Convergence of a family of series
Clash Royale CLAN TAG#URR8PPP
up vote
2
down vote
favorite
I want to show that for $D>0, Dequiv 0,1 pmod4$, $z in mathbbC$ with $ operatornameIm(z)>0$ and $k>1$ the sum
$$sum_a,b,c in mathbbZ \ b^2-4ac=D(az^2+bz+c)^-k$$ converges absolutely. I tried to take the absolute value and wrote $z=x+iy$. However, after calculating everything I just got $$|(az^2+bz+c)|^-k=sqrta^2l_1+al_2+acl_3+a^2cl_4+ac^2l_5+a^2c^2l_6+ac^3l_7+c^2l_8+c^4+l_9^-k,$$ where $l_i$ only depends on $x,y,D$ which can be seen as fixed. However, I have no idea how to go on.
sequences-and-series complex-analysis convergence
add a comment |Â
up vote
2
down vote
favorite
I want to show that for $D>0, Dequiv 0,1 pmod4$, $z in mathbbC$ with $ operatornameIm(z)>0$ and $k>1$ the sum
$$sum_a,b,c in mathbbZ \ b^2-4ac=D(az^2+bz+c)^-k$$ converges absolutely. I tried to take the absolute value and wrote $z=x+iy$. However, after calculating everything I just got $$|(az^2+bz+c)|^-k=sqrta^2l_1+al_2+acl_3+a^2cl_4+ac^2l_5+a^2c^2l_6+ac^3l_7+c^2l_8+c^4+l_9^-k,$$ where $l_i$ only depends on $x,y,D$ which can be seen as fixed. However, I have no idea how to go on.
sequences-and-series complex-analysis convergence
What does $Dequiv 0,1 pmod4$ mean?
– zhw.
Aug 2 at 20:40
$D equiv 0 pmod 4$ or $D equiv 1 pmod 4$ sorry if that was cloudy
– Deavor
Aug 2 at 20:50
You might wanted to put $b^2-4ac$ in two places, one under the sum, the other after your attempted proof.
– i707107
Aug 4 at 19:38
Oh yes thanks for noticing the typo
– Deavor
Aug 4 at 20:18
add a comment |Â
up vote
2
down vote
favorite
up vote
2
down vote
favorite
I want to show that for $D>0, Dequiv 0,1 pmod4$, $z in mathbbC$ with $ operatornameIm(z)>0$ and $k>1$ the sum
$$sum_a,b,c in mathbbZ \ b^2-4ac=D(az^2+bz+c)^-k$$ converges absolutely. I tried to take the absolute value and wrote $z=x+iy$. However, after calculating everything I just got $$|(az^2+bz+c)|^-k=sqrta^2l_1+al_2+acl_3+a^2cl_4+ac^2l_5+a^2c^2l_6+ac^3l_7+c^2l_8+c^4+l_9^-k,$$ where $l_i$ only depends on $x,y,D$ which can be seen as fixed. However, I have no idea how to go on.
sequences-and-series complex-analysis convergence
I want to show that for $D>0, Dequiv 0,1 pmod4$, $z in mathbbC$ with $ operatornameIm(z)>0$ and $k>1$ the sum
$$sum_a,b,c in mathbbZ \ b^2-4ac=D(az^2+bz+c)^-k$$ converges absolutely. I tried to take the absolute value and wrote $z=x+iy$. However, after calculating everything I just got $$|(az^2+bz+c)|^-k=sqrta^2l_1+al_2+acl_3+a^2cl_4+ac^2l_5+a^2c^2l_6+ac^3l_7+c^2l_8+c^4+l_9^-k,$$ where $l_i$ only depends on $x,y,D$ which can be seen as fixed. However, I have no idea how to go on.
sequences-and-series complex-analysis convergence
edited 2 days ago
asked Aug 2 at 17:57
Deavor
568412
568412
What does $Dequiv 0,1 pmod4$ mean?
– zhw.
Aug 2 at 20:40
$D equiv 0 pmod 4$ or $D equiv 1 pmod 4$ sorry if that was cloudy
– Deavor
Aug 2 at 20:50
You might wanted to put $b^2-4ac$ in two places, one under the sum, the other after your attempted proof.
– i707107
Aug 4 at 19:38
Oh yes thanks for noticing the typo
– Deavor
Aug 4 at 20:18
add a comment |Â
What does $Dequiv 0,1 pmod4$ mean?
