Countably many

There are N N rational numbers R R such that

5 × 1 0 8 + R 3 + 5 × 1 0 8 R 3 \sqrt[3] {5\times 10^8 + \sqrt{R} } + \sqrt[3]{5\times 10^8- \sqrt{R} }

is an integer. What are the last three digits of N N ?

Details and assumptions

You may use the fact that 0.7937 < 0.5 3 < 0.793701 0.7937 < \sqrt[3]{0.5} < 0.793701

We choose the real value of the roots (where applicable).


The answer is 587.

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.

2 solutions

Mark Hennings
Oct 14, 2013

Assume that R 0 R \ge 0 , so that R \sqrt{R} is real. Write L = 5 × 1 0 8 L=5\times10^8 , and suppose that K = ( L + R ) 1 3 + ( L R ) 1 3 K = (L+\sqrt{R})^{\frac13} + (L-\sqrt{R})^{\frac13} is an integer. It is clear that K 0 K \ge 0 . Then K 3 = L + R + 3 K ( L 2 R ) 1 3 + L R = 2 L + 3 K ( L 2 R ) 1 3 \begin{array}{rcl} K^3 & = &L + \sqrt{R} + 3K(L^2 - R)^{\frac13} + L - \sqrt{R} \\ & = & 2L + 3K(L^2 - R)^{\frac13} \end{array} and hence ( K 3 2 L ) 3 = 27 K 3 ( L 2 R ) (K^3 - 2L)^3 \; = \; 27K^3(L^2 - R) It is clear from this equation that K 0 K \neq 0 , so K 1 K \ge 1 . Then the requirement that R 0 R \ge 0 tells us that ( K 3 2 L ) 3 27 K 3 L 2 K 3 2 L 3 L 2 3 K K 3 3 L 2 3 K 2 L 0 ( K + L 1 3 ) 2 ( K 2 L 1 3 ) 0 \begin{array}{rcl} (K^3 - 2L)^3 & \le & 27K^3L^2 \\ K^3 - 2L & \le & 3L^{\frac23}K \\ K^3 - 3L^{\frac23}K - 2L & \le & 0 \\ \big(K + L^{\frac13}\big)^2\big(K - 2L^{\frac13}\big) & \le & 0 \end{array} so we deduce that 1 K 2 L 1 3 1 \le K \le 2L^{\frac13} .

Now R = L 2 ( K 3 2 L 3 K ) 3 R \; = \; L^2 - \left(\frac{K^3 - 2L}{3K}\right)^3 and hence d R d K = 3 ( K 3 2 L 3 K ) 2 ( 2 3 K + 2 L 3 K 2 ) < 0 \frac{dR}{dK} \; = \; -3\left(\frac{K^3 - 2L}{3K}\right)^2\left(\tfrac23K + \frac{2L}{3K^2}\right) \; < \; 0 for all K > 0 K > 0 . Thus we obtain a different rational value of R R for every different positive integer value of K K . Thus we deduce that there are 2 L 1 3 \big\lfloor 2L^{\frac13} \big\rfloor possible different values of R R . For our value of L L , this means that there are 1587 1587 possible different values of R R , making the desired answer 587 587 .

Moderator note:

Excellent solution!

This expression is the solution of a monic cubic equation, with a rational constant, and no x 2 x^2 term (depressed cubic). Thus the coefficient for x x is rational, which implies that R R is rational. See Scipione for more details.

Calvin Lin Staff - 7 years, 7 months ago

Log in to reply

You still need to establish that distinct K K correspond to distinct R R , whether you make the connection between K K and R R by hand or by Scipione.

Mark Hennings - 7 years, 7 months ago

Mark H., please see my solution. I think the calculation is easier.

A Brilliant Member - 7 years, 7 months ago

We rely on our traditional method of finding roots of surds.

