Which inequality for the fourth central moment do we need to apply here?












6












$begingroup$


Let





  • $dinmathbb N$ with $d>1$


  • $lambda^d$ denote the Lebesuge measure on $mathcal Bleft(mathbb R^dright)$


  • $fin C^2(mathbb R)$ be positive and $$pi(x):=prod_{i=1}^df(x_i);;;text{for }xinmathbb R^d$$


  • $(Omega,mathcal A,operatorname P)$ be a probability space


  • $X:Omegatomathbb R^d$ with $X_astoperatorname P=pilambda^d$


Now, let $$g(x):=frac1{d-1}sum_{i=2}^dleft|frac{f'(x_i)}{f(x_i)}right|^2;;;text{for }xinmathbb R^d.$$



Assume $$M:=intfrac{|f'|^8}{f^7}:{rm d}lambda^1<inftytag1.$$



Note that $$operatorname Eleft[g(X)right]=intfrac{left|f'right|^2}f:{rm d}lambda^1=:I.tag2$$




I want to show that $$operatorname Eleft[left|g(X)-Iright|^4right]le d^{-frac12}(d-1)^{-frac32}3Mtag3.$$




Is there an easy estimate which yields $(3)$? Clearly, we can expand the left-hand side using the multinomial theorem, but then we deal with a complicated expression and annoying computations.



On the other hand, by applying the Cauchy-Schwarz inequality twice, we obtain $$operatorname Eleft[left|g(X)-Iright|^4right]lefrac1{d-1}sum_{i=2}^doperatorname Eleft[left|left|frac{f'(X)}{f(X)}right|^2-Iright|^4right],$$ but I don't know how we need to proceed from here.










share|cite|improve this question











$endgroup$












  • $begingroup$
    What is the source of this problem?
    $endgroup$
    – LoveTooNap29
    Dec 9 '18 at 22:20






  • 1




    $begingroup$
    @LoveTooNap29 You can find it in the proof of Lemma 2.1 here: projecteuclid.org/euclid.aoap/1034625254
    $endgroup$
    – 0xbadf00d
    Dec 9 '18 at 22:27










  • $begingroup$
    @0xbadf00d Have you had a look at my answer? :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:01
















6












$begingroup$


Let





  • $dinmathbb N$ with $d>1$


  • $lambda^d$ denote the Lebesuge measure on $mathcal Bleft(mathbb R^dright)$


  • $fin C^2(mathbb R)$ be positive and $$pi(x):=prod_{i=1}^df(x_i);;;text{for }xinmathbb R^d$$


  • $(Omega,mathcal A,operatorname P)$ be a probability space


  • $X:Omegatomathbb R^d$ with $X_astoperatorname P=pilambda^d$


Now, let $$g(x):=frac1{d-1}sum_{i=2}^dleft|frac{f'(x_i)}{f(x_i)}right|^2;;;text{for }xinmathbb R^d.$$



Assume $$M:=intfrac{|f'|^8}{f^7}:{rm d}lambda^1<inftytag1.$$



Note that $$operatorname Eleft[g(X)right]=intfrac{left|f'right|^2}f:{rm d}lambda^1=:I.tag2$$




I want to show that $$operatorname Eleft[left|g(X)-Iright|^4right]le d^{-frac12}(d-1)^{-frac32}3Mtag3.$$




Is there an easy estimate which yields $(3)$? Clearly, we can expand the left-hand side using the multinomial theorem, but then we deal with a complicated expression and annoying computations.



