Skip to main content
14 events
when toggle format what by license comment
Aug 3, 2016 at 19:22 answer added Michael E2 timeline score: 9
Aug 2, 2016 at 16:53 comment added Michael @bbgodfrey: by analytical approximation I meant a closed form solution in term of elementary functions and other NPU-supported special functions ($erf, gamma$, etc) of $a,b,c$. Basically I want to plug $a,b,c$ into the analytical expression and get the above integral with accuracy $\epsilon$.
Aug 2, 2016 at 16:47 vote accept Michael
Aug 2, 2016 at 11:59 history edited J. M.'s missing motivation CC BY-SA 3.0
added 12 characters in body
Aug 2, 2016 at 11:26 answer added Mariusz Iwaniuk timeline score: 6
Aug 2, 2016 at 4:14 comment added J. M.'s missing motivation I don't agree with the proposed duplicate. OP's function has qualitatively different properties from the function in the other thread. Note that in this case both factors tend to a limit with increasing $b\gg c$, while the function in the other thread has an $\exp(x^2)$ factor whose blow-up is only mitigated by the complementary error function.
Aug 2, 2016 at 3:54 comment added bbgodfrey Note that an analytical solution does exist for c == 0, 1/4 Sqrt[π] (-Erf[a]^2 + Erf[b]^2).
Aug 2, 2016 at 1:12 comment added bbgodfrey What is your definition of analytical approximation? InterpolationFunction is an analytical approximation in the sense that it fits arrays of numbers to splines or Hermite polynomials, but I suspect that it is not what you are seeking.
Aug 2, 2016 at 0:25 review Close votes
Aug 2, 2016 at 4:15
Aug 2, 2016 at 0:06 comment added Artes Possible duplicate of Numerical underflow for a scaled error function
Aug 1, 2016 at 23:49 comment added Michael @J.M.: $-5<a,b,c<5$ would be good.
Aug 1, 2016 at 23:44 comment added J. M.'s missing motivation "within a certain range" - can you please specify those ranges for completeness?
Aug 1, 2016 at 23:44 history edited J. M.'s missing motivation
edited tags; edited tags
Aug 1, 2016 at 23:37 history asked Michael CC BY-SA 3.0