– zhw.
Aug 2 at 20:40
$D equiv 0 pmod 4$ or $D equiv 1 pmod 4$ sorry if that was cloudy
– Deavor
Aug 2 at 20:50
You might wanted to put $b^2-4ac$ in two places, one under the sum, the other after your attempted proof.
– i707107
Aug 4 at 19:38
Oh yes thanks for noticing the typo
– Deavor
Aug 4 at 20:18
What does $Dequiv 0,1 pmod4$ mean?
– zhw.
Aug 2 at 20:40
What does $Dequiv 0,1 pmod4$ mean?
– zhw.
Aug 2 at 20:40
$D equiv 0 pmod 4$ or $D equiv 1 pmod 4$ sorry if that was cloudy
– Deavor
Aug 2 at 20:50
$D equiv 0 pmod 4$ or $D equiv 1 pmod 4$ sorry if that was cloudy
– Deavor
Aug 2 at 20:50
You might wanted to put $b^2-4ac$ in two places, one under the sum, the other after your attempted proof.
– i707107
Aug 4 at 19:38
You might wanted to put $b^2-4ac$ in two places, one under the sum, the other after your attempted proof.
– i707107
Aug 4 at 19:38
Oh yes thanks for noticing the typo
– Deavor
Aug 4 at 20:18
Oh yes thanks for noticing the typo
– Deavor
Aug 4 at 20:18
add a comment |Â
1 Answer
1
active
oldest
votes
up vote
1
down vote
This can be done with Theory of Quadratic Forms. The constants $C$ appearing in this solution may not be the same. We begin with
Definition
Let $ax^2+bxy+cy^2$ be a quadratic form with $a, b, cin mathbbZ$ and $b^2-4ac=D$. We say that a quadratic form $a_1x_1^2+b_1x_1y_1+c_1y_1^2$ with $a_1,b_1,c_1inmathbbZ$ and $b_1^2-4a_1c_1=D$ is equivalent to $ax^2+bxy+cy^2$ if there is $beginbmatrixp & q \ r & s endbmatrixinmathrmSL_2(mathbbZ)$ such that
$$
a_1x_1^2+b_1x_1y_1+c_1y_1^2=a(px_1+qy_1)^2+b(px_1+qy_1)(rx_1+sy_1)+c(rx_1+sy_1)^2.
$$
This can be written again with matrices
$$
beginbmatrixa_1 & fracb_12\ fracb_12 & c_1 endbmatrix = beginbmatrixp & r \ q & s endbmatrix beginbmatrixa & fracb2\ fracb2 & c endbmatrix beginbmatrixp & q \ r & s endbmatrix
$$
and
$$
beginbmatrixx\yendbmatrix=beginbmatrixp & q \ r & s endbmatrixbeginbmatrixx_1\y_1endbmatrix.
$$
The next one is the reducing step.
Lemma 1
Any integer quadratic form with discriminant $D$ is equivalent to $ax^2+bxy+cy^2$ with $|b|leq |a|leq |c|$.
The proof is a repeated application of the matrices in $mathrmSL_2(mathbbZ)$.
$$
beginbmatrix 1 & 1 \ 0 & 1 endbmatrix, beginbmatrix 0 & -1 \ 1 & 0endbmatrix.
$$
Corollary 1
The number of equivalence classes of integer quadratic forms with discriminant $D$ is finite.
If $b^2-4ac=D$ and $|b|leq |a|leq |c|$, then $|D|geq 4|ac|-|b^2| geq 3|ac|$ gives only finitely many choices for $a$ and $c$, subsequently for $b$.
Then it suffices to prove the following for $A=beginbmatrixa& b/2\ b/2&cendbmatrix$ with $b^2-4ac=D$ and $|b|leq |a|leq |c|$, as the sum can be written as a finite sum of the following kind.
The sum is absolutely convergent for $k>1$:
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T A gamma beginbmatrixz\1endbmatrixright)^k.
$$
It suffices to show the following sum is convergent instead, as it differs only by a positive constant.
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T beginbmatrixalpha \ betaendbmatrix beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright)^k
$$
where $alpha=alpha'=2a$, $beta=b+sqrtD$, and $beta'=b-sqrtD$.
