# How to Prove It - Solutions

## Chapter - 6, Mathematical Induction

### Summary

• To prove a goal of the form $\forall n \in \mathbb N P(n)$:
Prove that $\forall n[(\forall k < nP(k)) \to P(n)]$, where both $n$ and $k$ range over the natural numbers in this statement. Of course, the most direct way to prove this is to let $n$ be an arbitrary natural number, assume that $\forall k < nP(k)$, and then prove $P(n)$.

• Note that no base case is necessary in a proof by strong induction. All that is needed is a modified form of the induction step in which we prove that if every natural number smaller than $n$ has the property $P$, then $n$ has the property $P$. In a proof by strong induction, we refer to the assumption that every natural number smaller than $n$ has the property $P$ as the inductive hypothesis.

Soln1

(a)

($\to$) Suppose $\forall n Q(n)$. Suppose $n$ is an arbitrary natural number. Since $n$ is natural number it follows $n + 1$ is also a natural number. Thus $Q(n+1)$ is true. Since $Q(n) = \forall k < n P(n)$, it follows $Q(n+1) = \forall k < n+1 P(k)$. Thus for $k = n$, $P(n)$ is true. Since $n$ is arbitrary, it follows $\forall n P(n)$.

($\leftarrow$) Suppose $\forall n P(n)$. It follows that $\forall k < n P(k)$ is true. Thus $Q(n)$ is true.

(b)

By ordinary induction:

Base Case:

For $n = 0$, $\forall k < n P(k)$ is true because $k$ is a natural number and there is no value of $k < 0$. Thus the statement $\forall k < n P(k)$ is vacuously true. Since $(\forall k < n P(k)) \to P(n)$ is true(given), it follows that $P(0)$ is true by modus ponens.

Induction Step:

Suppose $Q(n)$ is true. It follows that $\forall k < n P(k)$. But it is given that $(\forall k < n P(k)) \to P(n)$. Thus $P(n)$ is true. Thus $\forall k < (n+1) P(k)$ is true. It follows that $Q(n+1)$ is true.

Thus if $Q(n)$ is true then $Q(n+1)$ is also true.

Soln2

To Prove: $\forall q \in \mathbb N (q > 0 \to \lnot \exists p \in \mathbb N(p/q = \sqrt 2))$.
Or we can say $\forall q \in \mathbb N P(q)$, where $P(q) = q > 0 \to \lnot \exists p \in \mathbb N(p/q = \sqrt 2)$.

Suppose $q$ is an arbitrary natural number. Suppose $\forall k < q P(k)$ where $k \in \mathbb N$. Thus this is our induction hypothesis.

Note: We are using strong induction. Thus after assuming our induction hypothesis: $\forall k < q P(k)$, we will prove that $P(q)$ is true. Proving $P(q)$ will complete the proof.

Since $q$ is natural number, it follows $q > 0$. Thus we have to prove $\lnot \exists p \in \mathbb N(p/q = \sqrt 2)$. We shall prove this by contradiction. Thus lets suppose $\exists p \in \mathbb N(p/q = \sqrt 2)$. Thus there exist a $p \in \mathbb N$ such that $p/q = \sqrt 2$. From the example-6.4.5 from book, it follows that $p,q$ is even. Thus there are $p', q' \in \mathbb N$ such that $p' = p/2$ and $q' = q/2$. Since $p/q = \sqrt 2$, it follows $p'/q' = 2$. Since $q' = q/2$, it follows $q' < q$. Thus there exists a $q' < q$ such that $\exists p \in \mathbb N(p/q' = \sqrt 2)$. But this contradicts with our induction hypothesis. Thus our assumption that $\exists p \in \mathbb N(p/q = \sqrt 2)$ is true is wrong. Thus $\lnot \exists p \in \mathbb N(p/q = \sqrt 2)$. It follows that $P(q)$ is true.

Soln3

(a)

We will prove this by contradiction. We can follow the approach in Soln2 or we can also follow the approach used in book in example-6.4.5. Lets follows the approach used in example 6.4.5. Thus we will be using Well ordering principle.

Suppose $\sqrt 6$ is rational. Thus $\exists q \in \mathbb Z^+ \exists p \in \mathbb Z^+ (p/q = \sqrt 6)$. Thus the set $S = \{ q \in \mathbb Z^+ \, \vert \, \exists p \in \mathbb Z^+ (p/q = \sqrt 6) \}$ is not empty. Suppose $q$ is the smallest element of this set. Thus there exist $p \in \mathbb Z^+$ such that $p/q = \sqrt 6$. Thus $p^2 = 6q^2$. It follows that $p^2$ is even. Since $p^2$ is even it follows that $p$ is even. Thus suppose $p' = p/2$. Thus $p = 2p'$. Since $p^2 = 6q^2$, it follows $4{(p')}^2 = 6q^2$, or $2{(p')}^2 = 3q^2$. It follows that $q$ is also even. Thus $q = 2q'$ for some $q' \in \mathbb Z^+$. It follows $q' < q$. Also $p/q = (2p')/(2q') = p'/q' = \sqrt 6$. Thus $q' \in S$. But this contradicts with the assumption that $q$ is the smallest element. Thus $\sqrt 6$ is irrational.

(b)

Suppose $\sqrt 2 + \sqrt 3$ is rational. Thus $\exists q \in \mathbb Z^+ \exists p \in \mathbb Z^+ (p/q = \sqrt 2 + \sqrt 3)$. Suppose $p,q$ are the elements such that $p/q = \sqrt 2 + \sqrt 3$. Thus:
$\Rightarrow p^2/q^2 = 2 + 3 + 2 \sqrt 2 \sqrt 3$
$\Rightarrow p^2/q^2 = 5 + 2 \sqrt 6$
$\Rightarrow \sqrt 6 = \frac {p^2/q^2 - 5} 2$
$\Rightarrow \sqrt 6 = \frac {p^2 - 5q^2} {2q^2}$

Thus $\sqrt 6$ is rational. But this contradicts with part(a). Thus $\sqrt 2 + \sqrt 3$ is irrational.

Soln4

We need to prove that $\forall n \in \mathbb N(n \ge 12 \to \exists a \exists b (3a + 7b = n))$.

Suppose $P(n) = (n \ge 12) \to \exists a \exists b (3a + 7b = n)$. Thus we need to prove $\forall n \in \mathbb N P(n)$.

