Don't forget to carry

If the following infinite series S S is evaluated as a decimal,what is the 37th digit to the right of the decimal place?

S = 1 9 + 1 99 + 1 999 + + 1 10 n 1 + \large S=\frac { 1 }{ 9 } +\frac { 1 }{ 99 } +\frac { 1 }{ 999 } +\ldots + \frac { 1 }{ { 10 }^{ n }-1 } + \ldots


The answer is 2.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

5 solutions

Thaddeus Abiy
Apr 6, 2014

S = 1 9 + 1 99 + 1 999 + . . 1 10 n 1 = S=\frac { 1 }{ 9 } +\frac { 1 }{ 99 } +\frac { 1 }{ 999 } +..\frac { 1 }{ { 10 }^{ n }-1 }=

0.11111111111111... + 0.11111111111111...+

0.01010101010101... + 0.01010101010101...+

0.00100100100100... 0.00100100100100...

First,it is clear that the k t h kth column is composed only of 1 1 's and 0 0 's. The 1 1 's in column k k occur in the r r th row only if r r is a divisor of k k .Below the k t h kth row in column k k there are only 0 0 's. Since 37 37 is a prime,its only divisors are 1 1 and 37 37 and so the only 1 1 's in column 37 37 occur in rows 1 1 and 37 37 . Thus we may be tempted to jump to the conclusion that the required digit is 2 2 .

However since sums are preformed from right to left we must consider the possibility of a carry over from the 38 38 th place.Since 38 38 has four divisors 1 , 2 , 19 1,2,19 and 38 38 there are four 1 1 's in column 38 38 . Of course,there is a chance that there is a carry over from the 39 t h 39th place as well. However ,since the 38 38 th place holds only four 1 1 's,it would take a carry of at least 6 6 from the 39 39 th place. We can show that the carry from the 39 39 th place doesn't amount to 5 5 . Thus we conclude there is no carry over and that the answer is indeed 2 2 .

Nice. If the question did have a carry-over, that would have been more interesting.

Calvin Lin Staff - 7 years, 2 months ago

Log in to reply

Hi Calvin! Are you guys planning on making trignometry a seperate topic? It would be helpful.

A Former Brilliant Member - 7 years, 2 months ago

I guess the title is a bit of a misnomer..

Thaddeus Abiy - 7 years, 2 months ago

Why does this question have a rating, but this one doesn't have one?

Beakal Tiliksew - 7 years, 2 months ago

Log in to reply

@Thaddeus didn't set a level for this, and as such this problem is unrated for a much longer period.

Calvin Lin Staff - 7 years, 2 months ago

  • And notice, the nth digit is (the number of factors in n) (mod 10).

David Lee - 7 years, 2 months ago

Log in to reply

unless the n+1th digit has more than 9 factors, or the n +ith digit has more than 10^i factors for any i. So, for example, though 359 is prime, the 359th decimal place is not 2 but 4 because 360 has 24 factors

Michael Gilbart-Smith - 3 years, 10 months ago

My method was a little different:in 1/9 every digit is 1 and as 37 is a prime while adding the number 1 will occur once in 1/9 and second when 1 is divided by 37 nines.thus we get 2 simple

Adarsh Kumar - 7 years, 2 months ago

Log in to reply

Same here!

David Lee - 7 years, 2 months ago

Which diriclet theorem did you refer in this solution?

Eddie The Head - 7 years, 2 months ago

Log in to reply

My bad,I mixed it up with another problem.Dirichlet theorem on arithmetic progressions has nothing to do with this

Thaddeus Abiy - 7 years, 2 months ago

brilliant, absolutely brilliant

Reinaldo Limantara - 6 years, 10 months ago

Why is it true? “The 1’s in column k occur in the r th row only if r is a divisor of k “ Please explain.

Ravi Desai - 3 years, 8 months ago
Ivan Martinez
May 4, 2014

S= 9^-1 + 99^-1....
S/11= 99^-11+999^-1........
S-S/10=9^-1
then S=11/90 = 1.222222......


Moderator note:

As pointed out in the comments, this solution is incorrect. Can you spot the mistake?

Note that you do not have a geometric progression. Your answer is invalid.

