Hugely Harmonic

Let N N be the smallest positive integer for which 1 1 + 1 2 + 1 3 + + 1 N > 100. \frac{1}{1}+ \frac{1}{2} + \frac{1}{3} + \cdots + \frac{1}{N} > 100. How many digits does N N have?


Note: log 10 e 0.434. \log_{10}e \approx 0.434.


The answer is 44.

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.

4 solutions

Chew-Seong Cheong
Jul 26, 2017

Relevant wiki: Harmonic Number

k = 1 N 1 k > 100 H N > 100 where H n is the n th harmonic number. γ + ln N > 100 Approximate from γ = lim n ( H n ln n ) ln N > 100 γ where γ is Euler-Mascheroni constant. log 10 N > ( 100 γ ) log 10 e > ( 100 0.5772 ) log 10 e > 43.179 \begin{aligned} \sum_{k=1}^N \frac 1k & > 100 \\ \implies \color{#3D99F6} H_N & > 100 & \small \color{#3D99F6} \text{where }H_n \text{ is the }n\text{th harmonic number.} \\ \color{#3D99F6} \gamma + \ln N & > 100 & \small \color{#3D99F6} \text{Approximate from }\gamma = \lim_{n \to \infty} (H_n - \ln n) \\ \ln N & > 100 - \gamma & \small \color{#3D99F6} \text{where }\gamma \text{ is Euler-Mascheroni constant.} \\ \log_{10} N & > (100 - \gamma)\log_{10} e \\ & > (100 - 0.5772)\log_{10} e \\ & > 43.179 \end{aligned}

The number of digits of N N , n = log 10 N + 1 = 43 + 1 = 44 n = \left \lfloor \log_{10} N \right \rfloor + 1 = 43+1 = \boxed{44} .

You spelt the constant wrongly.

To move from step 2 to step 3, you need to prove that H n > ln n H_n > \ln n is true for all positive integers n n . Do you know how to do it?

Pi Han Goh - 3 years, 10 months ago

Log in to reply

Considering k k + 1 1 x d x > k k + 1 1 x d x \int_k^{k+1} \frac 1{\lfloor x \rfloor} \ dx > \int_k^{k+1} \frac 1x \ dx .

Chew-Seong Cheong - 3 years, 10 months ago

Log in to reply

So you have shown that H(n) > ln(n), but how do you know that H(n) > gamma + ln(n) is true as well?

Pi Han Goh - 3 years, 10 months ago

Log in to reply

@Pi Han Goh @Chew-Seong Cheong Pi Han Goh is correct. This estimate may have coincidentally yielded the correct number of digits, but it does not provide a valid proof as to why.

Daniel Juncos - 3 years, 10 months ago

@Pi Han Goh

No vigorous proof but the above diagram, the blue area converges to γ \gamma as n n \to \infty . This means that H n ln n < γ 0.5772 < 1 H_n - \ln n < \gamma \approx 0.5772 < 1 , therefore, the approximation is accurate.

Chew-Seong Cheong - 3 years, 10 months ago

Log in to reply

@Chew-Seong Cheong Your logic is not sound. If what you said is true, then H n < γ + ln n H_n < \gamma + \ln n is true, thus you cannot move from your 2nd inequality to your 3rd inequality.

Pi Han Goh - 3 years, 10 months ago

Log in to reply

@Pi Han Goh γ + ln n > H n > 100 \gamma + \ln n > H_n > 100 .

Chew-Seong Cheong - 3 years, 10 months ago

Log in to reply

@Chew-Seong Cheong This is obviously wrong. If you set n = 1 n = 1 , you get γ > H 1 > 100 \gamma > H_1 > 100 .

Pi Han Goh - 3 years, 10 months ago

Log in to reply

@Pi Han Goh But we are talking about the smallest H N H_{\color{#D61F06}N} larger than 100.

Chew-Seong Cheong - 3 years, 10 months ago

Log in to reply

@Chew-Seong Cheong This is a circular argument.

This goes back to the question of "How do you know that γ + ln n > H n \gamma + \ln n > H_n if you only know that H n > 100 H_n > 100 is true?"

Pi Han Goh - 3 years, 10 months ago

Log in to reply

@Pi Han Goh The difference H n ln n H_n - \ln n reduces asymptotically from 1 when n = 1 n=1 to γ \gamma when n n \to \infty . H n > ln n + γ \implies H_n > \ln n + \gamma .

Chew-Seong Cheong - 3 years, 10 months ago

Log in to reply

@Chew-Seong Cheong That's another circular argument. How do you know that H n ln n > γ H_n - \ln n > \gamma for sufficiently large n n ?

Pi Han Goh - 3 years, 10 months ago

Log in to reply

@Pi Han Goh I don't know.