On the other hand, by applying the Cauchy-Schwarz inequality twice, we obtain $$operatorname Eleft[left|g(X)-Iright|^4right]lefrac1{d-1}sum_{i=2}^doperatorname Eleft[left|left|frac{f'(X)}{f(X)}right|^2-Iright|^4right],$$ but I don't know how we need to proceed from here.










share|cite|improve this question











$endgroup$












  • $begingroup$
    What is the source of this problem?
    $endgroup$
    – LoveTooNap29
    Dec 9 '18 at 22:20






  • 1




    $begingroup$
    @LoveTooNap29 You can find it in the proof of Lemma 2.1 here: projecteuclid.org/euclid.aoap/1034625254
    $endgroup$
    – 0xbadf00d
    Dec 9 '18 at 22:27










  • $begingroup$
    @0xbadf00d Have you had a look at my answer? :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:01














6












6








6


4



$begingroup$


Let





  • $dinmathbb N$ with $d>1$


  • $lambda^d$ denote the Lebesuge measure on $mathcal Bleft(mathbb R^dright)$


  • $fin C^2(mathbb R)$ be positive and $$pi(x):=prod_{i=1}^df(x_i);;;text{for }xinmathbb R^d$$


  • $(Omega,mathcal A,operatorname P)$ be a probability space


  • $X:Omegatomathbb R^d$ with $X_astoperatorname P=pilambda^d$


Now, let $$g(x):=frac1{d-1}sum_{i=2}^dleft|frac{f'(x_i)}{f(x_i)}right|^2;;;text{for }xinmathbb R^d.$$



Assume $$M:=intfrac{|f'|^8}{f^7}:{rm d}lambda^1<inftytag1.$$



Note that $$operatorname Eleft[g(X)right]=intfrac{left|f'right|^2}f:{rm d}lambda^1=:I.tag2$$




I want to show that $$operatorname Eleft[left|g(X)-Iright|^4right]le d^{-frac12}(d-1)^{-frac32}3Mtag3.$$




Is there an easy estimate which yields $(3)$? Clearly, we can expand the left-hand side using the multinomial theorem, but then we deal with a complicated expression and annoying computations.



On the other hand, by applying the Cauchy-Schwarz inequality twice, we obtain $$operatorname Eleft[left|g(X)-Iright|^4right]lefrac1{d-1}sum_{i=2}^doperatorname Eleft[left|left|frac{f'(X)}{f(X)}right|^2-Iright|^4right],$$ but I don't know how we need to proceed from here.










share|cite|improve this question











$endgroup$




Let





  • $dinmathbb N$ with $d>1$


  • $lambda^d$ denote the Lebesuge measure on $mathcal Bleft(mathbb R^dright)$


  • $fin C^2(mathbb R)$ be positive and $$pi(x):=prod_{i=1}^df(x_i);;;text{for }xinmathbb R^d$$


  • $(Omega,mathcal A,operatorname P)$ be a probability space


  • $X:Omegatomathbb R^d$ with $X_astoperatorname P=pilambda^d$


Now, let $$g(x):=frac1{d-1}sum_{i=2}^dleft|frac{f'(x_i)}{f(x_i)}right|^2;;;text{for }xinmathbb R^d.$$



Assume $$M:=intfrac{|f'|^8}{f^7}:{rm d}lambda^1<inftytag1.$$



Note that $$operatorname Eleft[g(X)right]=intfrac{left|f'right|^2}f:{rm d}lambda^1=:I.tag2$$




I want to show that $$operatorname Eleft[left|g(X)-Iright|^4right]le d^{-frac12}(d-1)^{-frac32}3Mtag3.$$




Is there an easy estimate which yields $(3)$? Clearly, we can expand the left-hand side using the multinomial theorem, but then we deal with a complicated expression and annoying computations.



On the other hand, by applying the Cauchy-Schwarz inequality twice, we obtain $$operatorname Eleft[left|g(X)-Iright|^4right]lefrac1{d-1}sum_{i=2}^doperatorname Eleft[left|left|frac{f'(X)}{f(X)}right|^2-Iright|^4right],$$ but I don't know how we need to proceed from here.







probability-theory inequality expected-value






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Dec 11 '18 at 14:39







0xbadf00d

















asked Oct 31 '18 at 10:18









0xbadf00d0xbadf00d

1,91341531




1,91341531












  • $begingroup$
    What is the source of this problem?
    $endgroup$
    – LoveTooNap29
    Dec 9 '18 at 22:20