Let ( 5 × 1 0 8 ) + R = ( a + b ) 3 (5 \times 10^8) + \sqrt{R} = (a+\sqrt{b})^3 . Expanding RHS, & equating rational & irrational coefficients, we get:

a 3 + 3 a b = ( 5 × 1 0 8 ) b = ( 5 × 1 0 8 ) a 3 3 a a^3 + 3ab=(5 \times 10^8) \Rightarrow b= \frac{(5 \times 10^8)-a^3}{3a} .

Also, R = ( 3 a 2 + b ) b R = ( 3 a 2 + b ) 2 ( b ) = ( ( 5 × 1 0 8 ) + 8 a 3 ) 2 ( ( 5 × 1 0 8 ) a 3 ) 27 a 3 \sqrt{R}= (3a^2+b)\sqrt{b} \Rightarrow R = (3a^2+b)^2 (b) = \frac{((5 \times 10^8)+8a^3)^2((5 \times 10^8)-a^3)}{27a^3} . [on substituting for b b ]

Clearly a a must be positive, for R R to be positive. Also, a 3 < ( 5 × 1 0 8 ) a < . 5 3 × 1 0 3 < 793.701 a^3 < (5 \times 10^8) \Rightarrow a < \sqrt[3]{.5} \times 10^3 < 793.701 . For each value of a a , we get a rational R R .

The expression in the question evaluates to 2 a 2a . Thus for 2 a 2a to be integral, either a a is integral, or is rational with denominator 2 2 .

  • Case 1: a N a \in \mathbb{N} . Then a a can be any of 1 , 2 , . . . , 793 1,2,...,793 giving 793 793 choices.

  • Case 2: a = p 2 a= \frac{p}{2} (where p p is odd) Then p < 2 × 793.701 = 1587.402 1 p 1587 p < 2 \times 793.701 = 1587.402 \Rightarrow 1 \leq p \leq 1587 . There are 794 794 odd numbers from 1 1 to 1587 1587 (inclusive) & p p can be any of them.

Thus, there can be 793 + 794 = 1587 793+794 = 1587 values of a a ,& correspondingly that many values for R R .

You have assumed that the cube root of 5 × 1 0 8 + R 5\times10^8 + \sqrt{R} can be written in the particularly neat form a + b a + \sqrt{b} with 2 a 2a integral and b b rational, and found solutions for R R provided that 1 a ( 5 × 1 0 8 ) 1 3 1 \le a \le (5\times10^8)^{\frac13} .

How can you be sure that you have not missed out some values of R R for which the cube root cannot be written that way? It is a nontrivial assumption - the number 1 + 2 \sqrt{1+\sqrt{2}} is not equal to a + b 2 a + b\sqrt{2} for any rational a , b a,b .

The fact that the sum of the cube roots of 5 × 1 0 8 + R 5\times10^8 + \sqrt{R} and 5 × 1 0 8 R 5\times10^8 - \sqrt{R} is an integer forces your assumption to be valid, but this needs to be proven.

Mark Hennings - 7 years, 7 months ago

Log in to reply

Right. Since sum of the cube roots given is integral, a a is always rational. What about b b ? Either ways b b is always rational I guess,even in your quoted example ?

A Brilliant Member - 7 years, 7 months ago

Log in to reply

1 + 2 a + b c \sqrt{1 + \sqrt{2}} \neq a + b\sqrt{c} for any rational a , b , c a,b,c .

Mark Hennings - 7 years, 7 months ago

Hi Paramjit S.

I see there already a sub-thread about whether you will always have an appropriate a , b a, b rational. But I'd like to also point out that "equating rational & irrational coefficients" wouldn't work in the special case where R {R} is the square of a rational.

Peter Byers - 7 years, 7 months ago

1 obvious thing: if ( 5 × 1 0 8 ) + R = ( a + b ) 3 (5 \times 10^8) + \sqrt{R} = (a+\sqrt{b})^3 then ( 5 × 1 0 8 ) R = ( a b ) 3 (5 \times 10^8) - \sqrt{R} = (a-\sqrt{b})^3 . This is the benefit of surds. :) Mark H. has very aptly shown, using derivatives, that distinct a a give distinct R R . This might be included for a better proof.

A Brilliant Member - 7 years, 7 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...