Let $g(k)$ be the greatest odd divisor of $k$ show that $ 0< sum_{k=1}^n frac {g(k)}{k} - frac {2n}{3} lt...












8












$begingroup$



Prove that for all positive intergers $n$,



$$ 0< sum_{k=1}^n frac {g(k)}{k} - frac {2n}{3} < frac {2}{3}$$



Where $g(k)$ denotes the greatest odd divisor of $k$.




Here's my try:



All numbers $k$ can be written as $k= 2^ts$, for nonnegative $t$ and odd $s$, therefore if $k= 2^ts$, then $frac {g(k)}{k} = frac {1}{2^k}$ i.e $frac {g(k)}{k}$ is equal to $1$ divided by the highest power of $2$ dividing $k$. I first thought of proving the inequality for $k= 2^n (n>1)$. Let $Q= sum_{k=1}^{2^n} frac {g(k)}{k}$, then:



$$Q = frac {1}{2}q_1 +frac {1}{2^2}q_2 + cdots + frac {1}{2^{n -1}} q_{2^{n -1}}+ frac {1}{2^n} q_{2^n}+ 2^{n-1} $$.



$q_i$ is the number of times $frac {1}{2^i}$ is added in the summation. It's easy to notice that $q_{2^n} =1$. $q_i$ for $0< i < 2^n$ would be equal to $2^{n-1-i}$ (comes from this $(2^i)(2(2^{n-1-i})-1)$). Then:



$$Q= 2^{n-1}+ frac {1}{2^n} + sum_{i=1}^{n-1} frac{1}{2^i} cdot 2^{n-1-i} $$



$$Q = 2^{n-1}+ frac {1}{2^n} + 2^{n-1} sum_{i=1}^{n-1} (frac{1}{4})^i $$



EDITED



With some algebra we get that:



$$Q - frac {2}{3} cdot 2^n= - frac {4}{3} cdot 2^{n-1} + 2^{n-1}+ frac {1}{2^n} + 2^{n-1} cdot frac {4^{n-1}-1}{4^{n-1}} cdot frac {1}{3} < frac {1}{2^n} < frac {2}{3} $$



Also:



$$ Q - frac {2}{3} cdot 2^n = frac {1}{2^n} - frac {2^{n-1}}{4^{n-1}} cdot frac {1}{3}= frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{3}> frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{2}= 0$$



Then I would get:



$$ 0< Q - frac {2}{3} cdot 2^n < frac {2}{3}$$



Well, I thought proving the inequality for $k=2^n$ because I had some idea about how many times the powers of $2$ appeared. I then thought that I could go backwards with induction but I'm stuck. I would like to see some other approaches but I would also like to know if it's possible to solve the problem from the point where I am.



I'll appreciate any hints or help, thanks in advance.










share|cite|improve this question











$endgroup$












  • $begingroup$
    Check the summation of $(1/4)^i$ part again. The 3 should come out in the denominator.
    $endgroup$
    – Marco
    Dec 16 '18 at 2:03










  • $begingroup$
    Your summation is missing a $q_0$ term.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:15










  • $begingroup$
    @Marco you are right, thank you.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21










  • $begingroup$
    @JimmyK4542 Well, I forgot to say that but I did include it, it is in the part where I add $2^{n−1}$ since half of the numbers are odd.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21
















8












$begingroup$



Prove that for all positive intergers $n$,



$$ 0< sum_{k=1}^n frac {g(k)}{k} - frac {2n}{3} < frac {2}{3}$$



Where $g(k)$ denotes the greatest odd divisor of $k$.




Here's my try:



All numbers $k$ can be written as $k= 2^ts$, for nonnegative $t$ and odd $s$, therefore if $k= 2^ts$, then $frac {g(k)}{k} = frac {1}{2^k}$ i.e $frac {g(k)}{k}$ is equal to $1$ divided by the highest power of $2$ dividing $k$. I first thought of proving the inequality for $k= 2^n (n>1)$. Let $Q= sum_{k=1}^{2^n} frac {g(k)}{k}$, then:



$$Q = frac {1}{2}q_1 +frac {1}{2^2}q_2 + cdots + frac {1}{2^{n -1}} q_{2^{n -1}}+ frac {1}{2^n} q_{2^n}+ 2^{n-1} $$.



$q_i$ is the number of times $frac {1}{2^i}$ is added in the summation. It's easy to notice that $q_{2^n} =1$. $q_i$ for $0< i < 2^n$ would be equal to $2^{n-1-i}$ (comes from this $(2^i)(2(2^{n-1-i})-1)$). Then:



$$Q= 2^{n-1}+ frac {1}{2^n} + sum_{i=1}^{n-1} frac{1}{2^i} cdot 2^{n-1-i} $$



$$Q = 2^{n-1}+ frac {1}{2^n} + 2^{n-1} sum_{i=1}^{n-1} (frac{1}{4})^i $$



EDITED



With some algebra we get that:



$$Q - frac {2}{3} cdot 2^n= - frac {4}{3} cdot 2^{n-1} + 2^{n-1}+ frac {1}{2^n} + 2^{n-1} cdot frac {4^{n-1}-1}{4^{n-1}} cdot frac {1}{3} < frac {1}{2^n} < frac {2}{3} $$



Also:



$$ Q - frac {2}{3} cdot 2^n = frac {1}{2^n} - frac {2^{n-1}}{4^{n-1}} cdot frac {1}{3}= frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{3}> frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{2}= 0$$



Then I would get:



$$ 0< Q - frac {2}{3} cdot 2^n < frac {2}{3}$$



Well, I thought proving the inequality for $k=2^n$ because I had some idea about how many times the powers of $2$ appeared. I then thought that I could go backwards with induction but I'm stuck. I would like to see some other approaches but I would also like to know if it's possible to solve the problem from the point where I am.



