Raymond Wu
MIT EECS Undergraduate Research and Innovation Scholar
Optimizing Simit: A language for physical simulations
2016–2017
Electrical Engineering and Computer Science
- Artificial Intelligence & Machine Learning
Saman P. Amarasinghe
Optimizing Simit: A language for physical simulations
This project will focus on Simit which is a language for physical simulations. Many current systems that deal with physical simulations such as Matlab currently use linear algebra to describe the behavior of the physical system. However there are significant performance costs for the programmer and the machine when translating between linked data structures and matrices. Simit’s design allows it to be the first language for physical simulations that is concise expressive fast performance portable and interoperable. The language allows programmers to write code at a high level and get performance of optimized low-level code. Some further research to improve Simit’s performance can be accomplished with certain compiler optimizations and transformations.
I have taken 6.172 and 6.035. I hope to gain a greater understanding of compilers and optimizing them. I’m excited for the many challenges that the project will bring.