We establish the following uniform inequality:
Lemma 2
We have $$
left|beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright|geq C (|pz+q||rz+s|)^1/2
$$
where $C>0$ is an absolute constant depending at most on $a,b,c$, and $gamma=beginbmatrix p&q\r&sendbmatrixinmathrmSL_2(mathbbZ)$.
The proof is by AM-GM inequality.
Now, it suffices to prove the convergence of
$$
sum_gammain mathrmSL_2(mathbbZ)frac1 (*)
$$
Consider $Gamma=mathrmSL_2(mathbbZ)$ and $Gamma_infty=leftpmbeginbmatrix 1&m\0&1endbmatrix, minmathbbZright$.
Fix coprime integers $r$, $s$. Let $p_0,q_0inmathbbZ$ satisfy $p_0s-q_0r=1$. Then it is well-known that all integers $p,q$ satisfying $ps-qr=1$ are $p=p_0+rt$, $q=q_0+st$ with $tinmathbbZ$. Varying $t$ if necessary, we may choose $p_0, q_0$ such that a coset representative $beginbmatrixp_0&q_0\r&sendbmatrix$ in $Gamma_inftybackslash Gamma$ with $-frac12leqfracp_0z+q_0rz+s<frac12$. We have the above sums over $r,s$ with $|t|geq 1$, is convergent since
$$
sum_(r,s)=1 frac1rz+sleq Csum_sfrac1
$$
is convergent where $C>0$ is an absolute constant that depends only on $z$.
We also have that the area of a parallelogram with vertices $0$, $pz+q$, $rz+s$ is the same as that of vertices $0$, $1$, $z$, this is because of $ps-qr=1$. Call this area $C$. Then
$$
|pz+q||rz+s| |sin theta_gamma| = C.
$$
Here $theta_gamma$ is the angle of the corner formed by the three points $pz+q$, $0$, $rz+s$. This angle is also written as $theta_gamma=arg fracpz+qrz+s=arg gamma z$.
Then the convergence of above sum (*) over $r,s$ with $t=0$ follows from that of
$$
sum_gammainGamma_inftybackslashGamma|sin theta_gamma|^k (**)
$$
Lemma 3
Let $Y>0$. We have a constant $C>0$ depending on $z$ such that
$$
# gammainGamma_inftybackslashGamma biggvert leq 4+frac4CY.
$$
This is proved by applying a classical lemma
$$
# gammainGamma_inftybackslashGamma biggvert Im(gamma z)geq Yleq 1+frac CY.
$$
Now, the convergence of (**) follows by partial summation with application of Lemma 3 for $Y=frac1n$, $n=1,2,3,ldots$.
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
add a comment |Â
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
up vote
1
down vote
This can be done with Theory of Quadratic Forms. The constants $C$ appearing in this solution may not be the same. We begin with
Definition
Let $ax^2+bxy+cy^2$ be a quadratic form with $a, b, cin mathbbZ$ and $b^2-4ac=D$. We say that a quadratic form $a_1x_1^2+b_1x_1y_1+c_1y_1^2$ with $a_1,b_1,c_1inmathbbZ$ and $b_1^2-4a_1c_1=D$ is equivalent to $ax^2+bxy+cy^2$ if there is $beginbmatrixp & q \ r & s endbmatrixinmathrmSL_2(mathbbZ)$ such that
$$
a_1x_1^2+b_1x_1y_1+c_1y_1^2=a(px_1+qy_1)^2+b(px_1+qy_1)(rx_1+sy_1)+c(rx_1+sy_1)^2.
$$
This can be written again with matrices
$$
beginbmatrixa_1 & fracb_12\ fracb_12 & c_1 endbmatrix = beginbmatrixp & r \ q & s endbmatrix beginbmatrixa & fracb2\ fracb2 & c endbmatrix beginbmatrixp & q \ r & s endbmatrix
$$
and
$$
beginbmatrixx\yendbmatrix=beginbmatrixp & q \ r & s endbmatrixbeginbmatrixx_1\y_1endbmatrix.
$$
The next one is the reducing step.
Lemma 1
Any integer quadratic form with discriminant $D$ is equivalent to $ax^2+bxy+cy^2$ with $|b|leq |a|leq |c|$.
The proof is a repeated application of the matrices in $mathrmSL_2(mathbbZ)$.
$$
beginbmatrix 1 & 1 \ 0 & 1 endbmatrix, beginbmatrix 0 & -1 \ 1 & 0endbmatrix.
$$
Corollary 1
The number of equivalence classes of integer quadratic forms with discriminant $D$ is finite.