  • 1




    $begingroup$
    @LoveTooNap29 You can find it in the proof of Lemma 2.1 here: projecteuclid.org/euclid.aoap/1034625254
    $endgroup$
    – 0xbadf00d
    Dec 9 '18 at 22:27










  • $begingroup$
    @0xbadf00d Have you had a look at my answer? :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:01


















  • $begingroup$
    What is the source of this problem?
    $endgroup$
    – LoveTooNap29
    Dec 9 '18 at 22:20






  • 1




    $begingroup$
    @LoveTooNap29 You can find it in the proof of Lemma 2.1 here: projecteuclid.org/euclid.aoap/1034625254
    $endgroup$
    – 0xbadf00d
    Dec 9 '18 at 22:27










  • $begingroup$
    @0xbadf00d Have you had a look at my answer? :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:01
















$begingroup$
What is the source of this problem?
$endgroup$
– LoveTooNap29
Dec 9 '18 at 22:20




$begingroup$
What is the source of this problem?
$endgroup$
– LoveTooNap29
Dec 9 '18 at 22:20




1




1




$begingroup$
@LoveTooNap29 You can find it in the proof of Lemma 2.1 here: projecteuclid.org/euclid.aoap/1034625254
$endgroup$
– 0xbadf00d
Dec 9 '18 at 22:27




$begingroup$
@LoveTooNap29 You can find it in the proof of Lemma 2.1 here: projecteuclid.org/euclid.aoap/1034625254
$endgroup$
– 0xbadf00d
Dec 9 '18 at 22:27












$begingroup$
@0xbadf00d Have you had a look at my answer? :)
$endgroup$
– Roberto Rastapopoulos
Dec 15 '18 at 13:01




$begingroup$
@0xbadf00d Have you had a look at my answer? :)
$endgroup$
– Roberto Rastapopoulos
Dec 15 '18 at 13:01










1 Answer
1






active

oldest

votes


















2





+25







$begingroup$

Let us write $h = (f'/f)^2$.
By the definitions of $I$ and $M$, $E[h(X)] = I$ and $E[h^4(X)] = M$,
and we have
$$ g(x) = frac{1}{d-1} sum_{i=2}^{d} h(x_i).$$
Now we have
begin{align}
E left[ |g(X) - I|^4 right] &= E left[ left( frac{1}{d-1} sum_{i=2}^{d} (h(X_i) - I) right)^4 right] \
&= left( frac{1}{d-1} right)^4 E left[ left(sum_{i=2}^{d} (h(X_i) - I)^4 right) + 3 left( sum_{i=2}^{d} sum_{j=2, jneq i}^{d} (h(X_i) - I)^2(h(X_j) - I)^2 right) right],
end{align}

because the other terms of the product cancel out from the fact that $E(h(X_i) - I) = 0$.
Using the moment bound at our disposal and Holder's inequality,
we obtain
begin{align}
E left[ |g(X) - I|^4 right] &leq M , left( frac{1}{d-1} right)^4 left(d - 1 + 3 (d-1)(d-2) right), \
&= 3M , frac{d-5/3}{(d-1)^3} = 3 M , (d-1)^{-3/2} , frac{d - 5/3}{(d-1)^{3/2}}.
end{align}

and the last term is less than $d^{-1/2}$ when $d geq 2$.





EDIT: the fact that $E[(h(X_i) - I)^4] leq E[h(X_i)^4]$ follows from the non-negativity of $h$, see my answer here.






share|cite|improve this answer











$endgroup$













  • $begingroup$
    The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
    $endgroup$
    – Dap
    Dec 14 '18 at 14:06










  • $begingroup$
    @Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:00











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%2f2978938%2fwhich-inequality-for-the-fourth-central-moment-do-we-need-to-apply-here%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























