That's called ball arithmetic, and there are many libraries for it, such as arb. http://fredrikj.net/arb/ It's useful in mathematical calculation, but usually with floating point calculations you have some error bounds coming from your algorithm which are "good enough" if the result is going to be the value of something with 5% tolerance.
Comments
That's called ball arithmetic, and there are many libraries for it, such as arb. http://fredrikj.net/arb/ It's useful in mathematical calculation, but usually with floating point calculations you have some error bounds coming from your algorithm which are "good enough" if the result is going to be the value of something with 5% tolerance.