Suppose $n$ is an arbitrary natural number. Suppose $\forall k < n P(k)$ is our induction hypothesis. Thus we have following possible cases:

• Case $n < 12$:
Since $n < 12$, it follows $n \ge 12$ is false. Thus $P(n) = (n \ge 12) \to \exists a \exists b (3a + 7b = n)$ is vacuously true.

• Case $n = 12$:
Clearly $12 = 3 \times 4$. Thus $P(n)$ is true.

• Case $n = 13$: Clearly $n = 7 \times 1 + 3 \times 2$. Thus $P(n)$ is true.

• Case $n = 14$: Clearly $n = 7 \times 2$. Thus $P(n)$ is true.

• Case $n \ge 15$: Thus $n = 3 \times 1 + (n-3)$. Since $n-3 < n$, it follows from induction hypothesis that $P(n-3)$ is true. Thus there exist some $a$ and $b$ such that $n-3 = 3a + 7b$. Thus $n = 3 \times 1 + 3a + 7b$, or $n = 3 \times (a+1) + 7b$. It follows that $P(n)$ is true.

Soln5

We need to prove $\forall n \in \mathbb N (n \ge 1 \to x^n + \frac 1 {x^n})$ is integer. Suppose $n$ is an arbitrary integer. Suppose our induction hypothesis is $\forall k < n (k \ge 1 \to x^k + \frac 1 {x^k})$ is an integer. Thus we have following possible cases:

• Case $n < 1$:
Clearly $P(n)$ is vacuously true.

• Case $n = 1$:
We are given that $x + \frac 1 x$ is integer. Thus $P(n)$ is true.

• Case $n > 1$:
Consider the product: $P(1)P(n-1) = (x + \frac 1 x)(x^{n-1} + {\frac 1 {x^{n-1}} })$. Since $n-1 < n$, it follows by our induction hypothesis that $P(n-1)$ is an integer. Thus $P(1)P(n-1)$ is also an integer.

On simplifying $P(1)(n-1) = { x^n } + { \frac 1 {x^n} } + { x^{n-1} } + { \frac 1 {x^{n-1}} }$.
$\Rightarrow P(1)P(n-1) = P(n) + P(n-1)$
$\Rightarrow P(n) = P(1)P(n-1) - P(n-1)$

Thus $P(n)$ is also an integer since $P(1)P(n-1)$ is integer and $P(1)$ is integer.

Thus from all cases $P(n)$ is true.

Soln6

(a)

Suppose $n$ is arbitrary natural number. Suppose $\forall k < n (\sum_{i=0}^k F_i = F_{k+2} -1)$.

• Case $n = 1$
$\sum_{i=0}^k F_i = F_0 + F_1 = 0 + 1 = 1 = 2 - 1 = F_3 - 1$. Thus $P(1)$ is true.

• Case $n > 1$. Clearly $n-1$ is a natural number and $n - 1 < n$. Thus by induction hypothesis,
$\Rightarrow \sum_{i=0}^{n-1} F_i = F_{n+1} - 1$
Thus $\sum_{i=0}^n F_i = \sum_{i=0}^{n-1} F_i + F_n = F_{n+1} - 1 + F_n = F_{n+2} -1$.
Thus $P(n)$ is also true.

Thus from both cases, $P(n)$ is true.

(b)

Suppose $n$ is arbitrary natural number. Suppose $\forall k < n (\sum_{i=0}^k {(F_i)}^2 = F_k F_{k+1} )$.

• Case $n = 1$
$\sum_{i=0}^k {(F_i)}^2 = F_0^2 + F_1^2 = 0^2 + 1^2 = 1 = 1 \times 1 = F_1 \cdot F_2$. Thus $P(1)$ is true.

• Case $n > 1$. Clearly $n-1$ is a natural number and $n - 1 < n$. Thus by induction hypothesis,
$\Rightarrow \sum_{i=0}^{n-1} {(F_i)}^2 = F_{n-1}F_{n}$
Thus $\sum_{i=0}^n {(F_i)}^2 = \sum_{i=0}^{n-1} {(F_i)}^2 + F_n^2 = F_{n-1}F_{n} + F_n^2 = F_n(F_{n-1} + F_n) = F_n + F_{n+1}$.
Thus $P(n)$ is also true.

Thus from both cases, $P(n)$ is true.

(c)

Suppose $n$ is arbitrary natural number. Suppose $\forall k < n (\sum_{i=0}^k F_{2i+1} = F_{2k+2})$.

• Case $n = 1$
$\sum_{i=0}^k F_{2i+1} = F_1 + F_3 = 1 + 2 = 3 = F_4$. Thus $P(1)$ is true.

• Case $n > 1$. Clearly $n-1$ is a natural number and $n - 1 < n$. Thus by induction hypothesis,
$\Rightarrow \sum_{i=0}^{n-1} F_{2i+1} = F_{2(n-1)+1} = F_{2n}$
Thus $\sum_{i=0}^n F_{2i+1} = \sum_{i=0}^{n-1} F_{2i+1} + F_{2n+1} = F_{2n} + F_{2n+1} = F_{2n+2}$.
Thus $P(n)$ is also true.

Thus from both cases, $P(n)$ is true.

(d)

I made a guess as it is similar to part(c) for this formulae $\sum_{i=0}^n F_{2i+1} = F_{2n+1}$. But it does not cover the case for $n = 1$, so adjusted the formulae to $\sum_{i=0}^n F_{2i+1} = F_{2n+1} - 1$.

Suppose $n$ is arbitrary natural number. Suppose $\forall k < n (\sum_{i=0}^k F_{2i} = F_{2k+1} - 1)$.

• Case $n = 1$
$\sum_{i=0}^k F_{2i} = F_0 + F_2 = 0 + 1 = 1 = 2 - 1 = F_3 - 1$. Thus $P(1)$ is true.

• Case $n > 1$. Clearly $n-1$ is a natural number and $n - 1 < n$. Thus by induction hypothesis,
$\Rightarrow \sum_{i=0}^{n-1} F_{2i} = F_{2(n-1)+1} - 1 = F_{2n-1} - 1$
Thus $\sum_{i=0}^n F_{2i} = \sum_{i=0}^{n-1} F_{2i} + F_{2n} = F_{2n-1} - 1 + F_{2n} = F_{2n+1} -1$.
Thus $P(n)$ is also true.

