The Meaning of Life

1 a + 1 b + 1 c = 1 42 \large \frac1a+\frac1b+\frac1c= \frac1{42}

Let a b c a\leq b \leq c be positive integers that satisfy the equation above. Find the maximum possible value of c c .


The answer is 3263442.

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

The smallest possible value for a a is 43 , 43, which gives us the equation

1 b + 1 c = 1 42 1 43 = 1 1806 \dfrac{1}{b} + \dfrac{1}{c} = \dfrac{1}{42} - \dfrac{1}{43} = \dfrac{1}{1806}

1806 ( b + c ) = b c ( b 1806 ) ( c 1806 ) = 180 6 2 . \Longrightarrow 1806(b + c) = bc \Longrightarrow (b - 1806)(c - 1806) = 1806^{2}.

We can then maximize c c by letting b 1806 = 1 b - 1806 = 1 and

c 1806 = 180 6 2 c = 180 6 2 + 1806 = 1806 1807 = 3263442 . c - 1806 = 1806^{2} \Longrightarrow c = 1806^{2} + 1806 = 1806*1807 = \boxed{3263442}.

What if I choose a = 54 a=54 and b = 189 b=189 . This will give

1 a + 1 b = 1 42 \frac{1}{a}+\frac{1}{b}=\frac{1}{42} ,

making c c \to \infty .

Am I wrong somewhere??

Log in to reply

Good point, although I suppose then c c would no longer be an integer. Perhaps this issue could be dealt with if the wording of the last sentence was "Find the maximum possible (finite) value of c c ?"

Brian Charlesworth - 6 years ago

\infty is not a positive integer.

Patrick Corn - 6 years ago

It is given that a,b&c are positive integers and I guess infinity is not an integer.

Harish Sinha - 6 years ago

Infinity is not any number,it is a symbol which represents largest number..and also in number system,infinity is not defined.

Akhil Bansal - 6 years ago

Pls point my mistake... I substituted 1/a as x and two others as y and z respectively...therefore I got x+y+z=1/42 Here we know by substitution rule x+y+z+2=xyz Therefore x+y+z=xyz-2 Therefore I got abc =42/85 Thus I deduced further the maximum value of c thus is 42 @Brian Charlesworth @Janardhanan Sivaramakrishnan

sarvesh dubey - 6 years ago

Log in to reply

I'm not quite sure what the "substitution rule x + y + z + 2 = x y z x + y + z + 2 = xyz " is. By necessity x y z < 1 xyz \lt 1 since each of x , y , z x,y,z is less than 1 , 1, so there is no way for this equation to hold true. Also, a , b , c a,b,c are all integers, so we can't have a b c abc being a fraction.

Brian Charlesworth - 6 years ago

I did the Exact Same way :D

@Pi Han Goh , Sorry to say, But don't you think it is overrated? After all, I dumbo like me could solve it.

Mehul Arora - 6 years ago

Log in to reply

I set it as Level 3, but it appears that many people couldn't solve it, so it rose up. Yeah, it's overrated. It's very undeserving to be in level 5.

Don't call yourself a dumbo, everyone is learning ;)

Pi Han Goh - 6 years ago

Log in to reply

Haha, I won't ;) Thanks :D

Mehul Arora - 6 years ago

Nice solution! My approach was altogether different and way lengthier, though it did lead me to the answer at last. But after reading this one, it seems that the key is to think simple. Great problem, @Pi Han Goh ! :D

Sanchit Aggarwal - 5 years, 7 months ago

sir, how did you get a=43.please explain.

manish kumar singh - 5 years, 7 months ago

Log in to reply

To maximize c c we need to minimize a . a. If a = 42 a = 42 then 1 b + 1 c = 0 , \frac{1}{b} + \frac{1}{c} = 0, which has no solutions for positive integers b , c . b,c. If a < 42 a \lt 42 then 1 b + 1 c < 0 , \frac{1}{b} + \frac{1}{c} \lt 0, which again has no positive integer solutions. Thus the least possible value of a a is 43 , 43, which leads to the maximized value of c c as found in my solution.

Brian Charlesworth - 5 years, 7 months ago

Log in to reply

thank you sir, got it.

manish kumar singh - 5 years, 7 months ago

If the question is: "What is the smallest possible value of a+b+c?". Can you help me the solution please.

Nguyen Nam - 5 years ago

Interesting - I understand that the given solution maximizes c c for the fixed value a = 43 a=43 on the border, i.e. we have found a local maximum on the border. However, as far as I can tell we have not shown it is also a global maximum!

I'll do that below to demonstrate the difference.

Carsten Meyer - 2 months, 2 weeks ago

Thanks. Now I learnt another angle of looking at the problem. I was harping on the fact 1 4 + 1 4 + 1 2 = 1 A n d t h e n 1 42 2 x 42 t o g i v e r e c i p r o c a l o f a n d i n t e g e r . I got x=6, 8, 14, 16..I did not understand why I got these values. I a l w a y s a s s u m i m g b = c ! ! \dfrac 1 4 +\dfrac 1 4 +\dfrac 1 2 =1 \\ And~ then~\dfrac 1 {42}-\dfrac 2 {x*42} ~to~ give ~reciprocal~of~ and ~integer.~~\\\text{ I got x=6, 8, 14, 16..I did not understand why I got these values.}\\I ~always~ assumimg ~~~~~b=c!!

