Home » Recurrence

Tag: Recurrence

Prove that the recursive sequence 1 / xn+2 = 1 / xn+1 + 1 / xn converges

Prove that the sequence \{ x_n \} be defined by the recursive relationship,

    \[ x_0 = 1, \qquad x_1 = 1, \qquad \frac{1}{x_{n+2}} = \frac{1}{x_{n+1}} + \frac{1}{x_n} \]

converges and find the limit of the sequence.


Proof. First, we show that the sequence \{ x_n \} is monotonically decreasing for all n \geq 1. For the base case we have x_1 = 1 and

    \[ \frac{1}{x_2} = \frac{1}{1} + \frac{1}{1} = 2 \quad \implies \quad x_2 = \frac{1}{2}. \]

Hence, x_2 < x_1. Assume then that for all positive integers up to some k we have x_{k+1} < x_k. Then,

    \begin{align*}  && \frac{1}{x_{k+2}} &= \frac{1}{x_{k+1}} + \frac{1}{x_k} \\[9pt]  \implies && \frac{1}{x_{k+2}} &> \frac{1}{x_{k+1}} + \frac{1}{x_{k+1}} &(x_{k+1} < x_k \implies \frac{1}{x_{k+1}} > \frac{1}{x_k}) \\[9pt]  \implies && \frac{1}{x_{k+2}} &> \frac{2}{x_{k+1}} \\[9pt]  \implies && x_{k+2} &< \frac{x_{k+1}}{2} \\[9pt]  \implies && x_{k+2} &< x_{k+1}. \end{align*}

Thus, the sequences is monotonically decreasing. The sequence is certainly bounded below since all of the terms are greater than 0. Therefore, the sequence converges. \qquad \blacksquare

To compute the limit of the sequence, assume the sequence converges to a finite limit L (justified since we just proved that it does indeed converge). Therefore,

    \begin{align*}  && \lim_{n \to \infty} x_n &= L \\[9pt]  \implies && \lim_{n \to \infty} \frac{1}{x_n} &= \frac{1}{L} \\[9pt]  \implies && \lim_{n \to \infty} \left( \frac{1}{x_{n-1}} + \frac{1}{x_{n-2}} \right) &= \frac{1}{L} \\[9pt]  \implies && \frac{1}{L} + \frac{1}{L} &= \frac{1}{L} \\[9pt]  \implies && 2L &= L \\[9pt]  \implies && L &= 0. \end{align*}

Prove that the recursive sequence xn+1 = (1+xn)1/2 converges

Prove that the sequence \{ x_n \} whose terms are defined recursively by

    \[ x_1 = 1, \qquad x_{n+1} = \sqrt{1+x_n} \]

converges, and compute the limit of the sequence.


Proof. To show the sequence converges we show that it is monotonically increasing and bounded above. To see that it is monotonically increasing we use induction to prove that

    \[ x_{n+1} > x_n \qquad \text{for all } n \in \mathbb{Z}_{>0}. \]

For the case n = 1 we have

    \[ x_{n+1} = x_2 = \sqrt{2}, \qquad \text{and} \qquad x_n = x_1 = 1. \]

Since \sqrt{2}  >1, the statement holds in the case n =1. Assume then that the statement holds for some positive integer k. Then,

    \begin{align*}  x_{k+2} - x_{k+1} &= \sqrt{1+x_{k+1}} - \sqrt{1+x_k}\\[9pt]  &> 0 \end{align*}

since x_{k+1} > x_k by the induction hypothesis. Hence, x_{k+2} > x_{k+1} so by induction x_{n+1} > x_n for all positive integers n. Hence, the sequence is monotonically increasing.
Next we use induction again to prove the sequence is bounded above by 2. For n = 1 we have x_1 = 1 < 2 so the hypothesis holds. Assume then that x_k < 2 for all positive integers up to k. Then,

    \begin{align*}  x_{k+1} &= \sqrt{1+x_k} \\  &\leq \sqrt{1+2} \\  &\leq 2. \end{align*}

Hence, x_n < 2 for all positive integers n.
This shows that the sequence converges.\qquad \blacksquare

To compute the limit, assume the sequence converges to a number L (we just proved that it converges, so this assumption is valid). Then we have

    \begin{align*}  \lim_{n \to \infty} x_{n+1} &= L & \implies && \lim_{n \to \infty} \sqrt{1+x_n} &= L \\[9pt]  && \implies && \sqrt{1+L} &= L \\[9pt]  && \implies && L^2 - L - 1 &= 0 \\[9pt]  && \implies && L &= \frac{1 + \sqrt{5}}{2}. \end{align*}

(We can discard the negative solution since to the quadratic at the end since the sequence is certainly all positive terms.)

Establish a recurrence relation for a given integral

Define

    \[ I_n = \int_0^1 (1-x^2)^n \, dx. \]

Prove that this integral satisfies the recurrence relation,

    \[ (2n+1)I_n = 2n I_{n-1}. \]

Using this compute I_2, I_3, I_4, and I_5.


Proof. We start by integrating I_n by parts with

    \begin{align*}  u &= (1-x^2)^n & du &= -2xn(1-x^2)^{n-1} \, dx \\  dv &= dx & v &= x.  \end{align*}

Therefore we have

    \begin{align*}  I_n = \int_0^1 (1-x^2)^n \, dx &= \left( uv -  \int v \, du \right) \Bigr \rvert_0^1 \\[9pt]  &= x(1-x^2)^n \Bigr \rvert_0^1 + 2n \int_0^1 x^2 (1-x^2)^{n-1} \, dx \\[9pt]  &= 2n \int_0^1 x^2 (1-x^2)^{n-1} \, dx \\[9pt]  &= 2n \int_0^1 (1-1+x^2)(1-x^2)^{n-1} \, dx \\[9pt]  &= 2n \int_0^1 (1-x^2)^{n-1} \, dx - 2n \int_0^1 (1-x^2)(1-x^2)^{n-1} \, dx \\[9pt]  &= 2n I_{n-1} - 2n I_n \\[9pt] \implies (2n+1)I_n &= 2n I_{n-1}. \qquad \blacksquare \end{align*}

Now, to compute I_2, I_3, I_4, and I_5 using the recurrence relation we first evaluate I_1 to get ourselves started,

    \[ I_1 = \int_0^1 (1-x^2) \, dx = \left(x - \frac{x^3}{3}\right)\Bigr \rvert_0^1 = \frac{2}{3}. \]

Applying the recurrence relation with n = 2 we have

    \[ 5 I_2 = 4 I_1 \quad \implies \quad I_2 = \frac{8}{15}. \]

Then, applying the recurrence relation with n = 2 (and I_2 = \frac{8}{15}) we have

    \[ 7 I_3 = 6 I_2 \quad \implies \quad I_3 = \frac{16}{35}. \]

Next, applying the recurrence relation with n = 3 (and I_3 = \frac{16}{35}) we have

    \[ 9 I_4 = 8 I_3 \quad \implies \quad I_4 = \frac{128}{315}. \]

Finally, applying the recurrence relation with n = 4 (and I_4 =\frac{128}{315}) we have

    \[ 11 I_5 = 10 I_4 \quad \implies \quad I_5 = \frac{256}{693}. \]