Constant e: The Unique Intersection of Exponential and Linear Functions

  • Level: Undergrad 
  • Thread starter Jhenrique
  • Start date
  • Tags Tags
    Constant
5 replies · 2K views · last reply
Jhenrique
Messages
676
Reaction score
4
Find a number "k" such that exist only one intersection between the curve exponential k^x e and the straight x·k. This number is the constant e!

Have you noticed this? This relationship must have many important implications and indirect that we already know...
 
The [itex]e[/itex] after [itex]x^k[/itex] is unclear. Is it [itex]x^{ke}[/itex] or [itex]x^k \cdot e[/itex]?
 
Correction:

"between the curve exponential k^x and the straight x·k"

I press "e" not wanting... sorry!
 
Interesting. Do you know the proof? If you don't then I can post it if you are interested.
 
I don't know, I just observed... Yes, I'm interesting!
 
Actually, there is also only one intersection for [itex]0 < k \leq 1[/itex]. Anyway, here's the proof:

Let [itex]k > 0[/itex]. For [itex]0 < k \leq 1[/itex], we have [itex]k^x[/itex] constant or decreasing while [itex]kx[/itex] is increasing. Hence it is easy to show that there is only one intersection, in fact at [itex]x = 1[/itex].

Now assume [itex]k > 1[/itex] and let [itex]f(x) = k^x - kx[/itex] so that [itex]f(x) = 0[/itex] whenever [itex]k^x = kx[/itex]. Now we can notice a few things.

Firstly, as [itex]x \to -\infty[/itex], we have [itex]k^x \to 0[/itex] and [itex]kx \to -\infty[/itex], and so [itex]f(x) \to \infty[/itex].

Secondly, as [itex]x \to \infty[/itex], we have [itex]\frac{k^x}{kx} \to \infty[/itex] since [itex]k > 1[/itex] and so [itex]f(x) \to \infty[/itex].

Lastly, we can see that [itex]f(1) = k^1 - 1 \cdot k = 0[/itex] for any [itex]k[/itex].

Considering the limits shown above, if [itex]f(x) < 0[/itex] for some [itex]x[/itex], then there must be at least two zeros of [itex]f[/itex] by the intermediate value theorem. Thus, if there is only one intersection, we must have [itex]f(x) \geq 0[/itex] for all [itex]x[/itex]. So [itex]f[/itex] is a minimum at its only zero (which must be [itex]1[/itex] by our last observation above). Since [itex]f[/itex] differentiable, this means we must have [itex]f'(1) = 0[/itex]. Since [itex]f'(x) = \log(k)k^x - k[/itex] so the criteria [itex]f'(1) = 0[/itex] implies [itex]\log(k)k - k = 0[/itex]. But [itex]k > 0[/itex] and so [itex]\log(k) = 1 \implies k = e[/itex] which you observed.
 
  • Like
Likes   Reactions: DaTario