1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Fun math proof

  1. Nov 12, 2005 #1
    Any nice proofs for this?


    I hope the tex came out alright. have fun!

    ps. n is any natural number.
    Last edited: Nov 12, 2005
  2. jcsd
  3. Nov 12, 2005 #2
    ill post solution in the morning.
  4. Nov 12, 2005 #3


    User Avatar
    Gold Member

    I doesn't look that hard if you prove each half by induction, and if first you multiply everything by sqrt(n)/sqrt(n). But then again there could be a catch somewere. If I have time I'll try it.
  5. Nov 12, 2005 #4
    2\sqrt{n+1} - 2\sqrt{n} &= 2(\sqrt{n+1} - \sqrt{n}) \cdot \frac{\sqrt{n+1} + \sqrt{n}}{\sqrt{n+1} + \sqrt{n}} \\
    &= \frac{2}{\sqrt{n+1} + \sqrt{n}} \\
    &< \frac{2}{\sqrt{n} + \sqrt{n}} \\
    &= \frac{1}{\sqrt{n}}

    Alternatively, let [itex]f(x) = \sqrt{x}[/itex], then by the MVT on [n, n+1]:

    \frac{\sqrt{n+1} - \sqrt{n}}{1} &= f'(x) \\
    &= \frac{1}{2\sqrt{x}}, \text{ with x} \in (n, n+1) \\
    &< \frac{1}{2\sqrt{n}}

    Second one should be similar.
    Last edited: Nov 12, 2005
  6. Nov 13, 2005 #5
    Yes, my solution ressembled the first one. Although, i was very amused at seeing the MVT making such short work of it. nice job. heres the pickle. i forgot the solution to this, ill think about it over the next few days, but have a try:
    Using this inequality, evaluate to the [tex]\pm 1[/tex] the integer part of the sum
    [tex] 1+\frac{1}{\sqrt{2}}+\frac{1}{\sqrt{3}}+...+\frac{1}{\sqrt{1000000}}[/tex]
    any progress, you let me know.
  7. Nov 13, 2005 #6
    hehe, alright, it came back to me, but take a try at it anyways.
  8. Nov 15, 2005 #7
    Finally, heres the cherry on top. Prove that the integer part of the following expression is odd.


    have fun! im interested in seeing some original solutions as i think there are better ones than the ones i have seen.
    Last edited: Nov 15, 2005
  9. Nov 18, 2005 #8
    no takers? i post solutions tomorrow.
  10. Nov 18, 2005 #9
    I think some appropriate definite integration of 1/sqrt(x) will give the desired approximation.
  11. Nov 18, 2005 #10
    This follows from the following 2 observations:
    [tex](\sqrt{n}+\sqrt{n+1})^2 + (\sqrt{n}-\sqrt{n+1})^2 = 4n+2 [/tex]
    [tex](\sqrt{n}-\sqrt{n+1})^2[/tex] is really small.
  12. Nov 19, 2005 #11


    User Avatar
    Science Advisor
    Homework Helper

    [tex]2\sum_{k=a}^b\left(\sqrt{k+1}-\sqrt{k}\right)<\sum_{k=a}^b\frac{1}{\sqrt k}<2\sum_{k=a}^b\left(\sqrt{k}-\sqrt{k-1}\right)[/tex]

    [tex]2\left(\sqrt{b+1}-\sqrt{a}\right)<\sum_{k=a}^b\frac{1}{\sqrt k}<2\left(\sqrt{b}-\sqrt{a-1}\right)[/tex]

    [tex]2\left(\sqrt{1000001}-\sqrt{1}\right)<\sum_{k=1}^{1000000}\frac{1}{\sqrt k}<2\left(\sqrt{1000000}-\sqrt{0}\right)[/tex]

    [tex]1998<\sum_{k=1}^{1000000}\frac{1}{\sqrt k}<2000[/tex]

    By direct calculation
    [tex]1998.5401<\sum_{k=1}^{1000000}\frac{1}{\sqrt k}<1998.5402[/tex]
    Last edited: Nov 19, 2005
  13. Nov 19, 2005 #12
    well done, youve saved me a bit of typing! yes thats the solution i have as well.
  14. Nov 19, 2005 #13
    interesting, but i find your second premise a little weak, as in... not justified. it definitely is true though. i could be wrong though, its just what if n is very large?
  15. Nov 19, 2005 #14


    User Avatar
    Science Advisor
    Homework Helper

    It decreases strictly as n increases. In particular, if n>0, [tex]-1<\sqrt n-\sqrt{n+1}<0[/tex]. This gives the desired result.
  16. Nov 19, 2005 #15
    interesting! im curious, is there a nice proof of this inequality?
  17. Nov 19, 2005 #16


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    The original problem looks like an easy application of Taylor series! The 0-th order approximation is:

    f(x + e) = f(x) + f'(x^*) e

    for some [itex]x^*[/itex] between x and x+e inclusive.

    Letting f be the square root function and e be 1, we have:

    \sqrt{x + 1} - \sqrt{x} = \frac{1}{2 \sqrt{x^*}} \leq \frac{1}{2 \sqrt{x}}

    because the rightmost term is an upper bound for the middle term. Letting e = -1, we have:

    \sqrt{x - 1} - \sqrt{x} = - \frac{1}{2 \sqrt{x^*}} \geq -\frac{1}{2 \sqrt{x}}

    because the rightmost term is a lower bound for the middle term. (Note the signs!)


    [tex]\sqrt{x+1} - \sqrt{x} \leq \frac{1}{2 \sqrt{x}} \leq \sqrt{x} - \sqrt{x - 1}[/tex]


    [tex]2\sqrt{x+1} - 2\sqrt{x} \leq \frac{1}{\sqrt{x}} \leq 2\sqrt{x} - 2\sqrt{x - 1}[/tex]
    Last edited: Nov 19, 2005
  18. Nov 19, 2005 #17
    Yes. It's actually quite easy to see why this is true, since the terms in the middle are consecutive square roots. A formal proof goes something like this:

    [tex]\sqrt{n+1} - \sqrt{n} > 0 \text{ is obviously true for } n \in \mathbb{N}[/tex]

    It remains to show that:

    [tex]\sqrt{n+1} - \sqrt{n} < 1[/tex]

    Squaring both sides and playing around, we see that this follows from [itex]n < n+1[/itex].

    n < n+1
    \Rightarrow n^2 < n(n+1)
    \Rightarrow n < \sqrt{n(n+1)}
    \Rightarrow 2n - 2\sqrt{n(n+1)} < 0
    \Rightarrow n + 1 - 2\sqrt{n(n+1)} + n < 1
    \Rightarrow (\sqrt{n+1} + \sqrt{n})^2 < 1

    So we have:

    0 < \sqrt{n+1} - \sqrt{n} < 1
    \Rightarrow -1 < \sqrt{n} - \sqrt{n+1} < 0
  19. Nov 19, 2005 #18
    If you say so, i havent finished my cal 2 yet unfortunately, hehe. but ill keep it in mind!
  20. Nov 19, 2005 #19
    oustanding, thanks for that.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook