MHB Can You Prove This Fraction Sequence is Less Than 1/1000?

AI Thread Summary
The discussion centers on proving that the product of the fractions from 1/2 to 999999/1000000 is less than 1/1000. Participants share various methods, including induction, to approach the problem. The effectiveness of these methods is acknowledged, with specific praise for the induction technique. References to related threads and solutions are made to enhance understanding. The conversation emphasizes collaboration in finding a mathematical proof.
anemone
Gold Member
MHB
POTW Director
Messages
3,851
Reaction score
115
Show that $\dfrac{1}{2}\cdot\dfrac{3}{4}\cdot\dfrac{5}{6} \cdots\dfrac{999999}{1000000}<\dfrac{1}{1000}$
 
Mathematics news on Phys.org
We claim that

\frac{1 \cdot 3 \cdot 5 \cdots (2n-1)}{2 \cdot 4 \cdot 6\cdots (2n)} \leq \frac{1}{\sqrt{3n+1}}

for all positive integers n. The result then follows by setting n = 500000 and observing that

\frac{1}{\sqrt{1500001}} &lt; \frac{1}{1000}

The claim is proved by induction on n.
For n = 1, the claim is obvious.
Assume the claim is true for n. We have to show that

\frac{1 \cdot 3 \cdot 5 \cdots (2n-1)(2n+1)}{2 \cdot 4 \cdot 6\cdots (2n)(2n+2)} \leq \frac{1}{\sqrt{3n+4}}

But

\frac{1 \cdot 3 \cdot 5 \cdots (2n-1)(2n+1)}{2 \cdot 4 \cdot 6\cdots(2n)(2n+2)}=\frac{1 \cdot 3 \cdot 5 \cdots (2n-1)}{2 \cdot 4 \cdot 6\cdots (2n)} \frac{2n+1}{2n+2}\leq\frac{1}{\sqrt{3n+1}}\frac{2n+1}{2n+2}

So we have to show that

\frac{1}{\sqrt{3n+1}}\frac{2n+1}{2n+2}\leq \frac{1}{\sqrt{3n+4}}

This inequality follows by clearing fractions, squaring both sides and simplifying. The result is

19n \leq 20n

which holds for all positive n. This completes the proof of the claim.
 
As reported in...

http://mathhelpboards.com/questions-other-sites-52/unsolved-analysis-number-theory-other-sites-7479-4.html#post40097

... the the explicit expression of the sequence is...

$\displaystyle a_{n} = \prod_{k=1}^{n} (1 - \frac{1}{2\ k})\ (1)$

... and it is the solution of the difference equation...$\displaystyle a_{n+1} = a_{n}\ (1 - \frac{1}{2\ n}),\ a_{1}=1\ (2)$

The (2) is related to the ODE...

$\displaystyle y^{\ '} = - \frac{y}{2\ x}\ (3)$

... the solution of which is $\displaystyle y = \frac{c}{\sqrt{x}}$, so that we can suppose $\displaystyle a_{n} \sim r_{n}= \frac{c}{\sqrt{n}}$. If we suppose that $r_{500000}= \frac{1}{1000}$ then is $\displaystyle c = \frac{1}{\sqrt{2}}$. In the following table the first values os $a_{n}$ and $r_{n}$ are reported... $a_{1}= 1,\ r_{1} = .70710678...$

$a_{2}= .5,\ r_{2} = .5$

$a_{3}= .375,\ r_{3} = .40824829...$

$a_{4}= .3125,\ r_{4} = .35355339...$

$a_{5}= .273438...,\ r_{5} = .31627766...$

$a_{6}= .246094...,\ r_{6} = .2886751...$

$a_{7}= .225586...,\ r_{7} = .2672612...$

$a_{8}= .209473...,\ r_{8} = .25$

$a_{9}= .196381...,\ r_{9} = .2357022...$

$a_{10}= .185471...,\ r_{10} = .2236067...$

It is clear from the table that for n 'large enough' the relative increments of the $a_{n}$ and $r_{n}$ are pratically the same and that is verified considering that is... $\displaystyle \frac{a_{n+1}}{a_{n}} = 1 - \frac{1}{2\ n}$ $\displaystyle \frac{r_{n+1}}{r_{n}} = \sqrt{1 - \frac{1}{n}} = 1 - \frac{1}{2\ n} - \frac{1}{8\ n^{2}} - ...\ (4)$... so that we can conclude that is $\displaystyle a_{500000} < r_{500000} = \frac{1}{1000}$...

Kind regards

$\chi$ $\sigma$
 
Thanks for participating to both of you, Petek and chisigma! Your induction method looks nice and great, Petek!

@chisigma, your solution post reminds me of this thread(http://mathhelpboards.com/discrete-mathematics-set-theory-logic-15/find-a_%7B100000%7D-8448.html)! Bravo, chisigma!:)

Solution provided by other:

Let $x=\dfrac{1}{2}\cdot\dfrac{3}{4}\cdot\dfrac{5}{6} \cdots\dfrac{999999}{1000000}$.

Thus, what we need to show is that $x<\dfrac{1}{1000}$.

Now, note that

$x^2=\dfrac{1^2}{2^2}\cdot\dfrac{3^2}{4^2}\cdot \dfrac{5^2}{6^2} \cdots\dfrac{999999^2}{1000000^2}$

Since decreasing the denominator of a fraction makes it bigger, we have that

$\dfrac{1^2}{2^2}\le \dfrac{1^2}{2^2-1}= \dfrac{1^2}{(2-1)(2+1)}=\dfrac{1^2}{1\cdot3}$

$\dfrac{3^2}{4^2}\le \dfrac{3^2}{4^2-1}= \dfrac{3^2}{(4-1)(4+1)}=\dfrac{3^2}{3\cdot5}$

$\dfrac{5^2}{6^2}\le \dfrac{5^2}{6^2-1}= \dfrac{5^2}{(6-1)(6+1)}=\dfrac{5^2}{5\cdot7}$

$\vdots\;\;\;\;\;\;\;\;\;\;\;\vdots$

$\dfrac{999999^2}{1000000^2}\le \dfrac{999999^2}{1000000^2-1}= \dfrac{999999^2}{(1000000-1)(1000000+1)}=\dfrac{999999^2}{999999\cdot1000001}$

Multiplying all these together we get

$x^2<\dfrac{1}{1000001}<\dfrac{1}{1000000}$

Now, taking square roof of both sides we obtain

$x<\dfrac{1}{1000}$ or

$\dfrac{1}{2}\cdot\dfrac{3}{4}\cdot\dfrac{5}{6} \cdots\dfrac{999999}{1000000}<\dfrac{1}{1000}$ (Q.E.D.)
 
Last edited:
Seemingly by some mathematical coincidence, a hexagon of sides 2,2,7,7, 11, and 11 can be inscribed in a circle of radius 7. The other day I saw a math problem on line, which they said came from a Polish Olympiad, where you compute the length x of the 3rd side which is the same as the radius, so that the sides of length 2,x, and 11 are inscribed on the arc of a semi-circle. The law of cosines applied twice gives the answer for x of exactly 7, but the arithmetic is so complex that the...
Back
Top