If $b^2-4ac=D$ and $|b|leq |a|leq |c|$, then $|D|geq 4|ac|-|b^2| geq 3|ac|$ gives only finitely many choices for $a$ and $c$, subsequently for $b$.
Then it suffices to prove the following for $A=beginbmatrixa& b/2\ b/2&cendbmatrix$ with $b^2-4ac=D$ and $|b|leq |a|leq |c|$, as the sum can be written as a finite sum of the following kind.
The sum is absolutely convergent for $k>1$:
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T A gamma beginbmatrixz\1endbmatrixright)^k.
$$
It suffices to show the following sum is convergent instead, as it differs only by a positive constant.
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T beginbmatrixalpha \ betaendbmatrix beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright)^k
$$
where $alpha=alpha'=2a$, $beta=b+sqrtD$, and $beta'=b-sqrtD$.
We establish the following uniform inequality:
Lemma 2
We have $$
left|beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright|geq C (|pz+q||rz+s|)^1/2
$$
where $C>0$ is an absolute constant depending at most on $a,b,c$, and $gamma=beginbmatrix p&q\r&sendbmatrixinmathrmSL_2(mathbbZ)$.
The proof is by AM-GM inequality.
Now, it suffices to prove the convergence of
$$
sum_gammain mathrmSL_2(mathbbZ)frac1 (*)
$$
Consider $Gamma=mathrmSL_2(mathbbZ)$ and $Gamma_infty=leftpmbeginbmatrix 1&m\0&1endbmatrix, minmathbbZright$.
Fix coprime integers $r$, $s$. Let $p_0,q_0inmathbbZ$ satisfy $p_0s-q_0r=1$. Then it is well-known that all integers $p,q$ satisfying $ps-qr=1$ are $p=p_0+rt$, $q=q_0+st$ with $tinmathbbZ$. Varying $t$ if necessary, we may choose $p_0, q_0$ such that a coset representative $beginbmatrixp_0&q_0\r&sendbmatrix$ in $Gamma_inftybackslash Gamma$ with $-frac12leqfracp_0z+q_0rz+s<frac12$. We have the above sums over $r,s$ with $|t|geq 1$, is convergent since
$$
sum_(r,s)=1 frac1rz+sleq Csum_sfrac1
$$
is convergent where $C>0$ is an absolute constant that depends only on $z$.
We also have that the area of a parallelogram with vertices $0$, $pz+q$, $rz+s$ is the same as that of vertices $0$, $1$, $z$, this is because of $ps-qr=1$. Call this area $C$. Then
$$
|pz+q||rz+s| |sin theta_gamma| = C.
$$
Here $theta_gamma$ is the angle of the corner formed by the three points $pz+q$, $0$, $rz+s$. This angle is also written as $theta_gamma=arg fracpz+qrz+s=arg gamma z$.
Then the convergence of above sum (*) over $r,s$ with $t=0$ follows from that of
$$
sum_gammainGamma_inftybackslashGamma|sin theta_gamma|^k (**)
$$
Lemma 3
Let $Y>0$. We have a constant $C>0$ depending on $z$ such that
$$
# gammainGamma_inftybackslashGamma biggvert leq 4+frac4CY.
$$
This is proved by applying a classical lemma
$$
# gammainGamma_inftybackslashGamma biggvert Im(gamma z)geq Yleq 1+frac CY.
$$
Now, the convergence of (**) follows by partial summation with application of Lemma 3 for $Y=frac1n$, $n=1,2,3,ldots$.
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
add a comment |Â
up vote
1
down vote
This can be done with Theory of Quadratic Forms. The constants $C$ appearing in this solution may not be the same. We begin with
Definition
Let $ax^2+bxy+cy^2$ be a quadratic form with $a, b, cin mathbbZ$ and $b^2-4ac=D$. We say that a quadratic form $a_1x_1^2+b_1x_1y_1+c_1y_1^2$ with $a_1,b_1,c_1inmathbbZ$ and $b_1^2-4a_1c_1=D$ is equivalent to $ax^2+bxy+cy^2$ if there is $beginbmatrixp & q \ r & s endbmatrixinmathrmSL_2(mathbbZ)$ such that
$$
a_1x_1^2+b_1x_1y_1+c_1y_1^2=a(px_1+qy_1)^2+b(px_1+qy_1)(rx_1+sy_1)+c(rx_1+sy_1)^2.
