3 ms·
unums are at best very controversial. In actuality, I don't think I've heard a numerical analyst (save for Gustafson) come out in favor of them. The original p
by jcranmer 8y ago
unums are at best very controversial. In actuality, I don't think I've heard a numerical analyst (save for Gustafson) come out in favor of them.
The original proposal that Gustafson had involved a variable-width format, which is going to absolutely kill performance in practice (particularly on the GPU-based supercomputers that are all the rage nowadays). The posit format was essentially a recognition that variable-width doesn't work.
The more serious error is that, well, interval arithmetic turns out not to be useful in practice. Essentially, the potential accumulation of error is exponential in the number of steps in the worst case, so interval arithmetic in practice too often gives you the answer "the result is 1, but it could have been anywhere between -inf and +inf", which is completely useless. He may complain that he doesn't want to use a numerical analyst to get work done, but the rebuttal is that numerical analysis is still necessary anyways, so promising that one isn't necessary is a false promise.
- zlynx 8y agoAt least with (-inf, +inf) you know that the answer is nonsense and that you need to go back to the algorithm to find a better order of operations. With ordinary floating point, all too often people don't realize their results are nonsense. Adding measurement and computation error bounds to their graphs is illuminating.
- triska 8y agoThis is indeed a key point in John Gustafson's arguments: With floats, even though there are specialized flags and values that let you detect certain problematic situations, these flags are often not readily accessible in programming languages, and hence these issues tend to go unnoticed.
- jcranmer 8y ago> these flags are often not readily accessible in programming languages Unless you use C, C++, or Fortran. Which accounts for most numerical algorithms code. Yes, most people aren't aware of them, but not being aware of all the functions in the standard library isn't the same as them not being present there.
- jcranmer 8y agoJust because the error interval is [-inf, +inf] doesn't mean the actual error is [-inf, +inf]. There are other techniques (such as Monte Carlo simulation) that can give much narrower error bounds for the actual computation.
- Something1234 8y agoHow do I calculate measurement and error bounds for my plots when I'm using floating point? I honestly have on idea what to google.