I'll appreciate any hints or help, thanks in advance.










share|cite|improve this question











$endgroup$












  • $begingroup$
    Check the summation of $(1/4)^i$ part again. The 3 should come out in the denominator.
    $endgroup$
    – Marco
    Dec 16 '18 at 2:03










  • $begingroup$
    Your summation is missing a $q_0$ term.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:15










  • $begingroup$
    @Marco you are right, thank you.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21










  • $begingroup$
    @JimmyK4542 Well, I forgot to say that but I did include it, it is in the part where I add $2^{n−1}$ since half of the numbers are odd.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21














8












8








8


4



$begingroup$



Prove that for all positive intergers $n$,



$$ 0< sum_{k=1}^n frac {g(k)}{k} - frac {2n}{3} < frac {2}{3}$$



Where $g(k)$ denotes the greatest odd divisor of $k$.




Here's my try:



All numbers $k$ can be written as $k= 2^ts$, for nonnegative $t$ and odd $s$, therefore if $k= 2^ts$, then $frac {g(k)}{k} = frac {1}{2^k}$ i.e $frac {g(k)}{k}$ is equal to $1$ divided by the highest power of $2$ dividing $k$. I first thought of proving the inequality for $k= 2^n (n>1)$. Let $Q= sum_{k=1}^{2^n} frac {g(k)}{k}$, then:



$$Q = frac {1}{2}q_1 +frac {1}{2^2}q_2 + cdots + frac {1}{2^{n -1}} q_{2^{n -1}}+ frac {1}{2^n} q_{2^n}+ 2^{n-1} $$.



$q_i$ is the number of times $frac {1}{2^i}$ is added in the summation. It's easy to notice that $q_{2^n} =1$. $q_i$ for $0< i < 2^n$ would be equal to $2^{n-1-i}$ (comes from this $(2^i)(2(2^{n-1-i})-1)$). Then:



$$Q= 2^{n-1}+ frac {1}{2^n} + sum_{i=1}^{n-1} frac{1}{2^i} cdot 2^{n-1-i} $$



$$Q = 2^{n-1}+ frac {1}{2^n} + 2^{n-1} sum_{i=1}^{n-1} (frac{1}{4})^i $$



EDITED



With some algebra we get that:



$$Q - frac {2}{3} cdot 2^n= - frac {4}{3} cdot 2^{n-1} + 2^{n-1}+ frac {1}{2^n} + 2^{n-1} cdot frac {4^{n-1}-1}{4^{n-1}} cdot frac {1}{3} < frac {1}{2^n} < frac {2}{3} $$



Also:



$$ Q - frac {2}{3} cdot 2^n = frac {1}{2^n} - frac {2^{n-1}}{4^{n-1}} cdot frac {1}{3}= frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{3}> frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{2}= 0$$



Then I would get:



$$ 0< Q - frac {2}{3} cdot 2^n < frac {2}{3}$$



Well, I thought proving the inequality for $k=2^n$ because I had some idea about how many times the powers of $2$ appeared. I then thought that I could go backwards with induction but I'm stuck. I would like to see some other approaches but I would also like to know if it's possible to solve the problem from the point where I am.



I'll appreciate any hints or help, thanks in advance.










share|cite|improve this question











$endgroup$





Prove that for all positive intergers $n$,



$$ 0< sum_{k=1}^n frac {g(k)}{k} - frac {2n}{3} < frac {2}{3}$$



Where $g(k)$ denotes the greatest odd divisor of $k$.




Here's my try:



All numbers $k$ can be written as $k= 2^ts$, for nonnegative $t$ and odd $s$, therefore if $k= 2^ts$, then $frac {g(k)}{k} = frac {1}{2^k}$ i.e $frac {g(k)}{k}$ is equal to $1$ divided by the highest power of $2$ dividing $k$. I first thought of proving the inequality for $k= 2^n (n>1)$. Let $Q= sum_{k=1}^{2^n} frac {g(k)}{k}$, then:



$$Q = frac {1}{2}q_1 +frac {1}{2^2}q_2 + cdots + frac {1}{2^{n -1}} q_{2^{n -1}}+ frac {1}{2^n} q_{2^n}+ 2^{n-1} $$.



$q_i$ is the number of times $frac {1}{2^i}$ is added in the summation. It's easy to notice that $q_{2^n} =1$. $q_i$ for $0< i < 2^n$ would be equal to $2^{n-1-i}$ (comes from this $(2^i)(2(2^{n-1-i})-1)$). Then:



$$Q= 2^{n-1}+ frac {1}{2^n} + sum_{i=1}^{n-1} frac{1}{2^i} cdot 2^{n-1-i} $$



$$Q = 2^{n-1}+ frac {1}{2^n} + 2^{n-1} sum_{i=1}^{n-1} (frac{1}{4})^i $$



EDITED



With some algebra we get that:



$$Q - frac {2}{3} cdot 2^n= - frac {4}{3} cdot 2^{n-1} + 2^{n-1}+ frac {1}{2^n} + 2^{n-1} cdot frac {4^{n-1}-1}{4^{n-1}} cdot frac {1}{3} < frac {1}{2^n} < frac {2}{3} $$



Also:



$$ Q - frac {2}{3} cdot 2^n = frac {1}{2^n} - frac {2^{n-1}}{4^{n-1}} cdot frac {1}{3}= frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{3}> frac {1}{2^n} - frac {1}{2^{n-1}} cdot frac {1}{2}= 0$$



Then I would get:



$$ 0< Q - frac {2}{3} cdot 2^n < frac {2}{3}$$



Well, I thought proving the inequality for $k=2^n$ because I had some idea about how many times the powers of $2$ appeared. I then thought that I could go backwards with induction but I'm stuck. I would like to see some other approaches but I would also like to know if it's possible to solve the problem from the point where I am.



