back to list

Simultaneous temperament approximation

🔗Gene Ward Smith <gwsmith@svpal.org>

6/30/2004 5:58:31 PM

If we have several temperaments we want to approximate, we can treat
it as a modified multiple diophantine approximation problem, modified
since we step through the lcm of the periods. Hence, to approximate
meantone and pajara simultaneously, we look for even numbers n which
give a good simultaneous approximation to both fifths. Below I give
the maximum relative error, with the rms tuning treated as optimal,
times the square root of n, where 2n is the approximating et; and list
solutions under 1000 with this badness figure less than 0.6. We can,
of course, do this for more temperaments than just two. We've
discussed 198, and Paul mentioned 88. That 12 should be good is of
course obvious.

12 .215906 -.033520 .088143
88 .580526 .087518 -.020282
100 .479848 .053997 .067861
198 .528192 -.053085 -.045636
298 .271293 .000912 .022225