-1

This question was given to me today on a sheet of questions containing 'sum to infinity' questions. There are no hints as to how to solve it. The series is:

$$\frac11 + \frac14+\frac2{16}+\frac3{64}+\frac5{256}+\frac8{1024}+\frac{13}{4096}+\cdots$$

How can I solve it?


I wrote the following piece of code to empirically find the answer:

n1 = 0
n2 = 1

total = 0

for i in range(1, 100): nth = n1 + n2 n1 = n2 n2 = nth

total += (nth/(4**i))

total += 1 print(total)

The program returns the answer $\frac{16}{11}$, this doesn't match the findings of some users.

  • Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. – Community Apr 29 '22 at 17:05
  • 1
    Welcome to Math.SE! ... The community prefers/expects a question to include something of what the asker knows about the problem. (What have you tried? Where did you get stuck? etc) This helps answerers tailor their responses to best serve you, without wasting time (theirs or yours) explaining things you already understand or using techniques beyond your skill level. (It also helps convince people that you aren't simply trying to get them to do your homework for you. An isolated problem statement with no evidence of personal effort makes a poor impression, attracting down- and close-votes.) – Blue Apr 29 '22 at 17:06
  • 3
    Use Binet's formula for Fibonacci numbers to get two geometric series. – anon Apr 29 '22 at 17:12
  • 3
    @Mathematician Don't put that in a comment, edit your question to improve it. Also, type your work, don't use images – jjagmath Apr 29 '22 at 17:16
  • See https://math.stackexchange.com/q/1509099/42969, https://math.stackexchange.com/q/3322436/42969, https://math.stackexchange.com/q/980937/42969, https://math.stackexchange.com/q/102706/42969. – Martin R Apr 30 '22 at 13:18

3 Answers3

4

The closed formula for Fibonacci numbers is in fact completely unnecessary.

Hint 1: Show that the series converges; terms are non-negative and the $k$-th term is less than $1/2^k$ for $k > 1$.

Hint 2: Let $S$ be the sum of the series (which is why we need to show convergence first). Observe that:

$S+S/4 = \frac11+\frac14+\frac2{16}+\frac3{64}+\frac5{256}+\cdots \\ \quad\quad\quad\quad + \frac14·\left( \frac11+\frac14+\frac2{16}+\frac3{64}+\frac5{256}+\cdots\right)$

$\quad\quad\quad\quad = \frac11+\left( \frac?4+\frac?{16}+\frac?{64}+\frac?{256}+\cdots \right)$

$\quad\quad\quad\quad = \frac11+({?}·S-{?})$.

Now find $S$. (Note that we are using the fact that $\sum_{k=1}^∞ f(k) + \sum_{k=1}^∞ g(k) = \sum_{k=1}^∞ (f(k)+g(k))$ for any real functions $f, g$ on $ℕ$ such that both $\sum_{k=1}^∞ f(k)$ and $\sum_{k=1}^∞ g(k)$ converge.)

Mittens
  • 39,145
user21820
  • 57,693
  • 9
  • 98
  • 256
4

We recognize in the numerators the Fibonacci sequence

$$F_0=0, F_1=1, F_2=1, F_3=2, \cdots$$.

Therefore, the given series can be written

$$S(x):=\sum_{n=1}^{\infty}F_nx^{n-1} \ \ \text{for} \ \ x=\frac14$$

To be compared with the classical generating function of the $F_n$:

$$\sum_{n=0}^{\infty}F_nx^n=\dfrac{x}{1-x-x^2}$$

implying that

$$S(x)=\dfrac{1}{1-x-x^2}$$

(multiplying or dividing a generating function by the variable $x$ means a shifting operation on the indices of its coefficients) giving

$$S(\tfrac14)=\dfrac{1}{1-\tfrac14-\tfrac{1}{16}}=\dfrac{16}{11}$$

Jean Marie
  • 81,803
2

$F_{n}=\frac{1}{\sqrt{5}}(\phi^n-\bar{\phi^n})$, so $\sum\frac{F_{n}}{4^{n-1}}=4\sum\frac{F_{n}}{4^{n}}=\frac{4}{\sqrt{5}}\sum\frac{\phi^n-\bar{\phi^n}}{4^n}$.\ Okey so $\sum\frac{\phi^n}{4^n}=\frac{1}{1-\frac{\phi}{4}}=\frac{4}{4-\phi}=\frac{4}{\frac{8-1-\sqrt{5}}{2}}=\frac{8}{7-\sqrt{5}}$. The same process for $\sum\frac{\bar{\phi^n}}{4^n}$, gives us that $\sum\frac{\bar{\phi^n}}{4^n}=\frac{8}{7+\sqrt{5}}$. So $4\sum\frac{F_{n}}{4^n}=\frac{4}{\sqrt{5}}\sum\frac{\phi^n-\bar{\phi^n}}{4^n}=\frac{4}{\sqrt{5}}(\frac{8}{7-\sqrt{5}}-\frac{8}{7+\sqrt{5}})=\frac{32}{\sqrt{5}}\frac{2\sqrt{5}}{44}=\frac{16}{11}$

  • $F_{n}=\frac{1}{\sqrt{5}}(\phi^n-\bar{\phi^n})$, so $\sum\frac{F_{n}}{4^n}=\frac{1}{\sqrt{5}}\sum\frac{\phi^n-\bar{\phi^n}}{4^n}$.\ Okey so $\sum\frac{\phi^n}{4^n}=\frac{1}{1-\frac{\phi}{4}}=\frac{4}{4-\phi}=\frac{4}{\frac{8-1-\sqrt{5}}{2}}=\frac{8}{7-\sqrt{5}}$. The same process for $\sum\frac{\bar{\phi^n}}{4^n}$, gives us that $\sum\frac{\bar{\phi^n}}{4^n}=\frac{8}{7+\sqrt{5}}$. So $\sum\frac{F_{n}}{4^n}=\frac{1}{\sqrt{5}}\sum\frac{\phi^n-\bar{\phi^n}}{4^n}=\frac{1}{\sqrt{5}}(\frac{8}{7-\sqrt{5}}-\frac{8}{7+\sqrt{5}})=\frac{8}{\sqrt{5}}\frac{2\sqrt{5}}{44}=\frac{4}{11}$ – Guillermo García Sáez Apr 29 '22 at 19:26
  • 1
    Fibonacii numbers start indexing at $1$. So $F_1 = 1, F_2 = 1, \ldots$. This is also what the $n$ appeared in Binet's formula. The sum at hand should be $\sum_{n=1}^\infty \frac{F_n}{4^{n-1}}$ – achille hui Apr 29 '22 at 19:28
  • Oh my bad! Sorry and thanks :) – Guillermo García Sáez Apr 29 '22 at 19:32
  • See my answer; it is much better to use shifts than to use formulae, when possible, because in many cases the recurrence relation is more fundamental than the roots of the characteristic equation. – user21820 Apr 29 '22 at 20:40