More to the point, I guess: UX is hard ;-) The switching to scientific notation around x · 10¹⁶ hints at using IEEE 754 doubles underneath. I wonder what they

More to the point, I guess: UX is hard ;-) The switching to scientific notation around x · 10¹⁶ hints at using IEEE 754 doubles underneath. I wonder what they are using underneath for the math functions (they make it sound like it's custom-built). I spent a while in uni porting math function implementations from C to Java for a simulations framework and noticed that they usually all have the same pattern of selecting many different ways of calculating the result based on the intervals of the operands (to more or less guarantee 15 digits of precision instead of 0–15 as would otherwise be the case).