I'll appreciate any hints or help, thanks in advance.







sequences-and-series number-theory inequality divisibility analytic-number-theory






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Dec 18 '18 at 9:45









Batominovski

33.1k33293




33.1k33293










asked Dec 15 '18 at 5:19









VmimiVmimi

372212




372212












  • $begingroup$
    Check the summation of $(1/4)^i$ part again. The 3 should come out in the denominator.
    $endgroup$
    – Marco
    Dec 16 '18 at 2:03










  • $begingroup$
    Your summation is missing a $q_0$ term.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:15










  • $begingroup$
    @Marco you are right, thank you.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21










  • $begingroup$
    @JimmyK4542 Well, I forgot to say that but I did include it, it is in the part where I add $2^{n−1}$ since half of the numbers are odd.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21


















  • $begingroup$
    Check the summation of $(1/4)^i$ part again. The 3 should come out in the denominator.
    $endgroup$
    – Marco
    Dec 16 '18 at 2:03










  • $begingroup$
    Your summation is missing a $q_0$ term.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:15










  • $begingroup$
    @Marco you are right, thank you.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21










  • $begingroup$
    @JimmyK4542 Well, I forgot to say that but I did include it, it is in the part where I add $2^{n−1}$ since half of the numbers are odd.
    $endgroup$
    – Vmimi
    Dec 17 '18 at 4:21
















$begingroup$
Check the summation of $(1/4)^i$ part again. The 3 should come out in the denominator.
$endgroup$
– Marco
Dec 16 '18 at 2:03




$begingroup$
Check the summation of $(1/4)^i$ part again. The 3 should come out in the denominator.
$endgroup$
– Marco
Dec 16 '18 at 2:03












$begingroup$
Your summation is missing a $q_0$ term.
$endgroup$
– JimmyK4542
Dec 16 '18 at 2:15




$begingroup$
Your summation is missing a $q_0$ term.
$endgroup$
– JimmyK4542
Dec 16 '18 at 2:15












$begingroup$
@Marco you are right, thank you.
$endgroup$
– Vmimi
Dec 17 '18 at 4:21




$begingroup$
@Marco you are right, thank you.
$endgroup$
– Vmimi
Dec 17 '18 at 4:21












$begingroup$
@JimmyK4542 Well, I forgot to say that but I did include it, it is in the part where I add $2^{n−1}$ since half of the numbers are odd.
$endgroup$
– Vmimi
Dec 17 '18 at 4:21




$begingroup$
@JimmyK4542 Well, I forgot to say that but I did include it, it is in the part where I add $2^{n−1}$ since half of the numbers are odd.
$endgroup$
– Vmimi
Dec 17 '18 at 4:21










2 Answers
2






active

oldest

votes


















2












$begingroup$

Since $2^jmid k$ for each $jle v_2(k)$ and $sumlimits_{j=1}^n2^{-j}=1-2^{-n}$, we have
$$
2^{-v_2(k)}=1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}tag1
$$

where $[dots]$ are Iverson brackets. Thus,
$$
begin{align}
sum_{k=1}^nfrac{g(k)}k
&=sum_{k=1}^n2^{-v_2(k)}\
&=sum_{k=1}^nleft(1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}right)\
&=n-sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}tag2
end{align}
$$

Furthermore, since $leftlfloorfrac n{2^j}rightrfloorlefrac n{2^j}$ with equality iff $nequiv0pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&lesum_{j=1}^inftyfrac n{2^j}frac1{2^j}\
&=frac n3tag3
end{align}
$$

and $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$ with equality iff $nequiv-1pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&gesum_{j=1}^inftyfrac{n-(2^j-1)}{2^j}frac1{2^j}\
&=frac n3-frac23tag4
end{align}
$$

Equality in $(3)$ can only occur if $nequiv0pmod{2^j}$ for all $j$ ($implies n=0$) and equality in $(4)$ can only occur if $nequiv-1pmod{2^j}$ for all $j$ ($implies n=-1$). Therefore, for $nge1$,
$$
frac{2n}3ltsum_{k=1}^nfrac{g(k)}kltfrac{2n}3+frac23tag5
$$






share|cite|improve this answer











$endgroup$













  • $begingroup$
    I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:49










  • $begingroup$
    And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:56










  • $begingroup$
    For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
    $endgroup$
    – robjohn
    Dec 19 '18 at 21:40





















5












$begingroup$

Let $v(k)$ be the largest integer $m$ such that $2^m$ divides $k$. As you noted, $dfrac{g(k)}{k} = 2^{-v(k)}$.



So, let $S_n = displaystylesum_{k = 1}^{n}dfrac{g(k)}{k} = sum_{k = 1}^{n}2^{-v(k)}$.