1 Answer
1






active

oldest

votes








1 Answer
1






active

oldest

votes









active

oldest

votes






active

oldest

votes









2





+25







$begingroup$

Let us write $h = (f'/f)^2$.
By the definitions of $I$ and $M$, $E[h(X)] = I$ and $E[h^4(X)] = M$,
and we have
$$ g(x) = frac{1}{d-1} sum_{i=2}^{d} h(x_i).$$
Now we have
begin{align}
E left[ |g(X) - I|^4 right] &= E left[ left( frac{1}{d-1} sum_{i=2}^{d} (h(X_i) - I) right)^4 right] \
&= left( frac{1}{d-1} right)^4 E left[ left(sum_{i=2}^{d} (h(X_i) - I)^4 right) + 3 left( sum_{i=2}^{d} sum_{j=2, jneq i}^{d} (h(X_i) - I)^2(h(X_j) - I)^2 right) right],
end{align}

because the other terms of the product cancel out from the fact that $E(h(X_i) - I) = 0$.
Using the moment bound at our disposal and Holder's inequality,
we obtain
begin{align}
E left[ |g(X) - I|^4 right] &leq M , left( frac{1}{d-1} right)^4 left(d - 1 + 3 (d-1)(d-2) right), \
&= 3M , frac{d-5/3}{(d-1)^3} = 3 M , (d-1)^{-3/2} , frac{d - 5/3}{(d-1)^{3/2}}.
end{align}

and the last term is less than $d^{-1/2}$ when $d geq 2$.





EDIT: the fact that $E[(h(X_i) - I)^4] leq E[h(X_i)^4]$ follows from the non-negativity of $h$, see my answer here.






share|cite|improve this answer











$endgroup$













  • $begingroup$
    The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
    $endgroup$
    – Dap
    Dec 14 '18 at 14:06










  • $begingroup$
    @Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:00
















2





+25







$begingroup$

Let us write $h = (f'/f)^2$.
By the definitions of $I$ and $M$, $E[h(X)] = I$ and $E[h^4(X)] = M$,
and we have
$$ g(x) = frac{1}{d-1} sum_{i=2}^{d} h(x_i).$$
Now we have
begin{align}
E left[ |g(X) - I|^4 right] &= E left[ left( frac{1}{d-1} sum_{i=2}^{d} (h(X_i) - I) right)^4 right] \
&= left( frac{1}{d-1} right)^4 E left[ left(sum_{i=2}^{d} (h(X_i) - I)^4 right) + 3 left( sum_{i=2}^{d} sum_{j=2, jneq i}^{d} (h(X_i) - I)^2(h(X_j) - I)^2 right) right],
end{align}

because the other terms of the product cancel out from the fact that $E(h(X_i) - I) = 0$.
Using the moment bound at our disposal and Holder's inequality,
we obtain
begin{align}
E left[ |g(X) - I|^4 right] &leq M , left( frac{1}{d-1} right)^4 left(d - 1 + 3 (d-1)(d-2) right), \
&= 3M , frac{d-5/3}{(d-1)^3} = 3 M , (d-1)^{-3/2} , frac{d - 5/3}{(d-1)^{3/2}}.
end{align}

and the last term is less than $d^{-1/2}$ when $d geq 2$.





EDIT: the fact that $E[(h(X_i) - I)^4] leq E[h(X_i)^4]$ follows from the non-negativity of $h$, see my answer here.






share|cite|improve this answer











$endgroup$













  • $begingroup$
    The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
    $endgroup$
    – Dap
    Dec 14 '18 at 14:06










  • $begingroup$
    @Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:00














2





+25







2





+25



2




+25



$begingroup$

Let us write $h = (f'/f)^2$.
By the definitions of $I$ and $M$, $E[h(X)] = I$ and $E[h^4(X)] = M$,
and we have
$$ g(x) = frac{1}{d-1} sum_{i=2}^{d} h(x_i).$$
Now we have
begin{align}
E left[ |g(X) - I|^4 right] &= E left[ left( frac{1}{d-1} sum_{i=2}^{d} (h(X_i) - I) right)^4 right] \
&= left( frac{1}{d-1} right)^4 E left[ left(sum_{i=2}^{d} (h(X_i) - I)^4 right) + 3 left( sum_{i=2}^{d} sum_{j=2, jneq i}^{d} (h(X_i) - I)^2(h(X_j) - I)^2 right) right],
end{align}