Thus from both cases, $P(n)$ is true.

Soln7

(a)

Suppose $m$ is arbitrary. Suppose $\forall k < m (k \ge 1 \to (F_{k+n} = F_{k-1}F_{n} + F_kF_{n+1}))$. We have following possible cases:

• Case $m < 1$
Clearly $m \ge 1$ is false. Thus $m \ge 1 \to (F_{m+n} = F_{m-1}F_{n} + F_mF_{n+1})$ statement is vacuously true. Thus $P(m)$ is true.

• Case $m = 1$
$F_{1-1}F_n + F_{1}F_{n+1} = 0 + F_{n+1} = F_{m+1}$. Thus $P(m)$ is true.

• Case $m = 2$
$F_{2-1}F_n + F_{2}F_{n+1} = F_n + F_{n+1} = F_{n+2}$(fibonacci number). Thus $P(m)$ is true.

• Case $m > 2$
Thus $m - 1 \ge 1$ and $m-2 \ge 1$. Since $F_{m+n}$ is a fibonacci number, $F_{m+n} = F_{m+n-1} + F_{m+n-2} = F_{m-1+n} + F_{m-2+n}$. Since $m-1 < m$ and $m-2 < m$ and $m - 1 \ge 1 \land m-2 \ge 1$, it follows by induction hypothesis $F_{m-1+n} = F_{m-2}F_n + F_{m-1}F_{n+1}$ and $F_{m-2+n} = F_{m-3}F_n + F_{m-2}F_{n+1}$. Thus:

$F_{m+n} = F_{m+n-1} + F_{m+n-2}$
$= F_{m-2}F_n + F_{m-1}F_{n+1} + F_{m-3}F_n + F_{m-2}F_{n+1}$
$= F_n(F_{m-2} + F_{m-3}) + F_{n+1}(F_{m-1} + F_{m-2})$
$= F_nF_{m-1} + F_{n+1}F_m$
$= F_{m-1}F_n + F_mF_{n+1}$

Thus $P(m)$ is true.

Thus from all cases $P(m)$ is true.

(b)

Suppose $m$ is arbitrary. Suppose $\forall k < m (k \ge 1 \to (F_{k+n} = F_{k-1}F_{n} + F_kF_{n+1}))$. We have following possible cases:

• Case $m > 2$
Thus $m-1 \ge 1 \land m-2 \ge 1$. We have:
$F_{m+n} = F_{m+n-1} + F_{m+n-2}$, (fibonacci numbers property)
$= F_mF_{n+1} + F_{m-2}F_{n-1} + F_{m-1}F_{n+1} + F_{m-3}F_{n-1}$, (by induction hypothesis)
$= (F_m + F_{m-1})F_{n+1} + (F_{m-2} + F_{m-3})F_{n-1}$
$= F_{m+1}F_{n+1} + F_{m-1}F_{n-1}$

• Case $m = 2$
• Case $m = 1$
For both of these cases, it can be directly verified by putting the values for $m$.

(c)

Both parts can be proved easily by putting $m = n+1$ in part(a) and putting $m = n+1, n = n+1$ in part(b).

(d)

We need to prove following:

$\forall m \in \mathbb N \forall n \in \mathbb N ( \exists p \in \mathbb N(pm = n) \to \exists q \in \mathbb N(qF_m = F_n))$.

Suppose $m,n$ are arbitrary natural numbers such that $lm = n$ for some natural number $l$. Suppose $\forall k < n (\exists p \in \mathbb N(pm = k) \to \exists q \in \mathbb N(qF_m = F_k))$. We have following possible cases:

• Case $m = n$
Since $m = n$, it follows $F_m = F_n$. Thus for $q = 1$, $qF_m = F_n$.

• Case $n > m$
We have $F_n = F_{(n-m)+m}$. Since $n > m$, it follows $n-m$ is a natural number. Thus we can apply part(a). Applying part(a), we get $F_n = F_{n-m-1}F_m + F_{n-m}F_{m-1}$.

Since $n = lm$, it follows $n-m = lm-m = (l-1)m$. Thus $n-m$ is a multiple of $m$. Or we can say $tm = n-m$, where $t = l-1$. Since $n-m < n$ and $tm = n-m$, it follows from the induction hypothesis, there exist some natural number $q'$ such that $q'F_m = F_{n-m}$.

Thus $F_n = F_{n-m-1}F_m + F_{n-m}F_{m-1} = F_{n-m-1}F_m + q'F_mF_{m-1}$, or $F_n = (F_{n-m-1} + q'F_{m-1})F_m$. Thus for $q = (F_{n-m-1} + q'F_{m-1})$, we have $F_n = qF_m$.

• Case $n < m$
Since $lm = n$, for some natural number $l$, it follows that this case is not possible.

(e)

Suppose $n$ is arbitrary natural number. Suppose:

$\forall k < n (F_{2k-1} = \sum_{i=0}^{k-1} \binom {2k-i-2} k$
$\forall k < n (F_{2k} = \sum_{i=0}^{k-1} \binom {2k-i-1} k$

We know that $F_n = F_{n-1} + F_{n-2}$. Thus we have following possible cases:

• Case $n = 1$:
Thus $\sum_{i=0}^{n-1} \binom {2n-i-2} n = \binom {2-0-2} 0 = \binom 0 0 = 1 = F_1 = F_{2n-1}$.
Also $\sum_{i=0}^{n-1} \binom {2n-i-1} n = \binom {2-0-1} 0 = \binom 1 0 = 1 = F_2 = F_{2n}$.

• Case $n = 2$:
Thus $\sum_{i=0}^{n-1} \binom {2n-i-2} n = \binom {4-0-2} 0 + \binom {4-1-2} 1 = \binom 2 0 + \binom 1 1 = 1+1 = F_3 = F_{2n-1}$.
Also $\sum_{i=0}^{n-1} \binom {2n-i-1} n = \binom {4-0-1} 0 + \binom {4-1-1} 1 = \binom 3 0 + \binom 2 1 = 1+2 = F_4 = F_{2n}$.

• Case $n > 2$:

Since $n > 2$, it follows $2(n-1) \ge 4$ and $2(n-1)-3 \ge 1$. Thus $F_{2(n-1)}$ and $F_{2(n-1)-1}$ are both valid fibonacci numbers.