$$
This can be written again with matrices
$$
beginbmatrixa_1 & fracb_12\ fracb_12 & c_1 endbmatrix = beginbmatrixp & r \ q & s endbmatrix beginbmatrixa & fracb2\ fracb2 & c endbmatrix beginbmatrixp & q \ r & s endbmatrix
$$
and
$$
beginbmatrixx\yendbmatrix=beginbmatrixp & q \ r & s endbmatrixbeginbmatrixx_1\y_1endbmatrix.
$$
The next one is the reducing step.
Lemma 1
Any integer quadratic form with discriminant $D$ is equivalent to $ax^2+bxy+cy^2$ with $|b|leq |a|leq |c|$.
The proof is a repeated application of the matrices in $mathrmSL_2(mathbbZ)$.
$$
beginbmatrix 1 & 1 \ 0 & 1 endbmatrix, beginbmatrix 0 & -1 \ 1 & 0endbmatrix.
$$
Corollary 1
The number of equivalence classes of integer quadratic forms with discriminant $D$ is finite.
If $b^2-4ac=D$ and $|b|leq |a|leq |c|$, then $|D|geq 4|ac|-|b^2| geq 3|ac|$ gives only finitely many choices for $a$ and $c$, subsequently for $b$.
Then it suffices to prove the following for $A=beginbmatrixa& b/2\ b/2&cendbmatrix$ with $b^2-4ac=D$ and $|b|leq |a|leq |c|$, as the sum can be written as a finite sum of the following kind.
The sum is absolutely convergent for $k>1$:
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T A gamma beginbmatrixz\1endbmatrixright)^k.
$$
It suffices to show the following sum is convergent instead, as it differs only by a positive constant.
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T beginbmatrixalpha \ betaendbmatrix beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright)^k
$$
where $alpha=alpha'=2a$, $beta=b+sqrtD$, and $beta'=b-sqrtD$.
We establish the following uniform inequality:
Lemma 2
We have $$
left|beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright|geq C (|pz+q||rz+s|)^1/2
$$
where $C>0$ is an absolute constant depending at most on $a,b,c$, and $gamma=beginbmatrix p&q\r&sendbmatrixinmathrmSL_2(mathbbZ)$.
The proof is by AM-GM inequality.
Now, it suffices to prove the convergence of
$$
sum_gammain mathrmSL_2(mathbbZ)frac1 (*)
$$
Consider $Gamma=mathrmSL_2(mathbbZ)$ and $Gamma_infty=leftpmbeginbmatrix 1&m\0&1endbmatrix, minmathbbZright$.
Fix coprime integers $r$, $s$. Let $p_0,q_0inmathbbZ$ satisfy $p_0s-q_0r=1$. Then it is well-known that all integers $p,q$ satisfying $ps-qr=1$ are $p=p_0+rt$, $q=q_0+st$ with $tinmathbbZ$. Varying $t$ if necessary, we may choose $p_0, q_0$ such that a coset representative $beginbmatrixp_0&q_0\r&sendbmatrix$ in $Gamma_inftybackslash Gamma$ with $-frac12leqfracp_0z+q_0rz+s<frac12$. We have the above sums over $r,s$ with $|t|geq 1$, is convergent since
$$
sum_(r,s)=1 frac1rz+sleq Csum_sfrac1
$$
is convergent where $C>0$ is an absolute constant that depends only on $z$.
We also have that the area of a parallelogram with vertices $0$, $pz+q$, $rz+s$ is the same as that of vertices $0$, $1$, $z$, this is because of $ps-qr=1$. Call this area $C$. Then
$$
|pz+q||rz+s| |sin theta_gamma| = C.
$$
Here $theta_gamma$ is the angle of the corner formed by the three points $pz+q$, $0$, $rz+s$. This angle is also written as $theta_gamma=arg fracpz+qrz+s=arg gamma z$.
