Skip to main content
11 events
when toggle format what by license comment
Aug 23, 2022 at 21:57 comment added Daniel Shapero If you're asking about approximating a function by sums of rational powers, you'll want to have a look at Puiseux series. Whether or not this gives you some advantage for approximating a particular function like a Gaussian is another question; I have a hunch that you'll do better with the same number of coefficients using Padé but that's just a guess.
Aug 23, 2022 at 14:42 comment added Anti Earth @BrendanMcKay even if I have no nomination of which exponents of the uncountably infinitely many to fix, and optimise the coefficients thereof?
Aug 23, 2022 at 14:15 comment added Brendan McKay Software for nonlinear least-squares regression can find parameters. You should ask in a statistics forum.
Aug 23, 2022 at 13:53 history edited Anti Earth CC BY-SA 4.0
removing erroneous mention of Padé approximants
Aug 23, 2022 at 13:52 comment added Anti Earth @FedericoPoloni ah you helped me find a very unexpected bug in my notebook haha, thank you! Scratch my every mention of Padé approximants
Aug 23, 2022 at 13:24 comment added Federico Poloni Exactly -- how is something like $\frac{1+\frac{x}2 + \frac{x^2}{12}}{1-\frac{x}2 + \frac{x^2}{12}}$, which is the $(2,2)$ Padé approximant to $\exp(x)$, in that format?
Aug 23, 2022 at 13:05 comment added Anti Earth @FedericoPoloni I feed diagonal Padé approximants using this Mathematica function (second overload)
Aug 23, 2022 at 12:48 comment added Federico Poloni I am confused by the fact that you can feed the library Padé approximants. They don't seem to be in the acceptable form, because they have poles in places other than zero. What am I missing?
Aug 23, 2022 at 12:35 history edited YCor
edited tags
S Aug 23, 2022 at 11:57 review First questions
Aug 23, 2022 at 14:12
S Aug 23, 2022 at 11:57 history asked Anti Earth CC BY-SA 4.0