Researchers at the University of California, Berkeley have created a novel simulation visualizing over 700,000 years of star formation. The project took several months on NASA’s supercomputer, for a total of 1 million hours across thousands of processors. Since this sort of simulation is so computationally intensive, astrophysicist Richard Klein used a mathematical technique called “adaptive mesh refinement” to focus in on certain aspects of the evolving stars.

"This enables you to select what regions of a large-scale computational domain you want to focus on in a simulation, and enable you in a dynamic way to have high resolution in those areas that are critically important for understanding the astrophysical phenomenon, while letting other regions of the computation be at much lower resolution and therefore lower cost."

Before this technique, Klein says this type of simulation would have been computationally impossible.