Then the convergence of above sum (*) over $r,s$ with $t=0$ follows from that of
$$
sum_gammainGamma_inftybackslashGamma|sin theta_gamma|^k (**)
$$
Lemma 3
Let $Y>0$. We have a constant $C>0$ depending on $z$ such that
$$
# gammainGamma_inftybackslashGamma biggvert leq 4+frac4CY.
$$
This is proved by applying a classical lemma
$$
# gammainGamma_inftybackslashGamma biggvert Im(gamma z)geq Yleq 1+frac CY.
$$
Now, the convergence of (**) follows by partial summation with application of Lemma 3 for $Y=frac1n$, $n=1,2,3,ldots$.
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
add a comment |Â
up vote
1
down vote
up vote
1
down vote
This can be done with Theory of Quadratic Forms. The constants $C$ appearing in this solution may not be the same. We begin with
Definition
Let $ax^2+bxy+cy^2$ be a quadratic form with $a, b, cin mathbbZ$ and $b^2-4ac=D$. We say that a quadratic form $a_1x_1^2+b_1x_1y_1+c_1y_1^2$ with $a_1,b_1,c_1inmathbbZ$ and $b_1^2-4a_1c_1=D$ is equivalent to $ax^2+bxy+cy^2$ if there is $beginbmatrixp & q \ r & s endbmatrixinmathrmSL_2(mathbbZ)$ such that
$$
a_1x_1^2+b_1x_1y_1+c_1y_1^2=a(px_1+qy_1)^2+b(px_1+qy_1)(rx_1+sy_1)+c(rx_1+sy_1)^2.
$$
This can be written again with matrices
$$
beginbmatrixa_1 & fracb_12\ fracb_12 & c_1 endbmatrix = beginbmatrixp & r \ q & s endbmatrix beginbmatrixa & fracb2\ fracb2 & c endbmatrix beginbmatrixp & q \ r & s endbmatrix
$$
and
$$
beginbmatrixx\yendbmatrix=beginbmatrixp & q \ r & s endbmatrixbeginbmatrixx_1\y_1endbmatrix.
$$
The next one is the reducing step.
Lemma 1
Any integer quadratic form with discriminant $D$ is equivalent to $ax^2+bxy+cy^2$ with $|b|leq |a|leq |c|$.
The proof is a repeated application of the matrices in $mathrmSL_2(mathbbZ)$.
$$
beginbmatrix 1 & 1 \ 0 & 1 endbmatrix, beginbmatrix 0 & -1 \ 1 & 0endbmatrix.
$$
Corollary 1
The number of equivalence classes of integer quadratic forms with discriminant $D$ is finite.
If $b^2-4ac=D$ and $|b|leq |a|leq |c|$, then $|D|geq 4|ac|-|b^2| geq 3|ac|$ gives only finitely many choices for $a$ and $c$, subsequently for $b$.
Then it suffices to prove the following for $A=beginbmatrixa& b/2\ b/2&cendbmatrix$ with $b^2-4ac=D$ and $|b|leq |a|leq |c|$, as the sum can be written as a finite sum of the following kind.
The sum is absolutely convergent for $k>1$:
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T A gamma beginbmatrixz\1endbmatrixright)^k.
$$
It suffices to show the following sum is convergent instead, as it differs only by a positive constant.
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T beginbmatrixalpha \ betaendbmatrix beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright)^k
$$
where $alpha=alpha'=2a$, $beta=b+sqrtD$, and $beta'=b-sqrtD$.
We establish the following uniform inequality:
Lemma 2
We have $$
left|beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright|geq C (|pz+q||rz+s|)^1/2
$$
where $C>0$ is an absolute constant depending at most on $a,b,c$, and $gamma=beginbmatrix p&q\r&sendbmatrixinmathrmSL_2(mathbbZ)$.
The proof is by AM-GM inequality.
Now, it suffices to prove the convergence of
$$
sum_gammain mathrmSL_2(mathbbZ)frac1 (*)
$$
Consider $Gamma=mathrmSL_2(mathbbZ)$ and $Gamma_infty=leftpmbeginbmatrix 1&m\0&1endbmatrix, minmathbbZright$.
Fix coprime integers $r$, $s$. Let $p_0,q_0inmathbbZ$ satisfy $p_0s-q_0r=1$. Then it is well-known that all integers $p,q$ satisfying $ps-qr=1$ are $p=p_0+rt$, $q=q_0+st$ with $tinmathbbZ$. Varying $t$ if necessary, we may choose $p_0, q_0$ such that a coset representative $beginbmatrixp_0&q_0\r&sendbmatrix$ in $Gamma_inftybackslash Gamma$ with $-frac12leqfracp_0z+q_0rz+s<frac12$. We have the above sums over $r,s$ with $|t|geq 1$, is convergent since
$$
sum_(r,s)=1 frac1rz+sleq Csum_sfrac1
$$
is convergent where $C>0$ is an absolute constant that depends only on $z$.