Then, we have:



begin{align*}
S_{2n} &= displaystylesum_{k = 1}^{2n}2^{-v(k)}
\
&= sum_{k = 1}^{n}2^{-v(2k-1)} + sum_{k = 1}^{n}2^{-v(2k)}
\
&= sum_{k = 1}^{n}2^{-0} + sum_{k = 1}^{n}2^{-(1+v(k))}
\
&= sum_{k = 1}^{n}1 + dfrac{1}{2}sum_{k = 1}^{n}2^{-v(k)}
\
&= n + dfrac{1}{2}S_n
end{align*}



Also, $S_{2n+1} = displaystylesum_{k = 1}^{2n+1}2^{-v(k)} = 2^{-v(2n+1)} + sum_{k = 1}^{2n}2^{-v(k)} = 2^{0} + S_{2n} = S_{2n} + 1$.



We can now proceed by induction. Trivially, $S_1 = 1$, so $S_1 > dfrac{2}{3}$ and $S_1 < dfrac{4}{3}$.



Now, suppose that for some integer $n$, we have $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$.



Then, we have:



$$S_{2n} = dfrac{1}{2}S_n + n > dfrac{1}{2}left(dfrac{2n}{3}right) + n = dfrac{4n}{3} = dfrac{2 cdot 2n}{3}$$



$$S_{2n} = dfrac{1}{2}S_n + n < dfrac{1}{2}left(dfrac{2n}{3}+dfrac{2}{3}right) + n = dfrac{4n}{3} + dfrac{1}{3} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$$



$$S_{2n+1} = S_{2n}+1 > left(dfrac{4n}{3}right)+1 > dfrac{4n}{3}+dfrac{2}{3} = dfrac{2(2n+1)}{3}$$



$$S_{2n+1} = S_{2n}+1 < left(dfrac{4n}{3}+dfrac{1}{3}right)+1 = dfrac{2(2n+1)}{3} + dfrac{2}{3}.$$



Hence, $dfrac{2 cdot 2n}{3} < S_{2n} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$ and $dfrac{2(2n+1)}{3} < S_{2n+1} < dfrac{2(2n+1)}{3} + dfrac{2}{3}$.



So by induction, we have that $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$ for all $n$, as desired.






share|cite|improve this answer









$endgroup$









  • 1




    $begingroup$
    So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 2:35








  • 1




    $begingroup$
    If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:46








  • 1




    $begingroup$
    So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 9:52






  • 1




    $begingroup$
    If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 10:43








  • 1




    $begingroup$
    I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 11:10













Your Answer





StackExchange.ifUsing("editor", function () {
return StackExchange.using("mathjaxEditing", function () {
StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
});
});
}, "mathjax-editing");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "69"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});

function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
noCode: true, onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});


}
});














draft saved

draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3040201%2flet-gk-be-the-greatest-odd-divisor-of-k-show-that-0-sum-k-1n-frac%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























2 Answers
2






active

oldest

votes








2 Answers
2






active

oldest

votes









active

oldest

votes






active

oldest

votes









2












$begingroup$

Since $2^jmid k$ for each $jle v_2(k)$ and $sumlimits_{j=1}^n2^{-j}=1-2^{-n}$, we have
$$
2^{-v_2(k)}=1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}tag1
$$

where $[dots]$ are Iverson brackets. Thus,
$$
begin{align}
sum_{k=1}^nfrac{g(k)}k
&=sum_{k=1}^n2^{-v_2(k)}\
&=sum_{k=1}^nleft(1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}right)\
&=n-sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}tag2
end{align}
$$

Furthermore, since $leftlfloorfrac n{2^j}rightrfloorlefrac n{2^j}$ with equality iff $nequiv0pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&lesum_{j=1}^inftyfrac n{2^j}frac1{2^j}\
&=frac n3tag3
end{align}
$$

and $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$ with equality iff $nequiv-1pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&gesum_{j=1}^inftyfrac{n-(2^j-1)}{2^j}frac1{2^j}\
&=frac n3-frac23tag4
end{align}
$$

Equality in $(3)$ can only occur if $nequiv0pmod{2^j}$ for all $j$ ($implies n=0$) and equality in $(4)$ can only occur if $nequiv-1pmod{2^j}$ for all $j$ ($implies n=-1$). Therefore, for $nge1$,
$$
frac{2n}3ltsum_{k=1}^nfrac{g(k)}kltfrac{2n}3+frac23tag5
$$






share|cite|improve this answer











$endgroup$













  • $begingroup$
    I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:49










  • $begingroup$
    And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:56










  • $begingroup$
    For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
    $endgroup$
    – robjohn
    Dec 19 '18 at 21:40


















2












$begingroup$

Since $2^jmid k$ for each $jle v_2(k)$ and $sumlimits_{j=1}^n2^{-j}=1-2^{-n}$, we have
$$
2^{-v_2(k)}=1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}tag1
$$

where $[dots]$ are Iverson brackets. Thus,
$$
begin{align}
sum_{k=1}^nfrac{g(k)}k
&=sum_{k=1}^n2^{-v_2(k)}\
&=sum_{k=1}^nleft(1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}right)\
&=n-sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}tag2
end{align}
$$