because the other terms of the product cancel out from the fact that $E(h(X_i) - I) = 0$.
Using the moment bound at our disposal and Holder's inequality,
we obtain
begin{align}
E left[ |g(X) - I|^4 right] &leq M , left( frac{1}{d-1} right)^4 left(d - 1 + 3 (d-1)(d-2) right), \
&= 3M , frac{d-5/3}{(d-1)^3} = 3 M , (d-1)^{-3/2} , frac{d - 5/3}{(d-1)^{3/2}}.
end{align}

and the last term is less than $d^{-1/2}$ when $d geq 2$.





EDIT: the fact that $E[(h(X_i) - I)^4] leq E[h(X_i)^4]$ follows from the non-negativity of $h$, see my answer here.






share|cite|improve this answer











$endgroup$



Let us write $h = (f'/f)^2$.
By the definitions of $I$ and $M$, $E[h(X)] = I$ and $E[h^4(X)] = M$,
and we have
$$ g(x) = frac{1}{d-1} sum_{i=2}^{d} h(x_i).$$
Now we have
begin{align}
E left[ |g(X) - I|^4 right] &= E left[ left( frac{1}{d-1} sum_{i=2}^{d} (h(X_i) - I) right)^4 right] \
&= left( frac{1}{d-1} right)^4 E left[ left(sum_{i=2}^{d} (h(X_i) - I)^4 right) + 3 left( sum_{i=2}^{d} sum_{j=2, jneq i}^{d} (h(X_i) - I)^2(h(X_j) - I)^2 right) right],
end{align}

because the other terms of the product cancel out from the fact that $E(h(X_i) - I) = 0$.
Using the moment bound at our disposal and Holder's inequality,
we obtain
begin{align}
E left[ |g(X) - I|^4 right] &leq M , left( frac{1}{d-1} right)^4 left(d - 1 + 3 (d-1)(d-2) right), \
&= 3M , frac{d-5/3}{(d-1)^3} = 3 M , (d-1)^{-3/2} , frac{d - 5/3}{(d-1)^{3/2}}.
end{align}

and the last term is less than $d^{-1/2}$ when $d geq 2$.





EDIT: the fact that $E[(h(X_i) - I)^4] leq E[h(X_i)^4]$ follows from the non-negativity of $h$, see my answer here.







share|cite|improve this answer














share|cite|improve this answer



share|cite|improve this answer








edited Dec 15 '18 at 18:45

























answered Dec 13 '18 at 19:48









Roberto RastapopoulosRoberto Rastapopoulos

899424




899424












  • $begingroup$
    The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
    $endgroup$
    – Dap
    Dec 14 '18 at 14:06










  • $begingroup$
    @Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:00


















  • $begingroup$
    The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
    $endgroup$
    – Dap
    Dec 14 '18 at 14:06










  • $begingroup$
    @Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
    $endgroup$
    – Roberto Rastapopoulos
    Dec 15 '18 at 13:00
















$begingroup$
The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
$endgroup$
– Dap
Dec 14 '18 at 14:06




$begingroup$
The bound after "we obtain" seems to require $E[(h(X_i)-I)^4]leq E[h(X_i)^4],$ unless I've missed something, but this is not true in general: math.stackexchange.com/a/1591395.
$endgroup$
– Dap
Dec 14 '18 at 14:06












$begingroup$
@Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
$endgroup$
– Roberto Rastapopoulos
Dec 15 '18 at 13:00




$begingroup$
@Dap I think I proved the moment bound. Let me know if you see any mistakes. :)
$endgroup$
– Roberto Rastapopoulos
Dec 15 '18 at 13:00


















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%2f2978938%2fwhich-inequality-for-the-fourth-central-moment-do-we-need-to-apply-here%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

Probability when a professor distributes a quiz and homework assignment to a class of n students.

Aardman Animations

Are they similar matrix