We also have that the area of a parallelogram with vertices $0$, $pz+q$, $rz+s$ is the same as that of vertices $0$, $1$, $z$, this is because of $ps-qr=1$. Call this area $C$. Then
$$
|pz+q||rz+s| |sin theta_gamma| = C.
$$
Here $theta_gamma$ is the angle of the corner formed by the three points $pz+q$, $0$, $rz+s$. This angle is also written as $theta_gamma=arg fracpz+qrz+s=arg gamma z$.
Then the convergence of above sum (*) over $r,s$ with $t=0$ follows from that of
$$
sum_gammainGamma_inftybackslashGamma|sin theta_gamma|^k (**)
$$
Lemma 3
Let $Y>0$. We have a constant $C>0$ depending on $z$ such that
$$
# gammainGamma_inftybackslashGamma biggvert leq 4+frac4CY.
$$
This is proved by applying a classical lemma
$$
# gammainGamma_inftybackslashGamma biggvert Im(gamma z)geq Yleq 1+frac CY.
$$
Now, the convergence of (**) follows by partial summation with application of Lemma 3 for $Y=frac1n$, $n=1,2,3,ldots$.
This can be done with Theory of Quadratic Forms. The constants $C$ appearing in this solution may not be the same. We begin with
Definition
Let $ax^2+bxy+cy^2$ be a quadratic form with $a, b, cin mathbbZ$ and $b^2-4ac=D$. We say that a quadratic form $a_1x_1^2+b_1x_1y_1+c_1y_1^2$ with $a_1,b_1,c_1inmathbbZ$ and $b_1^2-4a_1c_1=D$ is equivalent to $ax^2+bxy+cy^2$ if there is $beginbmatrixp & q \ r & s endbmatrixinmathrmSL_2(mathbbZ)$ such that
$$
a_1x_1^2+b_1x_1y_1+c_1y_1^2=a(px_1+qy_1)^2+b(px_1+qy_1)(rx_1+sy_1)+c(rx_1+sy_1)^2.
$$
This can be written again with matrices
$$
beginbmatrixa_1 & fracb_12\ fracb_12 & c_1 endbmatrix = beginbmatrixp & r \ q & s endbmatrix beginbmatrixa & fracb2\ fracb2 & c endbmatrix beginbmatrixp & q \ r & s endbmatrix
$$
and
$$
beginbmatrixx\yendbmatrix=beginbmatrixp & q \ r & s endbmatrixbeginbmatrixx_1\y_1endbmatrix.
$$
The next one is the reducing step.
Lemma 1
Any integer quadratic form with discriminant $D$ is equivalent to $ax^2+bxy+cy^2$ with $|b|leq |a|leq |c|$.
The proof is a repeated application of the matrices in $mathrmSL_2(mathbbZ)$.
$$
beginbmatrix 1 & 1 \ 0 & 1 endbmatrix, beginbmatrix 0 & -1 \ 1 & 0endbmatrix.
$$
Corollary 1
The number of equivalence classes of integer quadratic forms with discriminant $D$ is finite.
If $b^2-4ac=D$ and $|b|leq |a|leq |c|$, then $|D|geq 4|ac|-|b^2| geq 3|ac|$ gives only finitely many choices for $a$ and $c$, subsequently for $b$.
Then it suffices to prove the following for $A=beginbmatrixa& b/2\ b/2&cendbmatrix$ with $b^2-4ac=D$ and $|b|leq |a|leq |c|$, as the sum can be written as a finite sum of the following kind.
The sum is absolutely convergent for $k>1$:
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T A gamma beginbmatrixz\1endbmatrixright)^k.
$$
It suffices to show the following sum is convergent instead, as it differs only by a positive constant.
$$
sum_gammain mathrmSL_2(mathbbZ) frac1left( beginbmatrixz&1endbmatrixgamma^T beginbmatrixalpha \ betaendbmatrix beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright)^k
$$
where $alpha=alpha'=2a$, $beta=b+sqrtD$, and $beta'=b-sqrtD$.