Furthermore, since $leftlfloorfrac n{2^j}rightrfloorlefrac n{2^j}$ with equality iff $nequiv0pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&lesum_{j=1}^inftyfrac n{2^j}frac1{2^j}\
&=frac n3tag3
end{align}
$$

and $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$ with equality iff $nequiv-1pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&gesum_{j=1}^inftyfrac{n-(2^j-1)}{2^j}frac1{2^j}\
&=frac n3-frac23tag4
end{align}
$$

Equality in $(3)$ can only occur if $nequiv0pmod{2^j}$ for all $j$ ($implies n=0$) and equality in $(4)$ can only occur if $nequiv-1pmod{2^j}$ for all $j$ ($implies n=-1$). Therefore, for $nge1$,
$$
frac{2n}3ltsum_{k=1}^nfrac{g(k)}kltfrac{2n}3+frac23tag5
$$






share|cite|improve this answer











$endgroup$













  • $begingroup$
    I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:49










  • $begingroup$
    And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:56










  • $begingroup$
    For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
    $endgroup$
    – robjohn
    Dec 19 '18 at 21:40
















2












2








2





$begingroup$

Since $2^jmid k$ for each $jle v_2(k)$ and $sumlimits_{j=1}^n2^{-j}=1-2^{-n}$, we have
$$
2^{-v_2(k)}=1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}tag1
$$

where $[dots]$ are Iverson brackets. Thus,
$$
begin{align}
sum_{k=1}^nfrac{g(k)}k
&=sum_{k=1}^n2^{-v_2(k)}\
&=sum_{k=1}^nleft(1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}right)\
&=n-sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}tag2
end{align}
$$

Furthermore, since $leftlfloorfrac n{2^j}rightrfloorlefrac n{2^j}$ with equality iff $nequiv0pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&lesum_{j=1}^inftyfrac n{2^j}frac1{2^j}\
&=frac n3tag3
end{align}
$$

and $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$ with equality iff $nequiv-1pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&gesum_{j=1}^inftyfrac{n-(2^j-1)}{2^j}frac1{2^j}\
&=frac n3-frac23tag4
end{align}
$$

Equality in $(3)$ can only occur if $nequiv0pmod{2^j}$ for all $j$ ($implies n=0$) and equality in $(4)$ can only occur if $nequiv-1pmod{2^j}$ for all $j$ ($implies n=-1$). Therefore, for $nge1$,
$$
frac{2n}3ltsum_{k=1}^nfrac{g(k)}kltfrac{2n}3+frac23tag5
$$






share|cite|improve this answer











$endgroup$



Since $2^jmid k$ for each $jle v_2(k)$ and $sumlimits_{j=1}^n2^{-j}=1-2^{-n}$, we have
$$
2^{-v_2(k)}=1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}tag1
$$

where $[dots]$ are Iverson brackets. Thus,
$$
begin{align}
sum_{k=1}^nfrac{g(k)}k
&=sum_{k=1}^n2^{-v_2(k)}\
&=sum_{k=1}^nleft(1-sum_{j=1}^inftyleft[,2^j,middle|,k,right]2^{-j}right)\
&=n-sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}tag2
end{align}
$$

Furthermore, since $leftlfloorfrac n{2^j}rightrfloorlefrac n{2^j}$ with equality iff $nequiv0pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&lesum_{j=1}^inftyfrac n{2^j}frac1{2^j}\
&=frac n3tag3
end{align}
$$

and $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$ with equality iff $nequiv-1pmod{2^j}$,
$$
begin{align}
sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}
&gesum_{j=1}^inftyfrac{n-(2^j-1)}{2^j}frac1{2^j}\
&=frac n3-frac23tag4
end{align}
$$

Equality in $(3)$ can only occur if $nequiv0pmod{2^j}$ for all $j$ ($implies n=0$) and equality in $(4)$ can only occur if $nequiv-1pmod{2^j}$ for all $j$ ($implies n=-1$). Therefore, for $nge1$,
$$
frac{2n}3ltsum_{k=1}^nfrac{g(k)}kltfrac{2n}3+frac23tag5
$$







share|cite|improve this answer














share|cite|improve this answer



share|cite|improve this answer








edited Dec 18 '18 at 14:00

























answered Dec 17 '18 at 6:27









robjohnrobjohn

269k27309635




269k27309635












  • $begingroup$
    I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:49










  • $begingroup$
    And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:56










  • $begingroup$
    For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
    $endgroup$
    – robjohn
    Dec 19 '18 at 21:40




















  • $begingroup$
    I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:49










  • $begingroup$
    And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
    $endgroup$
    – Vmimi
    Dec 19 '18 at 20:56










  • $begingroup$
    For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
    $endgroup$
    – robjohn
    Dec 19 '18 at 21:40


















$begingroup$
I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
$endgroup$
– Vmimi
Dec 19 '18 at 20:49




$begingroup$
I understood almost everything about your proof. In $(2)$, in the final summation I know per each $j$ you have to add bunches of $2^{-j}$. I think you had to add the amount of numbers that are divisible by $2^{-j}$, that's why it is $sum_{j=1}^inftyleftlfloorfrac n{2^j}rightrfloorfrac1{2^j}$?
$endgroup$
– Vmimi
Dec 19 '18 at 20:49












$begingroup$
And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
$endgroup$
– Vmimi
Dec 19 '18 at 20:56




$begingroup$
And in $4)$ you said that $leftlfloorfrac n{2^j}rightrfloorgefrac {n-left(2^j-1right)}{2^j}$, that's because if you let $leftlfloorfrac n{2^j}rightrfloor = x$, then $frac {n+1}{2^j} < x+1$ unless $n+1$ is divisible by $2^j$?.
$endgroup$
– Vmimi
Dec 19 '18 at 20:56