Since $F_{2n-1}$ is a fibonacci number, it follows:
$F_{2n-1} = F_{2n-2} + F_{2n-3} = F_{2(n-1)} + F_{2(n-1)-1}$

Since $n > 2$, it follows $n-1$ is a natural number. Since $n-1 < n$ and $n-1$ is a natural number, it follows from the induction hypothesis:

$F_{2(n-1)} = \sum_{i=0}^{n-2} \binom {2(n-1)-i-1} {n-1} = \sum_{i=0}^{n-2} \binom {2n-i-3} {n-1}$ and
$F_{2(n-1)-1} = \sum_{i=0}^{n-2} \binom {2(n-1)-i-2} {n-1} = \sum_{i=0}^{n-2} \binom {2n-i-4} {n-1}$

Thus $F_n$:
$= \sum_{i=0}^{n-2} \binom {2n-i-3} {n-1} + \sum_{i=0}^{n-2} \binom {2n-i-4} {n-1}$
$= {\binom {2n-3} 0} + {\binom {2n-4} 1} + \cdot \cdot \cdot + {\binom {n} {n-3} } + {\binom {n-1} {n-2}}$
$\quad\quad\quad\;\; + {\binom {2n-4} 0} + \cdot \cdot \cdot + {\binom {n} {n-4} } + {\binom {n-1} {n-3}} + {\binom {n-2} {n-2}}$

We can see above many pairs of the form of $\binom {n} {k}$ and $\binom {n} {k-1}$. Thus we can use the theorem $\binom {n+1} k = \binom n {k-1} + \binom n {k}$. Thus except the first and last term all the pairs in the above summation can be combined. Thus after combining we get the following:

$= {\binom {2n-3} 0} + {\binom {2n-3} 1} + \cdot \cdot \cdot + { \binom {n+1} {n-3} } + { \binom {n} {n-2} } + { \binom {n-2} {n-2} }$
$= 1 + {\binom {2n-3} 1} + \cdot \cdot \cdot + { \binom {n+1} {n-3} } + { \binom {n} {n-2} } + 1$
$= {\binom {2n-2} 0} + {\binom {2n-3} 1} + \cdot \cdot \cdot + { \binom {n+1} {n-3} } + { \binom {n} {n-2} } + { \binom {n-1} {n-1} }$
$= \sum_{i=0}^{n-1} \binom {2n-i-2} n$

Thus $F_{2n-1}$ is true in all cases.

Similarly we can prove for $F_{2n}$.

Soln8

(a)

We need to prove the following:
$\forall n \in \mathbb N ((a_n = a_{n-1} + a_{n-2}) \leftrightarrow c = \frac {1 \pm \sqrt 5} 2 )$

($\to$) Suppose $n$ is an arbitrary integer. Suppose $a_n = a_{n-1} + a_{n-2}$. Since $a_n = c^n$, it follows that $c^n = c^{n-1} + c^{n-2}$. Thus $c^n = c^{n-2}(c+1)$, or $c^{n-2}(c^2 - c - 1) = 0$.

If $n = 2$ then $c^{n-2} = c^0 = 1$. Thus $c \ne 0$. Since $0^0$ is not defined.

Since $c \ne 0$ and $c^{n-2}(c^2 - c - 1) = 0$, it follows $c^2 - c - 1 = 0$. Thus $c = \frac {1 \pm \sqrt 5} 2$.

($\leftarrow$) Suppose $c = \frac {1 \pm \sqrt 5} 2$. It follows $c^2 = 1 + c$, or $c^2 - c - 1 = 0$.

(b)

Suppose $c_1 = \frac {1 + \sqrt 5} 2$ and $c_2 = \frac {1 - \sqrt 5} 2$. Thus from part(a) we know that $c_1^2 = c_2^2 = 1+c_1 = 1+c_2$.

Suppose $n \ge 2$ is a arbitrary natural number. Since $a_n = s{(\frac {1 + \sqrt 5} 2)}^n + t{(\frac {1 - \sqrt 5} 2)}^n$, or $a_n = sc_1^n + tc_2^2$ it follows:

$a_{n-1} = sc_1^{n-1} + tc_1^{n-1}$
$a_{n-2} = sc_1^{n-2} + tc_1^{n-2}$

Thus $a_{n-1} + a_{n-2} = sc_1^{n-1} + tc_1^{n-1} + sc_1^{n-2} + tc_1^{n-2}$. Thus:
$\Rightarrow a_{n-1} + a_{n-2} = s(c_1^{n-1} + c_2^{n-2}) + t(c_2^{n-1} + c_2^{n-2})$
$\Rightarrow a_{n-1} + a_{n-2} = sc_1^{n-2}(c_1+1) + tc_2^{n-2}(c_2 + 1)$
But $1+c_1 = c_1^2$ and $1+ c_2 = c_2^2$
$\Rightarrow a_{n-1} + a_{n-2} = sc_1^{n-2}(c_1^2) + tc_2^{n-2}(c_2^2)$
$\Rightarrow a_{n-1} + a_{n-2} = sc_1^{n} + tc_2^{n}$
$\Rightarrow a_{n-1} + a_{n-2} = a_n$

Thus $a_0, a_1, a_2 ...$ is a gibonacci sequence.

(c)

Scratch Work

We will first compute $s$ and $t$ by assuming that formulae is correct.

Thus for $n = 0$, we have $a_0 = sc_1^0 + tc_2^0 = s+t$. Or $s+t = a_0$.
Similarly for $n = 1$, we have $a_1 = sc_1^1 + tc_2^1 = sc_1 + tc_2$. Or $a_1 = s(\frac {1 + \sqrt 5} 2) + t(\frac {1 - \sqrt 5} 2)$.
Thus we get $a_1 = \frac {s+t + \sqrt 5(s-t)} 2 = \frac {a_0 + \sqrt 5{s-t}} 2$.

Thus $s-t = \frac {2a_1 - a_0} {\sqrt 5}$ and $s+t = a_0$. We have two equations with two unknowns $s$ and $t$.

Solving them gives:
$s = \frac {5a_0 + \sqrt 5(2a_1 - a_0)} {10}$
$t = \frac {5a_0 - \sqrt 5(2a_1 - a_0)} {10}$

Now suppose $s = \frac {5a_0 + \sqrt 5(2a_1 - a_0)} {10}$ and $t = \frac {5a_0 - \sqrt 5(2a_1 - a_0)} {10}$.