We establish the following uniform inequality:
Lemma 2
We have $$
left|beginbmatrixalpha' &beta'endbmatrix gamma beginbmatrixz\1endbmatrixright|geq C (|pz+q||rz+s|)^1/2
$$
where $C>0$ is an absolute constant depending at most on $a,b,c$, and $gamma=beginbmatrix p&q\r&sendbmatrixinmathrmSL_2(mathbbZ)$.
The proof is by AM-GM inequality.
Now, it suffices to prove the convergence of
$$
sum_gammain mathrmSL_2(mathbbZ)frac1 (*)
$$
Consider $Gamma=mathrmSL_2(mathbbZ)$ and $Gamma_infty=leftpmbeginbmatrix 1&m\0&1endbmatrix, minmathbbZright$.
Fix coprime integers $r$, $s$. Let $p_0,q_0inmathbbZ$ satisfy $p_0s-q_0r=1$. Then it is well-known that all integers $p,q$ satisfying $ps-qr=1$ are $p=p_0+rt$, $q=q_0+st$ with $tinmathbbZ$. Varying $t$ if necessary, we may choose $p_0, q_0$ such that a coset representative $beginbmatrixp_0&q_0\r&sendbmatrix$ in $Gamma_inftybackslash Gamma$ with $-frac12leqfracp_0z+q_0rz+s<frac12$. We have the above sums over $r,s$ with $|t|geq 1$, is convergent since
$$
sum_(r,s)=1 frac1rz+sleq Csum_sfrac1
$$
is convergent where $C>0$ is an absolute constant that depends only on $z$.
We also have that the area of a parallelogram with vertices $0$, $pz+q$, $rz+s$ is the same as that of vertices $0$, $1$, $z$, this is because of $ps-qr=1$. Call this area $C$. Then
$$
|pz+q||rz+s| |sin theta_gamma| = C.
$$
Here $theta_gamma$ is the angle of the corner formed by the three points $pz+q$, $0$, $rz+s$. This angle is also written as $theta_gamma=arg fracpz+qrz+s=arg gamma z$.
Then the convergence of above sum (*) over $r,s$ with $t=0$ follows from that of
$$
sum_gammainGamma_inftybackslashGamma|sin theta_gamma|^k (**)
$$
Lemma 3
Let $Y>0$. We have a constant $C>0$ depending on $z$ such that
$$
# gammainGamma_inftybackslashGamma biggvert leq 4+frac4CY.
$$
This is proved by applying a classical lemma
$$
# gammainGamma_inftybackslashGamma biggvert Im(gamma z)geq Yleq 1+frac CY.
$$
Now, the convergence of (**) follows by partial summation with application of Lemma 3 for $Y=frac1n$, $n=1,2,3,ldots$.
answered 17 hours ago
i707107
11.2k21241
11.2k21241
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
add a comment |Â
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Thank you for your detailed answer. I hope you can answer two more questions for my comprehension. The first is, what is the motivation behind substituting $A$ by $beginbmatrixalpha \ betaendbmatrix * beginbmatrixalpha' &beta'endbmatrix$, since by multiplying them I got $4a*beginbmatrixa & fracb-sqrtD2 \ fracb+sqrtD2 & cendbmatrix$. And $D$ has not to be a square number. Did I make a mistake? The second question is why do we have to inspect the case $t=0$ seperately.
– Deavor
11 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
Since we can interpret quadratic forms as matrices:$$beginbmatrixx¥dbmatrixbeginbmatrixa&b\c&dendbmatrixbeginbmatrixx\yendbmatrix=ax^2+(b+c)xy+dy^2$$, we may have that decomposition. Please postpone accepting, because I found an error in Lemma 2 and need to fix it.
– i707107
9 hours ago
add a comment |Â
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2870328%2fconvergence-of-a-family-of-series%23new-answer', 'question_page');
);
Post as a guest
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
What does $Dequiv 0,1 pmod4$ mean?
– zhw.
Aug 2 at 20:40
$D equiv 0 pmod 4$ or $D equiv 1 pmod 4$ sorry if that was cloudy
– Deavor
Aug 2 at 20:50
You might wanted to put $b^2-4ac$ in two places, one under the sum, the other after your attempted proof.
– i707107
Aug 4 at 19:38
Oh yes thanks for noticing the typo
– Deavor
Aug 4 at 20:18