MIT EECS - Fano Undergraduate Research and Innovation Scholar
Designing an Energy Efficient Algorithm Compiler
Erik D. Demaine
Traditional models of computation include analysis of two main constraints, time and space. However with changes in technology, specifically relating to reversible logic devices, a new natural metric emerges with this reversible computation model, which we call “energy”. In our model, reversible computation (where the inputs can be recovered from the outputs) is free, but creating or destroying information costs energy. We predict that this model and the energy metric will become the dominating cost in computers within the next 10-20 years. We aim to develop and implement a simple language and compiler for coding and running energy-efficient algorithms. We would like to use this to measure the energy savings our energy-efficient algorithms achieve over existing algorithms developed for the current system.
I have extensive experience in working with algorithms and data structures through my classes at MIT, having taken 6.854 (Advanced Algorithms) and 6.851 (Advanced Data Structures). I am excited to translate what I have learned in these classes to a new application on the forefront of research. I have limited experience in implementing languages and compilers, but have resources available to me from the MIT class 6.035 (Computer Language Engineering).