Rational function that approximates e^x

  • Context: Undergrad 
  • Thread starter Thread starter Ledsnyder
  • Start date Start date
  • Tags Tags
    E^x Function Rational
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
6 replies · 3K views
Ledsnyder
Messages
26
Reaction score
0
Is there a rational function,not series, that approximates e^x
?
for example (x+1)/(x+3)
 
on Phys.org
What do you mean by "approximates"? A given function, rational or not, has a specific difference from [itex]e^x[/itex]. We can always find a function, for example by truncating the Taylor's series to a polynomial, that "approximates" [itex]e^x[/itex] to a desired degree of accuracy. The difference is that the series, [itex]\sum_{n=0}^\infty[/itex] does NOT 'approximate' [itex]e^x[/itex], it is exactly equal to it.
 
That is not at all the question you asked before. However, the standard Taylor's series for the exponential, [tex]\sum_{n=0}^\infty \frac{x^n}{n!}[/tex]
is a very "rapid" approximation to the exponential.
 
HallsofIvy said:
That is not at all the question you asked before. However, the standard Taylor's series for the exponential, [tex]\sum_{n=0}^\infty \frac{x^n}{n!}[/tex]
is a very "rapid" approximation to the exponential.

Truncating the series after a fixed number of terms is not a good method of evaluating [itex]e^x[/itex] numerically for large [itex]|x|[/itex], since one loses precision from calculating the large values of [itex]x^n[/itex] and [itex]n![/itex] and the remainder may be large.

To the OP: Woolfram gives a couple of continued fraction expressions for [itex]e^x[/itex]; truncating these will give you a rational function approximation. But it is better to use a library function for exp if at all possible.
 
It depends what type of approximation you want. One common one is an approximation such that the function and its derivatives match the approximation up to some degree

$$
e^{-x}\sim P(x)/Q(x)\\
\text{where}\\
P(x)=\sum_{k=0}^m \frac{(m+n-k)!m!}{(m+n)!k!(m-k)!}x^k
\\
Q(x)=\sum_{k=0}^n \frac{(m+n-k)!n!}{(m+n)!k!(n-k)!}x^k
$$

matches up to the n+m+1 derivative

see
http://mathworld.wolfram.com/PadeApproximant.html
http://en.wikipedia.org/wiki/Padé_table#An_example_.E2.80.93_the_exponential_function
http://wwwhome.math.utwente.nl/~vajtam/publications/temp00-pade.pdf