Suppose $n$ is an arbitrary natural number. Suppose $\forall k < n(a_n = sc_1^n + tc_2^n)$. Thus we have following possible cases:

• Case $n = 0$:
Clearly $sc_1^0 + tc_2^0 = s+t = \frac {5a_0 + \sqrt 5(2a_1 - a_0)} {10} + \frac {5a_0 - \sqrt 5(2a_1 - a_0)} {10} = a_0$. Thus $a_n = sc_1^n + t c_2^n$.

• Case $n = 1$:
Clearly $sc_1^1 + tc_2^1 = sc_1 + tc_2 = (\frac {1 + \sqrt 5} 2)(\frac {5a_0 + \sqrt 5(2a_1 - a_0)} {10}) + (\frac {1 - \sqrt 5} 2)(\frac {5a_0 - \sqrt 5(2a_1 - a_0)} {10}) = a_1$. Thus $a_n = sc_1^n + t c_2^n$.

• Case $n \ge 2$:
We have $sc_1^n + tc_2^n = sc_1^{n-2}c_1^2 + tc_2^{n-2}c_2^2$. From part(a) we know $c_1^2 = 1+ c_1$ and $c_2^2 = 1+c_2^2$. It follows $sc_1^n + tc_2^n = sc_1^{n-2}(1 + c_1) + tc_2^{n-2}(1 + c_2)$
$= sc_1^{n-2} + sc_1{n-1} + tc_2^{n-2} + tc_2{n-1}$
$= sc_1^{n-2} + tc_2^{n-2} + sc_1{n-1} + tc_2{n-1}$
$= a_{n-2} + a_{n-1}$
$= a_{n}$

Thus $a_n = sc_1^n + t c_2^n$.

Thus for all cases $a_n = sc_1^n + tc_2^n$.

Soln9

Since it is a gibonacci sequence and $a_0 = L_0 = 2$ and $a_1 = L_1 = 1$ is given. We can use part(c). Thus $a_n = sc_1^n + tc_2^n$, where $s = \frac {5a_0 + \sqrt 5(2a_1 - a_0)} {10}$ and $t = \frac {5a_0 - \sqrt 5(2a_1 - a_0)} {10}$.

Thus $s = \frac {10 + \sqrt 5(2 - 2)} {10} = 1$
and $t = \frac {10 - \sqrt 5(2 - 2)} {10} = 1$

Thus $a_n = c_1^n + c_2^n$, where $c_1, c_2$ are from Soln8.

Soln10

Lets suppose $a_n = sp^n - tq^n$ which is similar to Soln8 except negative sign. Addition will also generate same result but it appears from the formulae $a_n = 5a_{n-1} - 6a_{n-2}$ that negative sign may help.

Now we have:
$a_0 = -1$
$a_1 = 0$

Using $a_n = 5a_{n-1} - 6a_{n-2}$, we have $a_2 = 6$ and $a_3 = 30$.

Now using our assumption,
$a_0 = sp^0 - tq^0 = s-t = -1$
$a_1 = sp^1 - tq^1 = sp-tq = 0$
$a_2 = sp^2 - tq^2 = 6$
$a_2 = sp^3 - tq^3 = 30$

Thus we have four equations and four unknowns $s,t,p,q$.

Using first and seconds equation we can find values of $s$ and $t$ in terms of $p$ and $q$. Thus we get:
$s = \frac q {p-q}$
$t = \frac p {p-q}$

Now putting these values of $s$ and $t$ in $3^{rd}$ and $4^th$ equation:
In $3rd$ equation: $\Rightarrow \frac {qp^2 - pq^2} {p-q} = 6$, or $\frac {pq(p-q)} {p-q} = 6$. Thus $pq = 6$.
and In $4th$ equation: $\Rightarrow \frac {qp^3 - pq^3} {p-q} = 30$, or $\frac {pq(p^2-q^2)} {p-q} = 30$.
Using $a^2 - b^2 = (a-b)(a+b)$,
$\Rightarrow pq(p+q) = 30$
Using $pq = 6$:
we get $\Rightarrow p+q = 5$.

Thus with $pq = 6$ and $p+q = 5$ and eliminating $q$, we get the quadratic equation $p^2 - 2p + 6 = 0$. Solving it gives $p = 2,3$. Since $pq = 6$, we get $q = 3,2$.

Thus we have two pair of values for $(p,q) = (2,3)$ or $(p,q) = (3,2)$.

With $(p,q) = (2,3)$, we get $s = -3$ and $t = -2$. Thus $a_n = -3 \cdot 2^n + 2 \cdot 3^n$. Using $(p,q) = (3,2)$, we get $s = 2$ and $t = 3$. Thus $a_n = 2 \cdot 3^n - 3 \cdot 2^n$.

Thus from both values we get the same formulae $a_n = 2 \cdot 3^n - 3 \cdot 2^n$.

Lets prove that formulae is correct:

Suppose $n$ is arbitrary natural number. Suppose $\forall k < n (a_k = 2 \cdot 3^k - 3 \cdot 3^k)$. Thus we have following possible values of $n$:

• Case $n = 0$
$2 \cdot 3^0 - 3 \cdot 2^0 = 2-3 = -1 = a_0$. Thus $a_n = 2 \cdot 3^n - 3 \cdot 2^n$.

• Case $n = 1$
$2 \cdot 3^1 - 3 \cdot 2^1 = 2-3 = 0 = a_1$. Thus $a_n = 2 \cdot 3^n - 3 \cdot 2^n$.

• Case $n \ge 2$
Since $n-1 < n$, by induction hypothesis $a_{n-1} = 2 \cdot 3^{n-1} - 3 \cdot 2^{n-1}$. Similarly since $n-2 < n$, by induction hypothesis we get $a_{n-2} = 2 \cdot 3^{n-2} - 3 \cdot 2^{n-2}$.

