MIT Introduces a New Programming Language

A team of international researchers has developed a new programming language that can switch automatically between high-level and low-level descriptions of physical systems. It’s called – Simit. The new programming language can speed up computer simulations by 200 times and reduce the amount of code required by 90 percent. Check the blog post with other details.

Computer simulations of physical systems are widely used nowadays in science, engineering, the aerospace industry, and entertainment, but they use different tools for that. So sometimes you need to use a high-level programming language and sometimes a low-level description of the physical process and you also need to sometimes switch between these two levels of description. Such switch could be difficult not only for computer programmers, but for computers as well.

A group of international scientists from MIT’s Computer Science and Artificial Intelligence Laboratory, Adobe, the University of California at Berkley, the University of Toronto, Texas A&M, and the University of Texas have developed a programming language that handles such switches automatically and it is called Simit.

Simulations written in this language were even hundreds of times faster than those written in existing simulation languages. But it took only one-tenth of the standard code volume to run the same task.

pic2

The difference between the high-level and the low-level descriptions of physical systems is more properly described as the difference between descriptions that use linear algebra and and descriptions that use graphs. So developers who need to get into graphical descriptions of physical systems will often have to write their own code from scratch.

But Simit allows the programmers to describe the translation between the graphical description of a system and the matrix description. And then the programmer can use the language of linear algebra to program the simulation.

Programs and applications written in Simit can run on central process units (CPUs) or on graphic processing units (GPUs), with no change to the code. During experiments, the Simit code running on a GPU was between 4 and 20 times faster than on a standard chip.

Simit has applications not only for physical simulations, but in machine learning, data analytics, optimization, robotics, and other areas as well. The researchers have already successfully used the language to reproduce a version of Google’s original PageRank algorithm.

Source