Talk:Relative interval error

Unclear

Now that this article is linked to from the edo template, it will get a lot of views from newbies. But it's not written very clearly. "Closest mapping" sounds like it refers to the patent val aka nearest edomapping. I changed this to "direct mapping".

I rewrote the article to make it clearer, but the "Additivity" section needs more work.

"a ratio which is the product of some other ratios have their relative errors additive" -- this is often not true for a direct mapping. And yet the article doesn't make this clear.

I also question "an edo which is the sum of some other edos have their relative errors additive". 13-edo and 18-edo both sharpen 3/2 a lot, but 31-edo flattens it slightly. --TallKite (talk) 06:44, 11 December 2020 (UTC)

For 3/2, the 31 patent val (18\31) is different from the sum of 13edo and 18edo (8\13 + 11\18). It's the 31b val (19\31) that satisfies the additivity. To find the relative error of 18\31 you must reduce the error of 19\31 by exactly 100% (1 edostep). FloraC (talk) 09:28, 11 December 2020 (UTC)

I'm not sure, but it seems that there is a bit of overlapping with the article Relative cent, at least I tend to see both as synonyms of each other. --Xenwolf (talk) 07:40, 11 December 2020 (UTC)

Relative error is a type of error, whereas relative cent is a measurement of the relative error. It's possible to use other measurements. FloraC (talk) 09:28, 11 December 2020 (UTC)

Term "indirect mapping"

Can this interpreted as "derived mapping"? Direct means that a prime is mapped to EDO steps, while indirect means some expression involvin more than one prime is mapped. Is this correct? --Xenwolf (talk) 11:18, 11 December 2020 (UTC)

To be precise, "direct" means a ratio, no matter whether it's part of the ji basis (i.e. prime) or not, is simply rounded to the closest edostep; "indirect" means using vals and monzos and take an inner product. FloraC (talk) 12:52, 11 December 2020 (UTC)