Thus $5a_{n-1} - 6a_{n-2} = 5(2 \cdot 3^{n-1} - 3 \cdot 2^{n-1}) - 6(2 \cdot 3^{n-2} - 3 \cdot 2^{n-2})$
$= 10 \cdot 3^{n-1} - 12 \cdot 3^{n-2} - 15 \cdot 2^{n-1} + 18 \cdot 2^{n-2}$
$= 10 \times 3 \cdot 3^{n-2} - 12 \cdot 3^{n-2} - 15 \times 2 \cdot 2^{n-2} + 18 \cdot 2^{n-2}$
$= 30 \cdot 3^{n-2} - 12 \cdot 3^{n-2} - 30 \cdot 2^{n-2} + 18 \cdot 2^{n-2}$
$= 18 \cdot 3^{n-2} - 12 \cdot 2^{n-2}$
$= 2 \cdot 3^n - 3 \cdot 2^n$
$= a_n$

Thus $a_n = 2 \cdot 3^n - 3 \cdot 2^n$.

Thus $a_n = 2 \cdot 3^n - 3 \cdot 2^n$ from all the cases.

Soln11

Suppose $n$ is an arbitrary natural number. Suppose $\forall k < n (a_k = a_{k-1} + a_{k-2})$. Thus we have following possible cases:

• Case $n \le 2$
Clearly from the given values $a_0, a_1, a_2$ are fibonacci numbers.

• Case $n \ge 3$
We have $a_n = \frac {a_{n-3}} 2 + \frac {3a_{n-2}} 2 + \frac {a_{n-1}} 2$
$= \frac {a_{n-3}} 2 + a_{n-2} + \frac {a_{n-2}} 2 + \frac {a_{n-1}} 2$
$= \frac {a_{n-3} + a_{n-2}} 2 + a_{n-2} + \frac {a_{n-1}} 2$
Since $n-3 < n$ and $n-2 < n$, thus by our induction hypothesis, we have $a_{n-3} + a_{n-2} = a_{n-1}$.
$= \frac {a_{n-1}} 2 + a_{n-2} + \frac {a_{n-1}} 2$
$= {a_{n-1}} + a_{n-2}$

Thus $a_n = a_{n-1} + a_{n-2}$.

Thus the given sequence is fibonacci sequence.

Soln12

Suppose $n$ is an arbitrary natural number. Suppose $\forall k < n (P_n \text{ contains } F_{n+2} \text{ elements. })$. We have following possible cases:

• Case $n = 0$
Clearly $P_0 = \{ \phi \}$. Thus number of elements are $1$. Thus $P_n = F_{n+2}$.

• Case $n = 1$
Clearly $P_1 = \{ \phi, 1 \}$. Thus number of elements are $2$. Thus $P_n = F_{n+2}$.

• Case $n > 1$
We can easily see that $P_n$ will contain all the elements contained in $P_{n-1}$ and all elements of the set $S = \{ X \cup \{n \} \, \vert \, X \in P_{n-2} \}$. Since $X \in P_{n-2}$, it does not contain the element: “$n-1$”. Thus $X \cup \{ n \}$ will not have any consecutive numbers.
Clearly number of elements in $S$ is same as number of elements in $P_{n-2}$. Thus total number of elements:
$P_n = P_{n-1} + P_{n-2}$.
Since $n-1 < n$ and $n-2 < n$, it follows by induction hypothesis, $P_{n-1} = F_{n+1}$ and $P_{n-2} = F_n$. Thus $P_n = P_{n-1} + P_{n-2} = F_{n+1} + F_{n} = F_{n+2}$.

Thus from all the cases we get $P_n = F_{n+2}$.

Soln13

(a)

Suppose $n$ is an arbitrary integer. Consider the following cases:

• Case $n \ge 0$
Clearly it follows from the theorem 6.4.1 from the book that $n = mq+r$ and $0 \le r < m$.

• Case $n < 0$
Since $n < 0$, it follows $-n > 0$. Thus from theorem 6.4.1:
$\Rightarrow -n = mq+r$
$\Rightarrow n = -mq-r$
$\Rightarrow n = -mq-r-m+m$
$\Rightarrow n = -mq-m+m-r$
$\Rightarrow n = (-q-1)m+m-r$
$\Rightarrow n = q'm+r'$, where $q' = -q-1$ and $r' = m - r$.

Since $0 \le r < m$, it follows that $0 < m-r < m$, or $0 < r' < m$.

(b)

Suppose $q_1, r_1$ and $q_2, r_2$ are integers such that:
$n = mq_1 + r_1$, where $0 \le r_1 < m$.
and $n = mq_2 + r_2$ and $0 \le r_2 < m$.
It follows that $mq_1 + r_1 = mq_2 + r_2$.
Thus $(q_1-q_2)m = r_2 - r_1$.

Suppose $q_1 = q_2 + c$, where $c$ is an integer. Thus we get:
$\Rightarrow (q_2 + c - q_2)m = r_2 - r_1$
$\Rightarrow cm = r_2 - r_1$

We have following possible cases:

• Case $c \ne 0$
It follows $m = \frac {r_2 - r_1} c$. Thus $\vert \, m \, \vert = \frac {\vert \, r_2 - r_1 \, \vert} {\vert c \vert}$.
Since $m > 0, \Rightarrow \vert \, m \, \vert = m$. Thus $m = \frac {\vert \, r_2 - r_1 \, \vert} {\vert c \vert}$. Since $c \ne 0$, it follows $m \le \vert \, r_2 - r_1 \, \vert$.

Since $0 \le r_1 < m$ and $0 \le r_1 < m$, it follows that $0 \le \vert \, r_2 - r_1 \, \vert < m$. Thus it is a contradiction since $m \le \vert \, r_2 - r_1 \, \vert$.

Thus this case $c \ne 0$ is not possible.

• Case $c = 0$.
Since $q_1 = q_2 + c$, it follows $q_1 = q_2$. Also since $cm = r_2 - r_1$, it follows $r_1 = r_2$.

Thus from all possible cases $q_1 = q_2$ and $r_1 = r_2$. Thus we can conclude that $q$ and $r$ are unique integers.

(c)

Suppose $n$ is an arbitrary integer. From part(a), we can easily deduce that if $m = 2$, then $n = 2q + r$ such that $0 \le r < 2$. It follows that there are only two possible values($0,1$) of $r$. Thus if $r = 0$, it follows $n = 2q$ and if $r = 1$, it follows $n = 2q+1$.

Clearly if $n = 2q$, then $n$ is an even integer. and if $n = 2q+1$ then $n$ is an odd integer.

Also from part(b), since $n = 2q+r$, then $q$ and $r$ are unique. Thus only one case is possible i.e. either $n$ is even or $n$ is odd but not both.

