MEmory LOgic DEvice & DEsign (MeLoDe)

Memories and Logic Devices & Design lead to Intelligence.

Our group works on device physics, design and fabrication to enable advanced computing.

New devices often leverage new materials and physics e.g. memories using ionic transport and self-heating coupled with traditional electronic transport in ionic oxides. So materials exploration/understanding is an integral part.

New devices also enable new intelligent systems e.g. neurons and synapses enable neuromorphic architecture & algorithms. So we also develop a keen understanding of devices and their impact on circuits and systems.

Memory

Memory devices are essential for traditional computing. With the advent of artificial intelligence, memory-centric processing has become dominant. Ultimately, the brain requires synapses to learn and recognize – which are the memory functions.

Various memories are of interest which are non-volatile and nano-scale e,g. Resistance RAM, Ferro-electric RAM, Flash memory, etc.

Logic

Traditional computing is related to logic gates implemented in transistors. These suffer from variability and the emergence of new physics at the nano-scale.

Neuromorphic computing can leverage these new physics & related behaviors to design a rich plethora of neuronal behaviors.

Design

While traditional computing requires faster memory access which is related to memory-logic architecture co-design.

Further, neuromorphic systems require algorithms for learning and recognition, circuit implementation, and related device specifications & optimization.

Where do new device physics and design make an impact?

George Boole created the language of logic – Boolean Algebra. John von Neumann developed ideas of a von Neumann computer architecture to implement logic through digital gates where memory (programs) is stored separately. Moore’s Law leads to tremendous enhancement in high-precision computing. Yet, today there are significant challenges to such computing e.g. variability in nano-scale devices. Nano-scale devices also bring out new physics in traditional devices e.g. quantum confinement, ballistic transport, etc.

A natural computing system that works at low precision in noisy environments and performs fascinating tasks like learning and recognition is the brain. The brain has enmeshed memory and logic, analog computing, noisy behavior, and parallel information processing. The challenges to developing a neuromorphic system are at the level of algorithms (coding scheme, learning rules), architecture/circuits (networks), and devices (neurons & synapses).

Why MeLoDe?

Me Lo De (pronounced melody) is the essence of music where the individual note and the arrangement of notes in totality produce an emotion i.e. an idea that moves someone to act. A melody is beautiful and complex. It needs to be understood and felt and some more is created. Both the individual note and the arrangement are individually intricate worlds in themselves. But together, they are more that either individually. Our fascination with science and technology is an enduring one where we are fascinated by the microscopic world of device physics while observing the intricate behavior of the system. An example is how computing and cognition are built on an ensemble of simpler devices physics, and circuits. The devices and circuits are also complex interplays of different physics (e.g. joint current, heat, ion transport at the nano-scale). Another view is how microscopic statistical physics and network dynamics are able to explain systems-level computing/cognition behavior.