What is the ten's digit of $\zeta=7^{7^{7^{7^7}}}$. I got this question while doing binomial theorem. I think that $7^4=2401$ and we only need $\zeta\pmod{100}$. All I could think of is already presented (thought it is nothing actually), would you help?
-
About the ten's digit, please it's counted at leftside or rightside? – Piquito Apr 25 '15 at 11:43
-
1@LuisGomezSanchez rightside i'm sure – RE60K Apr 25 '15 at 11:45
-
2@LuisGomezSanchez "Ten's digit" seems fairly unambiguous, namely the second least significant digit or $\lfloor x/10\rfloor\bmod 10$. – Mario Carneiro Apr 25 '15 at 11:46
-
Are you aware of Euler's theorem? – Karolis Juodelė Apr 25 '15 at 11:46
-
@KarolisJuodelė nope – RE60K Apr 25 '15 at 11:48
-
@ADG Hopefully my answer explains Euler's theorem and the stronger Carmichael function well enough. – user26486 Apr 25 '15 at 15:25
2 Answers
Note that $$7^2=49,\ 7^3=343,\ 7^4=2401,\ 7^5=168\color{red}{07}.$$
So, all we need is to find $7^{7^{7^{7}}}$ in mod $4$.
Now $$7^{7^{7^{7}}}\equiv (-1)^{7^{7^{7}}}\equiv -1\equiv 3\pmod 4.$$ Hence, the answer we want is the same as the ten's digit of $343$, i.e. $\color{red}{4}$.
- 139,939
-
-
Yes, there are 5 7's. But all we need to check is the upper 4 7's. Let $[a]$ be right-most two digits of $a$. You'll see the patter : $[7^1]=07,[7^2]=49,[7^3]=43,[7^4]=01,[7^5]=07,[7^6]=49$ and so on. – mathlove Apr 25 '15 at 11:52
-
1
-
1
Euler's theorem: $(a,n)=1\,\Rightarrow\, a^{\phi(n)}\equiv 1\pmod {\!n}$, where $\phi(n)$ denotes Euler's totient function, which is a function outputting the amount of positive integers less than $n$ that are coprime to $n$.
So, e.g., $\phi (10)=4$, because $1,3,7,9$ (four positive integers) are coprime to $10$ and $2,4,5,6,8$ are not.
It can be proved that (there is a unique representation $n=p_1^{\alpha_1}p_2^{\alpha_2}\cdots p_k^{\alpha_k}$ with $p_i$ prime (because of Fundamental Theorem of Arithmetic))
$$\phi(n)=\phi(p_1^{\alpha_1}p_2^{\alpha_2}\cdots p_k^{\alpha_k})=(p^\alpha_1-p^{\alpha_1-1})(p_2^{\alpha_2}-p_2^{\alpha_2-1})\cdots(p_k^{\alpha_k}-p_k^{\alpha_k-1})$$
So, e.g., $\phi(2700)=\phi(2^23^35^2)=(4-2)(27-9)(25-5)=2\cdot 18\cdot 20=720$.
$\phi(n)$ and other numbers $m$ such that $a^m\equiv 1\pmod{n}$ can be used to reduce exponents ($a^E\equiv a^{E\!\pmod {\! m}}\!\pmod {\!n}$), as will be shown later in this answer.
Your problem isn't easily approachable using Euler's theorem (using it to reduce the exponent won't be as simple, since it is too large and we can find smaller such numbers - $4$ in this example), but it is an exception. The best strategy for this kind of problems with exponentiation ($a^E\!\pmod {\!n}$) is finding the order (I'll explain), denoted $\text{ord}_n a$, of $a$ modulo $n$, or a multiple of it, and this is why finding and using $\phi(n), \lambda(n)$ (I'll introduce you to it later in the answer) is useful.
The multiplicative order, denoted $\text{ord}_n a$, of $a$ modulo $n$ is the least positive $s$ such that $a^s\equiv 1\!\pmod {\!p}$.
So, in our example, $\text{ord}_{100} 7=4$, since $7^4\equiv 2401\equiv 1\!\pmod {\! 100}$ while $7^1,7^2,7^3\not\equiv 1\!\pmod {\!100}$.
Then any $t$ such that $a^t\equiv 1\!\pmod{\!n}$ will be divisible by $s$, i.e. $t=sl$ for some $l\in\mathbb Z$. Notice that $\phi(n),\lambda(n)$ are instances of such $t$ and so they're multiples of the orders.
The reason why finding orders or multiples of them (like $\phi(n),\lambda(n)$) is important in this kind of problems is because it lets us reduce the exponent. E.g., in the case of this problem, $$7^{7^{7^{\ldots}}}\equiv 7^{7^{7^{\ldots}}\!\pmod{\!4}}\equiv 7^{(-1)^{7^{\ldots}}\!\pmod {\!4}}\equiv 7^{-1\!\pmod {\!4}}\equiv 7^3\equiv 343\!\!\!\pmod{\!100}$$
The reason why we could reduce the exponent modulo $4$ is because $7^4\equiv 1\!\pmod{\!100}$, so, e.g., $7^{13}\equiv 7^{4+4+4+1}\equiv 7^47^47^47^1\equiv 1\cdot 1\cdot 1\cdot 7\equiv 7\!\pmod {\!100}$. The reason I said "Your problem isn't easily approachable using Euler's theorem" is because $\phi(100)=40$ is too large compared to $4$. Mathlove's answer uses the same idea.
Carmichael function, denoted $\lambda (n)$, is defined as the smallest $m$ such that $(a,n)=1\,\Rightarrow\, a^m\equiv 1\!\pmod{\!n}$.
Notice how $\phi(n)$ satisfies the conditions, except for the 'smallest'. So $\lambda(n)$ is just a smaller version of $\phi(n)$ (we have $\lambda(n)\le \phi(n)$).
It can be proved that $$\lambda(n)=\begin{cases}\phi(n),\ \ \ n=2,4,p_{\text{odd}}^k\\\frac{1}{2}\phi(n),\ \ \ n=2^k,k\ge 3\\\text{lcm}(\lambda(p_1^{\alpha_1}),\lambda(p_2^{\alpha_2}),\ldots,\lambda(p_k^{\alpha_k}))\ \ \ \text{ otherwise}\end{cases}$$
So, e.g., (if $k\ge 4$)
$$\lambda(10^k)=\text{lcm}(\lambda(2^k),\lambda(5^k))=\text{lcm}(2^{k-2},4\cdot 5^{k-1})=2^{k-2}5^{k-1}$$
while $\phi(10^k)=(2^k-2^{k-1})(5^{k}-5^{k-1})=2^{k-1}\cdot 4\cdot 5^{k-1}=2^{k+1}5^{k-1}$.
Notice how for $k\ge 4$ we have $\lambda(10^k)=\frac{\phi(10^k)}{8}$.
$\lambda (n)$, compared to $\phi(n)$, makes it quite significantly easier to find the last $\ge 4$ digits of an exponentiated integer (depending on the exponent - the smaller lambda function may usually reduce the exponent more, but not necessarily).
- 11,331