Calvin Lin Staff - 4 years, 8 months ago

totally unreadable. at least use different lines

Jean-Denis Muys - 4 years, 7 months ago

99 * 11 = 1089

Sarthak Upadhyay - 3 years, 5 months ago

Your error is as sarthak said below- actually the decimal value of S if you begin adding them starts looking like: s=.01223242435262445

Donal Stones - 3 years, 3 months ago

I CAN'T do this

Kain Hakim - 2 years, 5 months ago

This is to hard

Jean-Pierre Goupy - 1 week, 5 days ago
Carsten Meyer
Sep 9, 2019

Rewrite the series

First rewrite the series into a more accessible form: S = k = 1 1 1 0 k 1 = k = 1 1 0 k 1 1 0 k = k = 1 l = 1 ( 1 0 k ) l = k = 1 l = 1 1 0 k l geometric series, 1 0 k 1 10 < 1 \begin{aligned} S&=\sum_{k=1}^\infty\frac{1}{10^k-1}=\sum_{k=1}^\infty \frac{10^{-k}}{1-10^{-k}}=\sum_{k=1}^\infty\sum_{l=1}^\infty (10^{-k})^l=\sum_{k=1}^\infty\sum_{l=1}^\infty10^{-kl}&\left|\text{geometric series, }|10^{-k}|\leq\frac{1}{10}<1\right. \end{aligned}

The series converges absolutely for all k , l 1 k,\:l\geq 1 - we may reorder the series by the exponent k l = ! d N kl\overset{!}{=}d\in\mathbb{N} : S = k = 1 l = 1 1 0 k l = d = 1 ( k , l ) Ω d 1 0 n k Ω d : = { ( k , l ) N 2 n k = d } \begin{aligned} S&=\sum_{k=1}^\infty\sum_{l=1}^\infty10^{-kl}=\sum_{d=1}^\infty \sum_{(k,\:l)\in\Omega_d}10^{-nk}&\left|\Omega_d:=\{(k,\:l)\in\mathbb{N}^2|\quad nk=d\}\right.\\ \end{aligned}

Notice that the inner sum is constant because the exponent equals d d ? In other words, the inner sum counts how many pairs exist such that k l = d kl=d - that is the number of positive factors of d d : S = d = 1 c d 1 0 d , c d = Ω d = "number of positive factors of d " 2 d 1 \begin{aligned} S&= \sum_{d=1}^\infty c_d 10^{-d},&c_d=|\Omega_d|=\text{"number of positive factors of }d\text{"}\leq 2\sqrt{d}-1 \end{aligned}

How many positive factors can d = k l d=kl have? If it is not a perfect square, one factor has to be smaller than d \sqrt{d} and the other has to be greater than d \sqrt{d} - in total we have at most 2 ( d 1 ) 2(\sqrt{d}-1) factors. If d d is a perfect square, we have one more factor d \sqrt{d} !


Converge back to the task

Let's find the 3 7 t h 37^{th} digit of S S ! We may be tempted to say "that's c 37 c_{37} , easy", but that might not be right: The coefficients c d c_d are not digits and can be greater than 10. In other words, later coefficients might influence the 3 7 t h 37^{th} digit as well! To make sure we get a correct estimate for the digit, we need to know how many terms we need to consider or how fast the series converges: c d 1 0 d < 2 d 1 0 d = : a d , a d + 1 a d = 2 d + 1 1 0 ( d + 1 ) 2 d 1 0 d = 1 + 1 d 1 10 2 10 = : q < 1 c d 1 0 d < a d a N q d N , d N ( ) \begin{aligned} |c_d10^{-d}|&<2\sqrt{d}\cdot10^{-d}=:a_d,&&&\frac{a_{d+1}}{a_d}&=\frac{ 2\sqrt{d+1}\cdot 10^{-(d+1)} }{ 2\sqrt{d}\cdot 10^{-d} }=\sqrt{1+\frac{1}{d}}\cdot\frac{1}{10}\leq\frac{\sqrt{2}}{10}=:q<1\\[1em] \Rightarrow\quad |c_d10^{-d}|&<a_d\leq a_Nq^{d-N},&&&d&\geq N&(*) \end{aligned}