Niranjan Khanderia - 6 years ago

Log in to reply

You're welcome. In general, for the equation

1 a + 1 b + 1 c = 1 n \dfrac{1}{a} + \dfrac{1}{b} + \dfrac{1}{c} = \dfrac{1}{n} for integers 0 < n < a b c , 0 \lt n \lt a \le b \le c,

the maximum possible value for c c is ( n 2 + n ) ( n 2 + n + 1 ) . (n^{2} + n)(n^{2} + n + 1).

Brian Charlesworth - 6 years ago

Log in to reply

Thanks a lot.. Where on web can we study such problems ?

Niranjan Khanderia - 6 years ago

Log in to reply

@Niranjan Khanderia Well, sums of unit fractions are referred to as Egyptian fractions , which have been studied extensively. The representation of 1 1 as the sum of unit fractions has also garnered a fair bit of analysis, such as in this paper .

Brian Charlesworth - 6 years ago

Log in to reply

@Brian Charlesworth Thank you. I will go through the links. Incidentally,
1 a + 1 b + 1 c = 1 n \dfrac{1}{a} + \dfrac{1}{b} + \dfrac{1}{c} = \dfrac{1}{n} for integers 0 < n < a b c , 0 \lt n \lt a \le b \le c,

the maximum possible value for c c is ( n 2 + n ) ( n 2 + n + 1 ) = n ( n + 1 ) ( n 2 + n + 1 ) . = n 4 + 2 n 3 + 2 n 2 + n p l e a s i n g t o l o o k a t ! (n^{2} + n)(n^{2} + n + 1)=\color{#D61F06}{n(n+1)(n^2+n+1)}.=\color{#3D99F6}{n^4+2n^3+2n^2+n}\\pleasing ~to ~look~ at!

Niranjan Khanderia - 6 years ago

Actually a great observation sir !.

Gaurav Jain - 6 years ago

i can't understand
please elaborate this theorem or give me related links where i can find this broadly.

Nabil Rafiq - 6 years ago

Log in to reply

@Nabil Rafiq The smallest possible value for a a is n + 1 , n + 1, which gives us

1 b + 1 c = 1 n 1 n + 1 = 1 n 2 + n \dfrac{1}{b} + \dfrac{1}{c} = \dfrac{1}{n} - \dfrac{1}{n + 1} = \dfrac{1}{n^{2} + n}

( n 2 + n ) ( b + c ) = b c \Longrightarrow (n^{2} + n)(b + c) = bc

b c ( n 2 + n ) b ( n 2 + n ) c + ( n 2 + n ) 2 = ( n 2 + n ) 2 \Longrightarrow bc - (n^{2} + n)b - (n^{2} + n)c + (n^{2} + n)^{2} = (n^{2} + n)^{2}

( b ( n 2 + n ) ) ( c ( n 2 + n ) ) = ( n 2 + n ) 2 . \Longrightarrow (b - (n^{2} + n))(c - (n^{2} + n)) = (n^{2} + n)^{2}.

We can then maximize c c by letting

c ( n 2 + n ) = ( n 2 + n ) 2 c - (n^{2} + n) = (n^{2} + n)^{2}

c = ( n 2 + n ) 2 + ( n 2 + n ) = ( n 2 + n ) ( n 2 + n + 1 ) . \Longrightarrow c = (n^{2} + n)^{2} + (n^{2} + n) = (n^{2} + n)(n^{2} + n + 1).

Brian Charlesworth - 6 years ago

Log in to reply

@Brian Charlesworth Yeah , please post it as a note! Thanks!

Nihar Mahajan - 6 years ago

@Brian Charlesworth Can you please post this as a note and wiki ? It will be very helpful. Thank you.

