bidirectional communication, respectively. Each scenario was coded in a different simulator app. A complete tutorial ChipAI Neuromorphic System Simulator Free-software app can be downloaded here (link).
Requirements to use the app
The three apps have been coded on MATLAB™ and their GUIs have been designed on MATLAB GUIDE™. The standalone versions of the apps have been compiled with MATLAB Compiler™. The standalone apps require MATLAB™ and MATLAB Runtime™ installed on a computer to run (although MATLAB™ is not required to be running).
Download the app
Figure 2. Screenshot of App 1: ChipAI Tx Neuromorphic Node simulator app. (a) I-V characteristic pop-up menu. (b) RTD parameters edit fields. (c) Input bias voltage pop-up menu and edit field. (d) LD specifications pop-up menu. (e) LD Input bias current edit field. (f) Input square voltage pulse edit fields. (g) Number of realizations edit field. (h) Action buttons. (i) RTD phase plane and nullclines. (j) LD bifurcation diagram, current threshold and total injected current. (k) Tx RTD output. (l) Tx LD output. (m) Informative static text.
Fig. 3. Screenshot of App 2: ChipAI Tx-Rx Neuromorphic Network simulator app. (a) I-V characteristic pop-up menu. (b) RTD parameters edit fields. (c) Input bias voltage pop-up menu and edit field. (d) LD specifications pop-up menu and LD Input bias current edit field. (e) Tx PD conversion factor edit field. (f) Input square voltage pulse edit fields. (g) Number of realizations edit field. (h) Action buttons. (i) RTD phase plane and nullclines. (j) LD bifurcation diagram, threshold and total injected currents. (k) Tx RTD output. (l) Tx LD output. (m) Rx RTD and (n) LD outputs. (o) Informative static text.
Fig. 4. Screenshot of App 3: ChipAI simulator app accounting for the Tx-Rx Neuromorphic Network with bidirectional communication. (a) Pop-up menus and edit fields to specify the simulation parameters. (b) Action buttons. (c) RTD phase plane and nullclines. (d) LD bifurcation diagram, threshold and total injected currents. (e) Tx RTD and (f) LD outputs. (g) Rx RTD and (h) LD outputs. (i) Informative static text.
References
[1] Resonant Tunneling Diode Nano-Optoelectronic Excitable Nodes for Neuromorphic Spike-Based Information Processing. M. Hejda, J. Arturo Alanis, I. Ortega-Piwonka, J. Lourenço, J. Figueiredo, J. Javaloyes, B. Romeira, and A. Hurtado, Phys. Rev. Applied 17, 024072 (2022). DOI: https://doi.org/10.1103/PhysRevApplied.17.024072
[2] Spike propagation in a nanolaser-based optoelectronic neuron. I. Ortega-Piwonka, M. Hejda, J. Alanis, J. Lourenço, A. Hurtado, J. Figueiredo, B. Romeira, and J. Javaloyes, Opt. Mater. Express 12, 2679-2696 (2022). DOI: https://doi.org/10.1364/OME.451706