Since $n$ is arbitrary, it follows that every integer is either even or odd but not both.

Soln14

Suppose $a$ is the maximum of $5k$ and $k(k+1)$. Suppose $n > a$ is an arbitrary integer. Thus by the division algorithm we can choose unique $q$ and $r$ such that $n = kq + r$ and $0 \le r < k$.

Suppose $q \le 4$. Thus $n = kq + r \le 4k + r$. Since $0 \le r < k$, it follows $n \le 4q + r < 4k + k = 5k \le a$. Thus $n < a$ which is a contradiction. Thus $q > 4$, or $q \ge 5$.

Suppose $q < k+1$. Thus $n = kq + r < k(k+1) + r < k(k+1) + r < k(k+1) \le a$. Thus $n < a$ which is a contradiction. Thus $q \ge k+1$.

Since $0 \le r < k$, it follows $r \le k+1$. Since $q \ge k+1$ and $r \le k+1$, it follows $q \ge r$.

Thus $q^2 \ge q(k+1) = qk + q \ge qk + r = n$. Thus $q^2 \ge n$.

From example 6.1.3, we know that $2^n > n^2$, if $n \ge 5$. Since $q \ge 5$, it follows $2^q > q^2$. Since $q^2 \ge n$, it follows $2^q > n$.

Since $n = kq + r$ and $0 \le r < k$, it follows $2^n \ge 2^{kq} = {2^q}^k$. Since $2^q > n$, it follows $2^n \ge {2^q}^k > n^k$.

Soln15

(a)

Suppose $k$ is a positive integer. Suppose $\forall p < k (a_1 f_1 + a_2 f_2 + \cdot \cdot \cdot + a_p f_p) \in \mathcal O(g)$.

Since $k-1 < k$, it follows from induction hypothesis that there exist some $a \in \mathbb Z^+$ and $c \in R$ such that $\forall n > a ( \vert \, a_1 f_1(n) + a_2 f_2(n) + \cdot \cdot \cdot + a_{k-1} f_{k-1}(n) \, \vert \le c \vert \, g(n) \, \vert$.

Since $f_k \in \mathcal O(g)$, it follows that for some $a' \in \mathbb Z^+$ and $c' \in \mathbb R$, $\vert \, f_k(n) \, \vert \le c' \vert \, g(n) \, \vert$.

Now consider $\vert \, a_1 f_1(n) + a_2 f_2(n) + \cdot \cdot \cdot + a_{k-1} f_{k-1}(n) + a_k f_k(n) \, \vert$
$\le \vert \, a_1 f_1(n) + a_2 f_2(n) + \cdot \cdot \cdot + a_{k-1} f_{k-1}(n) \, \vert + \vert \, a_k f_k(n) \, \vert$ (By triangle inequality)
$\le c \vert \, g(n) \, \vert + \vert \, a_k f_k(n) \, \vert$ for all $n > a$ (By induction hypothesis for k-1)
$\le c \vert \, g(n) \, \vert + c' \vert \, g(n) \, \vert$ for all $n > a''$ where $a''$ is maximum of $a$ and $a'$
$= c'' \vert \, g(n) \, \vert$ where $c'' = c+c'$

Thus $f \in \mathcal O(g)$ where $f(n) = a_1 f_1(n) + a_2 f_2(n) + \cdot \cdot \cdot + a_k f_k(n)$.

(b)

From Ex14 part(a), we know that for every positive integer $k$, $\forall n > a (2^n \ge n^k)$ for some positive integer $a$. It follows that functions $f_1 = 1 = n^0$, $f_2 = n^2$, … $f_k = n^k$ belongs to O(g) where $g = 2^n$.

Now the required proof directly follows from part(a) of this solution.

Soln16

(a)

By division algorithm we know that for some $q$ and $r$, $a = dq + r$ where $0 \le r < d$.

Also since $d \in S$, it follows that for some $s,t \in \mathbb Z$, $d = as + bt$. Putting $d = as + bt$ in $a = dq + r$, we get:
$\Rightarrow a = (as+bt)q + r$
$\Rightarrow a-asq = btq + r$
$\Rightarrow (1-sq)a + (-tq)b = r$
Thus s’ = 1-sq $and$ t’ = -tq $,$ r = s’a + t’b .

Since $0 \le r$, we have two cases:

• Case $r \ne 0$
Thus $r > 0$ or $r \in Z^+$. It follows that $r \in S$. Since $d$ is the smallest element in $S$, it follows $d \le r$. But this contradicts with $0 \le r < d$. Thus $r \ne 0$ is not possible.

• Case $r = 0$
Thus $r \notin Z^+$. Thus $r \notin S$. It follows $a = dq+r = dq+0 = dq$. Thus $d \, \vert \, a$.

Similarly it can be proved that $d \, \vert \, b$.

(b)

Since $c \, \vert \, a$, it follows that for some integer $p$, $a = pc$. Similarly since $c \, \vert \, b$, it follows that for some integer $q$, $b = qc$.
Since $d \in S$, it follows that $d = as + bt$. Thus $d = pcs + qct = c(ps+qt)$. Thus $c \, \vert \, d$.

Also since from part(a), $d$ divides both $a$ and $b$. Thus any number which divides $a$ and $b$ must divides $d$. Thus $c \le d$. It follows that $d$ is greatest common divisor of $a$ and $b$.

Soln17

(a)

Suppose $a,b,p$ are natural numbers such that $p \, \vert \, ab$ and $p$ is prime. Suppose $d$ is the greatest common divisor of $a$ and $p$. Thus from Ex-16 part(b), $d = as + pt$ for some integers $s, t$. Since $d$ divides $p$ and $p$ is a prime number, there are following possible cases:

• Case $d = p$
Since $d$ is greatest common divisor of $a$ and $p$, it follows $d$ divides $a$. Since $d = p$, it follows $p$ divides $a$, or $p \, \vert \, a$.

• Case $d = 1$
Since $d = as + pt$, it follows $1 = as + pt$. Since $p \, \vert \, ab$, it follows for some integer $q$, $ab = pq$. Thus $a = \frac {pq} b$. Putting $a = \frac {pq} b$ in $as + pt = 1$, we get $\frac {pqs} b + pt = 1$, or $pqs + pbt = b$, or $p(qs + bt) = b$. Thus $p \, \vert \, b$.