Niranjan Khanderia - 6 years ago
Lew Sterling Jr
Jun 11, 2015

Carsten Meyer
Mar 31, 2021

Constraints on a , b a,\:b

By definition we have 1 a b c 1\leq a\leq b\leq c . Let's find some additional constraints on a a : 0 < 1 b + 1 c = 1 42 1 a a 43 , 1 42 = 1 a + 1 b + 1 c 3 a a 3 42 = 126 \begin{aligned} 0&< \frac{1}{b}+\frac{1}{c}=\frac{1}{42}-\frac{1}{a} &\Rightarrow &&a&\geq 43,&&&&& \frac{1}{42} &=\frac{1}{a} +\frac{1}{b} +\frac{1}{c} \leq \frac{3}{a} &\Rightarrow && a\leq 3\cdot 42=126 \end{aligned} We now know a { 43 ; ; 126 } a\in\{43;\:\ldots;\:126\} . Lets find a lower bound for b b : 0 < 42 a b c = 42 a b ( 1 42 1 a 1 b ) = a b 42 ( a + b ) = ( a 42 ) ( b 42 ) 4 2 2 1 + 4 2 2 ( a 42 ) ( b 42 ) b 42 + 1765 a 42 \begin{aligned} &&0&<\frac{42ab}{c}=42ab\left(\frac{1}{42}-\frac{1}{a}-\frac{1}{b}\right)=ab-42(a+b)=(a-42)(b-42)-42^2 \\\\ \Rightarrow && 1+42^2 &\leq (a-42)(b-42) \quad\Rightarrow \quad b \geq 42+\frac{1765}{a-42} \end{aligned} Together with b a b\geq a we have two lower bounds for b b we may combine into l ( a ) l(a) and get l ( a ) : = max { a ; 42 + 1765 a 42 } l ( a ) b c \begin{aligned} l(a)&:=\max\left\{a;\:42+\frac{1765}{a-42}\right\} &\Rightarrow &&l(a) &\leq b\leq c \end{aligned} Let's summarize our results into the set Ω \Omega of remaining values for a , b a,\:b : Ω : = { ( a , b ) N 2 43 a 126 , l ( a ) b } and 1 c = 1 42 1 a 1 b = : 1 f ( a , b ) \begin{aligned} \Omega&:=\left\{(a,\:b)\in\mathbb{N}^2|\:43\leq a\leq 126, \quad l(a)\leq b\right\} &&&\text{and}&&\frac{1}{c}=\frac{1}{42}-\frac{1}{a}-\frac{1}{b}=:\frac{1}{f(a,\:b)} \end{aligned} However, it is really difficult to solve this problem over the discrete set. Instead, we will consider a larger set Ω \Omega' with the same borders but a , b , c R a,\:b,\:c\in\mathbb{R} .


Global Maximum for c c

We now want to maximize c = f ( a , b ) c=f(a,\:b) on Ω \Omega' . For points within Ω \Omega' we notice f ( a , b ) > f ( a , b ) for b < b f ( a , b ) > f ( a , b ) for a < a \begin{aligned} \bullet &&f(a,\:b) &>f(a,\:b') &\text{for}&& b&<b'\\ \bullet &&f(a,\:b)&>f(a',\:b) &\text{for}&& a&< a' \end{aligned} i.e. f f strictly increases whenever we move down or to the left within Ω \Omega' . Our maximum (if it exists) must lie on the lower left border δ Ω \delta\Omega defined by δ Ω : = { ( 43 , b ) R 2 b l ( 43 ) } { ( a , l ( a ) ) R 2 a [ 43 ; 84 ] } \begin{aligned} \delta\Omega &:= \left\{ (43,\:b)\in\mathbb{R}^2|\: b\geq l(43) \right\}\cup\left\{ (a,\:l(a))\in\mathbb{R}^2|\:a \in[43;\:84] \right\} \end{aligned} The first part of the border is easy to check - f ( 43 , b ) f(43,\:b) increases as we move down and our only candidate is ( 43 , l ( 43 ) ) (43,\:l(43)) . It's also possible to show for the second part that f ( a , l ( a ) ) f(a,\:l(a)) increases as we move to the left so we get the global maximum of f f on Ω \Omega' at a = 43 , b = l ( 43 ) = 42 + 1765 = 1807 c = 1 f ( a , b ) = 3263442 \begin{aligned} a&=43, & b&=l(43)=42+1765=1807 &&&\Rightarrow&&&& c&=\frac{1}{f(a,\:b)}=\boxed{3263442} \end{aligned} The solution is integer-valued and also satisfies b c b\leq c , so it is indeed a solution to the problem.

Wow, you're right. All the other solutions implicitly assumed that c c is maximized when a = 42 , a=42, but that wasn't proven at all. Thank you for this explanation

Pi Han Goh - 2 months, 2 weeks ago

Thank you for your reply - I really hope the proof is correct so far. I did leave out some details like why the left border begins at x = 84 x=84 and the monotony on the second border part, but as far as I can tell there is no way around all that :(

Perhaps a nice little sketch of Ω \Omega' is in order - I came up with the proof only after the sketch, so it might really help to follow this (rather lengthy) elaboration.

Carsten Meyer - 2 months, 2 weeks ago

Log in to reply

Yes, I'm still reading through your solution. If I recall correctly, I also made the same assumption as everyone else.

Pi Han Goh - 2 months, 2 weeks ago
Pop Wong
Jun 7, 2020

The smallest possible value for a is 43

1 b \frac{1}{b} + 1 c \frac{1}{c} = 1 42 \frac{1}{42} - 1 43 \frac{1}{43} = 1 42 43 \frac{1}{42*43}

The smallest possible value for b will be 42*43+1

1 c \frac{1}{c} = 1 ( 42 43 ) ( 42 43 + 1 ) \frac{1}{(42*43)*(42*43+1)}

Therefore, the maximum c = c = ( 42 43 ) ( 42 43 + 1 ) (42*43)*(42*43+1) = 3263442 \boxed{3263442}

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...