On Mon 13 Jul 2020 at 19:02:38 (-0400), The Wanderer wrote: > On 2020-07-13 at 08:23, Greg Wooledge wrote: > > On Mon, Jul 13, 2020 at 06:47:51AM -0400, The Wanderer wrote: > > > >> In some contexts it could even need fewer; for example, calc (as > >> shipped in Debian) provides the built-in function 'pi()', which > >> takes a precision - expressed as a value between zero and one - and > >> returns pi to that level of precision. > > > > I do not understand this description. How many decimal places do > > you get for a precision of, say, 0.5? > > (It turns out I got the specification of what the precision can be > expressed as wrong, and "precision" may not be the correct term for it > anyway; see below.) > > I don't understand the full algorithm myself, but the output seems > coherent: > > $ calc > C-style arbitrary precision calculator (version 2.12.7.2) > Calc is open software. For license details type: help copyright > [Type "exit" to exit, or "help" for help.] > > ; pi(.5) > 3 > ; pi(.1) > 3.1 > ; pi(.9) > 2.7 > ; pi(.05) > 3.15 > ; pi(.0005) > 3.1415 > ; pi(.00005) > 3.1416 > ; pi(.00001) > 3.14159 > ; pi(.00009) > 3.1416 >
It appears to use the argument as an integral factor of the answer quoted. However, the last example seems wrong. The answer ought to be 3.14163, which is 34907 * 0.00009. Was that a copy/paste failure? […] > Also, it turns out that it accepts values >=1 and <=0 as well: > > ; pi(5) > 5 > ; pi(10) > 0 > ; pi(50) > 0 > ; pi(3) > 3 > ; pi(4) > 4 > ; pi(2) > 4 > ; pi(-1) > 3 > ; pi(0) > zero epsilon value for pi > > ; pi(1) > 3 So it looks like the argument's sign is ignored. > The results just aren't very useful. Perhaps it's connected with the facility you mentioned earlier, the option to present non-integer output in the form of a ratio of two integers. Cheers, David.