Chew-Seong Cheong - 3 years, 10 months ago
Zach Abueg
Jul 25, 2017

For any decreasing function f f \ :

k = a N f ( k ) x = a N + 1 f ( x ) d x \displaystyle \sum_{k \ = \ a}^{N} f(k) \ \geq \ \int_{x \ = \ a}^{N + 1} f(x) \ dx

For f ( x ) = 1 x f(x) = \dfrac 1x , we have

k = 1 N 1 k 1 N + 1 1 x d x k = 1 N 1 k ln ( N + 1 ) \displaystyle \sum_{k \ = \ 1}^{N} \frac 1k \ \geq \ \int_{1}^{N + 1} \frac 1x \ dx \\ \sum_{k \ = \ 1}^{N} \frac 1k \ \geq \ \ln \left(N + 1\right)

Because the growth of the harmonic series is so slow, we can say that k = 1 N + 1 1 k 101 \displaystyle \sum_{k \ = \ 1}^{N + 1} \frac 1k \ \cancel{\geq} \ 101 . Thus,

k = 1 N 1 k = 100 100 k = 1 N 1 k < 101 \displaystyle \Bigg\lfloor{\sum_{k \ = \ 1}^{N} \frac 1k\Bigg\rfloor} = 100 \\ \implies 100 \ \leq \ \sum_{k \ = \ 1}^{N} \frac 1k \ < \ 101

For k = 1 N 1 k = 100 \displaystyle \sum_{k \ = \ 1}^{N} \frac 1k = 100 , we have

ln ( N + 1 ) > 100 N > e 100 1 2.7 × 1 0 43 \displaystyle \begin{aligned} \ln\left(N + 1\right) & > 100 \\ \implies N & > e^{100} - 1 \ \approx \ 2.7 \times 10^{43} \end{aligned}

In the extreme case that k = 1 N 1 k = 101 \displaystyle \sum_{k \ = \ 1}^{N} \frac 1k = 101 , we have

ln ( N + 1 ) < 101 N < e 101 1 7.3 × 1 0 43 \displaystyle \begin{aligned} \ln\left(N + 1\right) & < 101 \\ \implies N & < e^{101} - 1 \ \approx \ 7.3 \times 10^{43}\end{aligned}

Thus, 2.7 × 1 0 43 < N < 7.3 × 1 0 43 2.7 \times 10^{43} < N < 7.3 \times 10^{43} , the numbers of which all have 43 + 1 = 44 43 + 1 = \boxed{44} digits.

You have the right ideas, but it could be much better expressed, especially at the end.

Essentially, what you want is a n + 1 f ( x ) d x < k = a n f ( k ) < a 1 n f ( x ) d x \int_{a}^{n+1} f(x) \, dx < \sum_{k=a}^n f(k) < \int_{a-1}^n f(x) \, dx .
Note in particular, that the ln ( N + 1 ) < 101 \ln (N+1) < 101 inequality isn't exactly correct.

Calvin Lin Staff - 3 years, 10 months ago

Log in to reply

I see. Please, guide me in the right direction.

Zach Abueg - 3 years, 10 months ago
Kyle T
Aug 4, 2017

The exact number is given here
as 15092688622113788323693563264538101449859497
which is 44 digits in length

wow! that's a big #!!!!!!!!

A Former Brilliant Member - 3 years, 10 months ago

the beauty of maths lies in finding the number of digits without having to find the actual number.....anyway you listed the number by using a computer or calculator...didnt you...?🤣

Ben John Toms - 3 years, 6 months ago

The sum is about 100.0000000000000000000000000000000000000000000090043212089...

Joshua Powles - 3 years, 10 months ago
Aneesh Saripalli
Aug 3, 2017

We can rexpress the sum as:

x = 0 n 1 x 1 N 1 x d x \sum_{x=0}^n\frac{1}{x}\approx\int_{1}^{N}\frac{1}{x}dx

Evalutating the integral:

1 N 1 x d x = ln x 1 N \int_{1}^{N}\frac{1}{x}dx =\ln{|x|} \big|_1^N

ln x 1 N = ln N ln 1 \ln{|x|} \big|_1^N=\ln{N}-\ln{1}

ln N ln 1 = ln N 0 = ln N \ln{N}-\ln{1}=\ln{N} - 0 = \ln{N}

We know that this value must be greater than 100, so

ln N > 100 \ln{N} > 100

Consequently:

N = e 100 N=e^{100}

The next step requires a bit of intuition:

We're told that log 10 e . 434 \log_{10}{e}\approx.434

Furthermore, we know that 1 digit requires 1 0 1 10^1 power. The number of power of 10's in a number is the log 1 0 \log_10 of that number.

We can think of this like e having .434 digits. Hence e 100 e^{100} would have log 10 e 100 \log_{10}{e^{100}} digits.

log 10 e 100 = 100 log 10 e = 100 . 434 = 43.4 \log_{10}{e^{100}}=100\log_{10}{e}=100*.434=43.4

Because numbers can only have a whole number of digits, and N has at least 43.4 digits, it must have 44 digits.

You make the argument that it must have 43.4 \geq 43.4 digits, so it must have 44 44 . How do we know it doesn't have 45 \geq 45 digits?

Zach Abueg - 3 years, 10 months ago

Quite elegantly done!

Tanay Rathore - 3 years, 10 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...