Stochastic optimization application in molecular electronics
Husband, Christopher P.
Tour, James M.; Symes, William W.
Doctor of Philosophy
Molecular electronics is an exciting emerging technology in today's science involving interesting and challenging projects in a variety of interdisciplinary fields. There are dozens of proposed molecular electronic designs each with their advantages and disadvantages. The nanocell is our lowest level device which we believe trumps the major disadvantage inherit in most, if not all, other designs, that of the nanoscale construction of devices. As miniaturization continues into the nanoscale regime, the financial cost and the difficulty of engineering techniques to decreasing size are expected to increase sharply. The nanocell alleviates this problem by allowing molecules to self-assemble into random networks. These molecules are capable of retaining different conductive states. The challenge then is to program or reconfigure the molecular states of the nanocell post-fabrication. In essence, the nanocell idea is to shift the burden from fabrication to post-fabrication reconfiguration. We present in this thesis a description, simulations, and an analysis of a class of algorithms, called stochastic gradient descent, that may be useful in the post-fabrication programming phase.
Mathematics; Chemistry; Electronics; Electrical engineering