Thus from both cases, either $p \, \vert \, a$ or $p \, \vert \, b$.

(b)

Suppose $n$ is arbitrary. Suppose $\forall k < n ( p \, \vert \, a_1 a_2 \cdot \cdot \cdot a_k \to (p \, \vert \, a_1 \lor p \, \vert \, a_2 \lor \cdot \cdot \cdot \lor p \, \vert \, a_k) )$.

Suppose $p \, \vert \, a_1 a_2 \cdot \cdot \cdot a_{n-1} a_n$, or $p \, \vert \, (a_1 a_2 \cdot \cdot \cdot a_{n-1}) \, a_n$. Thus from part(a), it follows either either $p \, \vert \, a_1 a_2 \cdot \cdot \cdot a_{n-1}$ or $p \, \vert \, a_n$.

Since $n-1 < n$, it follows from induction hypothesis that if $p \, \vert \, a_1 a_2 \cdot \cdot \cdot a_{n-1}$, then $p \, \vert \, a_1 \lor p \, \vert \, a_2 \lor \cdot \cdot \cdot \lor p \, \vert \, a_{n-1}$.

Thus it follows that $p \, \vert \, a_1 \lor p \, \vert \, a_2 \lor \cdot \cdot \cdot p \, \vert \, \lor a_{n-1} \lor p \, \vert \, a_n$.

Soln18

Base Case:

Suppose $j = 1$. Since $p_1 = q_1 q_2 \cdot \cdot \cdot q_k$ and $p_1$ is prime it follows that $k = 1$. Thus $p_1 = q_1$.

Induction Step:

Suppose for $j \ge 1$ and for all $k \ge 1$, if $p_1 p_2 \cdot \cdot \cdot p_j = q_1 q_2 \cdot \cdot \cdot q_k$, then $j = k$ and $p_i = q_i$ where $1 \le i \le j$.

Now suppose $p_1 p_2 \cdot \cdot \cdot p_{j+1} = q_1 q_2 \cdot \cdot \cdot q_k$. Clearly $k \ge 2$, since if $k = 1$ then $p_1 p_2 \cdot \cdot \cdot p_{j+1} = q_1$. Thus $q_1$ will not be prime.

Note: $j$ is same in last two statements. But $k$ is different. In former statement $j = k$ if $p_1 p_2 \cdot \cdot \cdot p_j = q_1 q_2 \cdot \cdot \cdot q_k$, but in the later statement above, $k$ may or may not be equal to $j$. Or first statement is hypothesis $P(j)$ and second statement is $P(j+1)$ which we shall prove to be true if hypothesis is true.

Clearly $p_{j+1} \; \vert \; p_1 p_2 \cdot \cdot \cdot p_{j+1}$. It follows $p_{j+1} \; \vert \; q_1 q_2 \cdot \cdot \cdot q_k$. Thus from Ex17 part(b), it follows $p_{j+1} \; \vert \; q_i$, where $1 \le i \le k$. But $q_i \le q_k$. Thus $p_{j+1} \le q_k$.

Also since $q_k \; \vert \; q_1 q_2 \cdot \cdot \cdot q_k$, it follows $q_k \; \vert \; p_1 p_2 \cdot \cdot \cdot p_{j+1}$. Thus from Ex17 part(b), it follows $q_k \; \vert \; p_i$, where $1 \le i \le j+1$. But $p_i \le p_{j+1}$. Thus $q_k \le p_{j+1}$.

Thus $p_{j+1} = q_k$. Now since $p_1 p_2 \cdot \cdot \cdot p_{j+1} = q_1 q_2 \cdot \cdot \cdot q_k$, it follows $p_1 p_2 \cdot \cdot \cdot p_{j} = q_1 q_2 \cdot \cdot \cdot q_{k-1}$.

But from the induction hypothesis, if $p_1 p_2 \cdot \cdot \cdot p_{j} = q_1 q_2 \cdot \cdot \cdot q_{k-1}$ then $p_i = q_i$ and $j = k-1$. Thus $j+1 = k$.

Thus if $p_1 p_2 \cdot \cdot \cdot p_{j+1} = q_1 q_2 \cdot \cdot \cdot q_k$, then $j+1 = k$ and $p_i = q_i$.

Soln19

Since $a_{n+1} = 1 + \sum_{i=0}^{n} a_i$, it follows $a_n = 1 + \sum_{i=0}^{n-1} a_i$.

$n$ $a_n = 1 + \sum_{i=0}^{n-1} a_i$
$0$ $1 = 2^0$
$1$ $2 = 2^1$
$2$ $4 = 2^2$
$3$ $8 = 2^3$
$4$ $16 = 2^4$
$5$ $32 = 2^5$

Thus it appears that $a_n = 2^n$.

Now we shall prove the formulae is correct.

Base Case:

For $n = 0$, $2^n = 2^0 = 1 = a_0 = a_n$.

Induction step:

Suppose formulae is correct for $n \ge 1$. Thus $a_n = 2^p$.

Thus $a_{n+1} = 1 + \sum_{i=0}^n a_i$
$= 1 + \sum_{i=0}^{n-1} a_i + a_n$
$= a_n + a_n$
Using hypothesis:
$= 2^n + 2^n$,
$= 2^{n+1}$

Thus $P(n+1)$ is true if $P(n)$ is true.

Soln20

$n$ $a_n = 1 + \frac 1 {a_{n-1} }$
$0$ $\frac 1 1$
$1$ $\frac 2 1$
$2$ $\frac 3 2$
$3$ $\frac 5 3$
$4$ $\frac 8 5$
$5$ $\frac {13} 8$

It appears from above table that formulae is $a_n = \frac {F_{n+2} } {F_{n+1} }$.

Proof:

Base Case:

For $n = 0$, it follows directly from the table.

Induction Step:

Suppose it is true for $n$. Thus $a_n = \frac {F_{n+2} } {F_{n+1} }$.

Now $a_{n+1} = 1 + \frac 1 {a_n}$
$= 1 + \frac {F_{n+1} } {F_{n+2} }$
$= \frac {F_{n+2} + F_{n+1} } {F_{n+2} }$
$= \frac {F_{n+3} } {F_{n+2} }$
$= \frac {F_{n+1+2} } {F_{n+1+1} }$.

Thus if $P(n)$ is true then $P(n+1)$ is also true.