$begingroup$
For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
$endgroup$
– robjohn
Dec 19 '18 at 21:40






$begingroup$
For $(2)$: yes. For $(4)$: $leftlfloorfrac n{2^j}rightrfloorgefrac {n+1}{2^j}-1iffleftlfloorfrac n{2^j}rightrfloorgtfrac {n}{2^j}-1$, which is the same thing.
$endgroup$
– robjohn
Dec 19 '18 at 21:40













5












$begingroup$

Let $v(k)$ be the largest integer $m$ such that $2^m$ divides $k$. As you noted, $dfrac{g(k)}{k} = 2^{-v(k)}$.



So, let $S_n = displaystylesum_{k = 1}^{n}dfrac{g(k)}{k} = sum_{k = 1}^{n}2^{-v(k)}$.



Then, we have:



begin{align*}
S_{2n} &= displaystylesum_{k = 1}^{2n}2^{-v(k)}
\
&= sum_{k = 1}^{n}2^{-v(2k-1)} + sum_{k = 1}^{n}2^{-v(2k)}
\
&= sum_{k = 1}^{n}2^{-0} + sum_{k = 1}^{n}2^{-(1+v(k))}
\
&= sum_{k = 1}^{n}1 + dfrac{1}{2}sum_{k = 1}^{n}2^{-v(k)}
\
&= n + dfrac{1}{2}S_n
end{align*}



Also, $S_{2n+1} = displaystylesum_{k = 1}^{2n+1}2^{-v(k)} = 2^{-v(2n+1)} + sum_{k = 1}^{2n}2^{-v(k)} = 2^{0} + S_{2n} = S_{2n} + 1$.



We can now proceed by induction. Trivially, $S_1 = 1$, so $S_1 > dfrac{2}{3}$ and $S_1 < dfrac{4}{3}$.



Now, suppose that for some integer $n$, we have $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$.



Then, we have:



$$S_{2n} = dfrac{1}{2}S_n + n > dfrac{1}{2}left(dfrac{2n}{3}right) + n = dfrac{4n}{3} = dfrac{2 cdot 2n}{3}$$



$$S_{2n} = dfrac{1}{2}S_n + n < dfrac{1}{2}left(dfrac{2n}{3}+dfrac{2}{3}right) + n = dfrac{4n}{3} + dfrac{1}{3} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$$



$$S_{2n+1} = S_{2n}+1 > left(dfrac{4n}{3}right)+1 > dfrac{4n}{3}+dfrac{2}{3} = dfrac{2(2n+1)}{3}$$



$$S_{2n+1} = S_{2n}+1 < left(dfrac{4n}{3}+dfrac{1}{3}right)+1 = dfrac{2(2n+1)}{3} + dfrac{2}{3}.$$



Hence, $dfrac{2 cdot 2n}{3} < S_{2n} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$ and $dfrac{2(2n+1)}{3} < S_{2n+1} < dfrac{2(2n+1)}{3} + dfrac{2}{3}$.



So by induction, we have that $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$ for all $n$, as desired.






share|cite|improve this answer









$endgroup$









  • 1




    $begingroup$
    So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 2:35








  • 1




    $begingroup$
    If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:46








  • 1




    $begingroup$
    So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 9:52






  • 1




    $begingroup$
    If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 10:43








  • 1




    $begingroup$
    I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 11:10


















5












$begingroup$

Let $v(k)$ be the largest integer $m$ such that $2^m$ divides $k$. As you noted, $dfrac{g(k)}{k} = 2^{-v(k)}$.



So, let $S_n = displaystylesum_{k = 1}^{n}dfrac{g(k)}{k} = sum_{k = 1}^{n}2^{-v(k)}$.



Then, we have:



begin{align*}
S_{2n} &= displaystylesum_{k = 1}^{2n}2^{-v(k)}
\
&= sum_{k = 1}^{n}2^{-v(2k-1)} + sum_{k = 1}^{n}2^{-v(2k)}
\
&= sum_{k = 1}^{n}2^{-0} + sum_{k = 1}^{n}2^{-(1+v(k))}
\
&= sum_{k = 1}^{n}1 + dfrac{1}{2}sum_{k = 1}^{n}2^{-v(k)}
\
&= n + dfrac{1}{2}S_n
end{align*}



Also, $S_{2n+1} = displaystylesum_{k = 1}^{2n+1}2^{-v(k)} = 2^{-v(2n+1)} + sum_{k = 1}^{2n}2^{-v(k)} = 2^{0} + S_{2n} = S_{2n} + 1$.



We can now proceed by induction. Trivially, $S_1 = 1$, so $S_1 > dfrac{2}{3}$ and $S_1 < dfrac{4}{3}$.



Now, suppose that for some integer $n$, we have $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$.



