Magnetic materials are extremely difficult to find. They're rare in nature, and creating one in the lab usually involves both a lot of experimentation and a little luck. Duke University, however, has found a way to take the mystery out of the process: its researchers have used computer modelling to help generate two new kinds of magnetic materials. The models whittled down the potential atomic structures from a whopping 236,115 combinations to just 14 candidates by subjecting the structures to increasingly tougher tests. How stable are they? Do they have a "magnetic moment" that determines the strength of their reaction to an outside magnetic field? After that, it was just a matter of synthesizing the few remaining materials to see how well they worked in real life.