Let's find out which digits the later coefficients of the series can influence: d = N c d 1 0 d < ( ) d = N a N q d N = a N d = 0 q d = q < 1 a N 1 q = 2 1 0.1 2 N 1 0 N 2.4 N 1 0 N d = 39 c d 1 0 d 2.4 39 1 0 39 < 1.5 1 0 38 \begin{aligned} \sum_{d=N}^\infty |c_d10^{-d}|&\underset{(*)}{<}\sum_{d=N}^\infty a_Nq^{d-N}=a_N\sum_{d=0}^\infty q^d\underset{|q|<1}{=}\frac{a_N}{1-q}=\frac{2}{1-0.1\sqrt{2}}\cdot \sqrt{N}\cdot10^{-N}\leq 2.4\cdot \sqrt{N}\cdot10^{-N}\\ \Rightarrow\quad \sum_{d=39}^\infty |c_d10^{-d}|&\leq 2.4\cdot\sqrt{39}\cdot10^{-39}<1.5\cdot 10^{-38} \end{aligned}

We only need to consider two coefficients - c 37 = 2 c_{37}=2 and c 38 = 4 c_{38}=4 . Any following coefficients can at most flip the 3 8 t h 38^{th} digit to 5 5 and can be ignored - the 3 7 t h 37^{th} digit of S S is indeed 2 \boxed{2} !


Rem.: For a challenge with overflow, prove that the 4 7 t h 47^{th} digit of S S is 3 c 47 3\neq c_{47} !

James Lee
Aug 24, 2017

37 can be divided by 1 and 37 so the first 1/9 and the eventual 1/10^37-1 will add a 1 in the 37th digit giving it a 2.

Ariel Carvajal
Oct 13, 2018

Here you can easily notice:

0.1111111111 1111111111 1111111111 1111111
0.0101010101 0101010101 0101010101 0101010
0.0010010010 0100100100 1001001001 0010010
0.0001000100 0100010001 0001000100 0100010
0.0000100001 0000100001 0000100001 0000100
0.0000010000 0100000100 0001000001 0000010
0.0000001000 0001000000 1000000100 0000100
0.0000000100 0000010000 0001000000 0100000
0.0000000010 0000000100 0000001000 0000010
0.0000000001 0000000001 0000000001 0000000
0.0000000000 1000000000 0100000000 0010000
0.0000000000 0100000000 0001000000 0000010
0.0000000000 0010000000 0000010000 0000000
0.0000000000 0001000000 0000000100 0000000
0.0000000000 0000100000 0000000001 0000000
0.0000000000 0000010000 0000000000 0100000
0.0000000000 0000001000 0000000000 0001000
0.0000000000 0000000100 0000000000 0000010
0.0000000000 0000000010 0000000000 0000000
0.0000000000 0000000001 0000000000 0000000
0.0000000000 0000000000 1000000000 0000000
0.0000000000 0000000000 0100000000 0000000
0.0000000000 0000000000 0010000000 0000000
0.0000000000 0000000000 0001000000 0000000
0.0000000000 0000000000 0000100000 0000000
0.0000000000 0000000000 0000010000 0000000
0.0000000000 0000000000 0000001000 0000000
0.0000000000 0000000000 0000000100 0000000
0.0000000000 0000000000 0000000010 0000000
0.0000000000 0000000000 0000000001 0000000
0.0000000000 0000000000 0000000000 1000000
0.0000000000 0000000000 0000000000 0100000
0.0000000000 0000000000 0000000000 0010000
0.0000000000 0000000000 0000000000 0001000
0.0000000000 0000000000 0000000000 0000100
0.0000000000 0000000000 0000000000 0000010
0.0000000000 0000000000 0000000000 0000001

Python

for i in range(1, 38):
    print("0.", end="")

    for k in range(1, 38):
        if(k % i == 0):
            print("1", end="")
        else:
            print("0", end="")

    print("")

Ariel Carvajal - 2 years, 8 months ago

Using the following code on arbitrary precision calculator:

scale=42; s=0; for (i=9;i<10^42;i=i*10+9) {s+=1/i}; print s

We get: 0.122324243426244526264428344628264449 2 44828

Therefor, the digit is 2.

Big Chiroptera - 2 years, 3 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...