Then, we have:



$$S_{2n} = dfrac{1}{2}S_n + n > dfrac{1}{2}left(dfrac{2n}{3}right) + n = dfrac{4n}{3} = dfrac{2 cdot 2n}{3}$$



$$S_{2n} = dfrac{1}{2}S_n + n < dfrac{1}{2}left(dfrac{2n}{3}+dfrac{2}{3}right) + n = dfrac{4n}{3} + dfrac{1}{3} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$$



$$S_{2n+1} = S_{2n}+1 > left(dfrac{4n}{3}right)+1 > dfrac{4n}{3}+dfrac{2}{3} = dfrac{2(2n+1)}{3}$$



$$S_{2n+1} = S_{2n}+1 < left(dfrac{4n}{3}+dfrac{1}{3}right)+1 = dfrac{2(2n+1)}{3} + dfrac{2}{3}.$$



Hence, $dfrac{2 cdot 2n}{3} < S_{2n} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$ and $dfrac{2(2n+1)}{3} < S_{2n+1} < dfrac{2(2n+1)}{3} + dfrac{2}{3}$.



So by induction, we have that $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$ for all $n$, as desired.






share|cite|improve this answer









$endgroup$









  • 1




    $begingroup$
    So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 2:35








  • 1




    $begingroup$
    If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:46








  • 1




    $begingroup$
    So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 9:52






  • 1




    $begingroup$
    If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 10:43








  • 1




    $begingroup$
    I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 11:10
















5












5








5





$begingroup$

Let $v(k)$ be the largest integer $m$ such that $2^m$ divides $k$. As you noted, $dfrac{g(k)}{k} = 2^{-v(k)}$.



So, let $S_n = displaystylesum_{k = 1}^{n}dfrac{g(k)}{k} = sum_{k = 1}^{n}2^{-v(k)}$.



Then, we have:



begin{align*}
S_{2n} &= displaystylesum_{k = 1}^{2n}2^{-v(k)}
\
&= sum_{k = 1}^{n}2^{-v(2k-1)} + sum_{k = 1}^{n}2^{-v(2k)}
\
&= sum_{k = 1}^{n}2^{-0} + sum_{k = 1}^{n}2^{-(1+v(k))}
\
&= sum_{k = 1}^{n}1 + dfrac{1}{2}sum_{k = 1}^{n}2^{-v(k)}
\
&= n + dfrac{1}{2}S_n
end{align*}



Also, $S_{2n+1} = displaystylesum_{k = 1}^{2n+1}2^{-v(k)} = 2^{-v(2n+1)} + sum_{k = 1}^{2n}2^{-v(k)} = 2^{0} + S_{2n} = S_{2n} + 1$.



We can now proceed by induction. Trivially, $S_1 = 1$, so $S_1 > dfrac{2}{3}$ and $S_1 < dfrac{4}{3}$.



Now, suppose that for some integer $n$, we have $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$.



Then, we have:



$$S_{2n} = dfrac{1}{2}S_n + n > dfrac{1}{2}left(dfrac{2n}{3}right) + n = dfrac{4n}{3} = dfrac{2 cdot 2n}{3}$$



$$S_{2n} = dfrac{1}{2}S_n + n < dfrac{1}{2}left(dfrac{2n}{3}+dfrac{2}{3}right) + n = dfrac{4n}{3} + dfrac{1}{3} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$$



$$S_{2n+1} = S_{2n}+1 > left(dfrac{4n}{3}right)+1 > dfrac{4n}{3}+dfrac{2}{3} = dfrac{2(2n+1)}{3}$$



$$S_{2n+1} = S_{2n}+1 < left(dfrac{4n}{3}+dfrac{1}{3}right)+1 = dfrac{2(2n+1)}{3} + dfrac{2}{3}.$$



Hence, $dfrac{2 cdot 2n}{3} < S_{2n} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$ and $dfrac{2(2n+1)}{3} < S_{2n+1} < dfrac{2(2n+1)}{3} + dfrac{2}{3}$.



So by induction, we have that $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$ for all $n$, as desired.






share|cite|improve this answer









$endgroup$



Let $v(k)$ be the largest integer $m$ such that $2^m$ divides $k$. As you noted, $dfrac{g(k)}{k} = 2^{-v(k)}$.



So, let $S_n = displaystylesum_{k = 1}^{n}dfrac{g(k)}{k} = sum_{k = 1}^{n}2^{-v(k)}$.



Then, we have:



begin{align*}
S_{2n} &= displaystylesum_{k = 1}^{2n}2^{-v(k)}
\
&= sum_{k = 1}^{n}2^{-v(2k-1)} + sum_{k = 1}^{n}2^{-v(2k)}
\
&= sum_{k = 1}^{n}2^{-0} + sum_{k = 1}^{n}2^{-(1+v(k))}
\
&= sum_{k = 1}^{n}1 + dfrac{1}{2}sum_{k = 1}^{n}2^{-v(k)}
\
&= n + dfrac{1}{2}S_n
end{align*}



Also, $S_{2n+1} = displaystylesum_{k = 1}^{2n+1}2^{-v(k)} = 2^{-v(2n+1)} + sum_{k = 1}^{2n}2^{-v(k)} = 2^{0} + S_{2n} = S_{2n} + 1$.



We can now proceed by induction. Trivially, $S_1 = 1$, so $S_1 > dfrac{2}{3}$ and $S_1 < dfrac{4}{3}$.



Now, suppose that for some integer $n$, we have $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$.



