This is a set of IPython notebooks and data investigating the properties of nanowire-based neuromorphic networks with the Nano Functionality Integration Group, MANA, NIMS, Tsukuba, Japan.
In this brief project, the networks investigated by the group were simulated and briefly compared to experimental results. This project is not intended for publication (the findings are not near as robustly explored) but to guide future explorations of the group.
This set of files is organized for archival purposes only. The author or lab group might be able to use the contents for future research, but documentation is probably not thorough enough for the casual unaffiliated reader to make heads or tails of the contents. This is partly because the author feels the particular research topic a dry and desolate place to look for results, and partly because only so much can be done in two months while enjoying Japan. :)
Nevertheless, an abstract seems fitting:
Alternative electronic architectures promise to transcend limitations of standard 2D photolithographic CMOS computers. For example, optical computing replaces electrons with photons for speed, while neural networks and lossy computing trade precision for efficiency. Nanowires, whose dimensions can be precisely controlled, can also be assembled into circuits approaching or exceeding CMOS performace. Here, we investigate the electrical properties of a random network of PVP@Ag nanowires for potential applications. We find evidence that individual nanowire junctions exhibit unipolar memristive behavior that, though limited, may find use in ephemeral memory or custom finite state machines.
'simulations/' contains Python Jupyter notebooks documenting most of the results.
'documents/' contains a summary paper, poster, and presentation.
'experiment/' contains experimental data, while 'pictures/' and 'plots/' contain their namesakes, respectively.
'pictures/' and 'plots/' contain various media.
This project is the result of the author's (Daniel Teal) REU-like internship organized between the US NNCI, funded by the NSF, and the Japanese NIMS.
Speaking as the author: thanks go to my lab group (the Nano Functionality Integration Group), especially my PI, Dr. Tomonobu Nakayama, the project organizer, Dr. Lynn Rathbun, NSF grant OISE-1559368, and all the others.