Then, we have:



$$S_{2n} = dfrac{1}{2}S_n + n > dfrac{1}{2}left(dfrac{2n}{3}right) + n = dfrac{4n}{3} = dfrac{2 cdot 2n}{3}$$



$$S_{2n} = dfrac{1}{2}S_n + n < dfrac{1}{2}left(dfrac{2n}{3}+dfrac{2}{3}right) + n = dfrac{4n}{3} + dfrac{1}{3} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$$



$$S_{2n+1} = S_{2n}+1 > left(dfrac{4n}{3}right)+1 > dfrac{4n}{3}+dfrac{2}{3} = dfrac{2(2n+1)}{3}$$



$$S_{2n+1} = S_{2n}+1 < left(dfrac{4n}{3}+dfrac{1}{3}right)+1 = dfrac{2(2n+1)}{3} + dfrac{2}{3}.$$



Hence, $dfrac{2 cdot 2n}{3} < S_{2n} < dfrac{2 cdot 2n}{3} + dfrac{2}{3}$ and $dfrac{2(2n+1)}{3} < S_{2n+1} < dfrac{2(2n+1)}{3} + dfrac{2}{3}$.



So by induction, we have that $dfrac{2n}{3} < S_n < dfrac{2n}{3} + dfrac{2}{3}$ for all $n$, as desired.







share|cite|improve this answer












share|cite|improve this answer



share|cite|improve this answer










answered Dec 16 '18 at 2:12









JimmyK4542JimmyK4542

41.2k245107




41.2k245107








  • 1




    $begingroup$
    So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 2:35








  • 1




    $begingroup$
    If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:46








  • 1




    $begingroup$
    So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 9:52






  • 1




    $begingroup$
    If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 10:43








  • 1




    $begingroup$
    I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 11:10
















  • 1




    $begingroup$
    So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 2:35








  • 1




    $begingroup$
    If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 2:46








  • 1




    $begingroup$
    So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
    $endgroup$
    – marty cohen
    Dec 16 '18 at 9:52






  • 1




    $begingroup$
    If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 10:43








  • 1




    $begingroup$
    I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
    $endgroup$
    – JimmyK4542
    Dec 16 '18 at 11:10










1




1




$begingroup$
So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
$endgroup$
– marty cohen
Dec 16 '18 at 2:35






$begingroup$
So does $lim_{n to infty} S_n-dfrac{2n}{3}$ exist; if so, what is it?
$endgroup$
– marty cohen
Dec 16 '18 at 2:35






1




1




$begingroup$
If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
$endgroup$
– JimmyK4542
Dec 16 '18 at 2:46






$begingroup$
If $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right) = L$, then $displaystylelim_{n to infty}left(S_{2n} - tfrac{2 cdot 2n}{3}right) = L$ and $displaystylelim_{n to infty}left(S_{2n+1} - tfrac{2(2n+1)}{3}right) = L$ as well. But since $S_{2n+1}-tfrac{2(2n+1)}{3} = left(S_{2n}-tfrac{2 cdot 2n}{3}right) + dfrac{1}{3}$ holds for all $n$, taking the limit of both sides as $n to infty$ yields, $L = L + tfrac{1}{3}$, a contradiction. Thus, $displaystylelim_{n to infty}left(S_n - tfrac{2n}{3}right)$ doesn't exist.
$endgroup$
– JimmyK4542
Dec 16 '18 at 2:46






1




1




$begingroup$
So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
$endgroup$
– marty cohen
Dec 16 '18 at 9:52




$begingroup$
So $S_n$ seems to behave differently for even and odd $n$. How about $lim S_{2n+k}-2(2n+k)/3$ for $k=0$ a nd $1$?
$endgroup$
– marty cohen
Dec 16 '18 at 9:52




1




1




$begingroup$
If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
$endgroup$
– JimmyK4542
Dec 16 '18 at 10:43






$begingroup$
If we let $x_n = S_n - tfrac{2n}{3}$, then we have the relations $x_{2n} = tfrac{1}{2}x_n$ and $x_{2n+1} = x_{2n}+tfrac{1}{3}$. From generating the first $2^{20}$ terms of $x_n$ in MATLAB and plotting it, It appears that $displaystylelim_{n to infty}x_{2n+k}$ not exist for $k = 0,1$. Also, I conjecture that $displaystylelim_{n to infty}x_{n cdot 2^m+ k}$ does not exist for any $0 le k le 2^m-1$ and $m in mathbb{N}$.
$endgroup$
– JimmyK4542
Dec 16 '18 at 10:43






1




1




$begingroup$
I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
$endgroup$
– JimmyK4542
Dec 16 '18 at 11:10






$begingroup$
I can also conjecture that for any $m in mathbb{N}$ and $0 le k le 2^m-1$, we have $displaystyleliminflimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k)}{3 cdot 2^{m-1}}$ and $displaystylelimsuplimits_{n to infty}x_{n cdot 2^m + k} = dfrac{r_m(k) + 1}{3 cdot 2^{m-1}}$ where $r_m(k)$ is the number obtained by reversing the $m$ digit binary representation of $k$, e.g. $r_3(1) = r_3(001_2) = 100_2 = 4$.
$endgroup$
– JimmyK4542
Dec 16 '18 at 11:10




















draft saved

draft discarded




















































Thanks for contributing an answer to Mathematics Stack Exchange!


  • Please be sure to answer the question. Provide details and share your research!

But avoid



  • Asking for help, clarification, or responding to other answers.

  • Making statements based on opinion; back them up with references or personal experience.


Use MathJax to format equations. MathJax reference.


To learn more, see our tips on writing great answers.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3040201%2flet-gk-be-the-greatest-odd-divisor-of-k-show-that-0-sum-k-1n-frac%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown





















































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown

































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown







Popular posts from this blog

Plaza Victoria

Puebla de Zaragoza

Musa