Mutual information: Proving $I(X;Y)=underset{r(x),s(y)}{min}D_{KL}[p(x,y)midmid r(x)cdot s(y)]$












1












$begingroup$


I was asked to show that for two discrete RVs $X,Y$ with joint distribution $pleft(x,yright)$ and marginal distributions $pleft(xright)$ and $pleft(yright)$, it holds that for any two distributions $rleft(xright)$ and $sleft(yright)$ it holds that



$$Ileft(X;Yright)=underset{rleft(xright),sleft(yright)}{min}D_{KL}left[pleft(x,yright)midmid rleft(xright)cdot sleft(yright)right]$$



Now,



$$Ileft(X;Yright)=D_{KL}left[pleft(x,yright)midmid pleft(xright)cdot pleft(yright)right]=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



so this amounts to showing $$underset{rleft(xright),sleft(yright)}{min}sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{rleft(xright)sleft(yright)}=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



but how does one show this? Do I need to differentiate w.r.t. $rleft(xright)$ and $sleft(yright)$? Is that possible?










share|cite|improve this question











$endgroup$












  • $begingroup$
    This amounts to showing that for every pair of distributions $r,s$ it holds that $$ sum_{x,y} p_{xy} log frac{p_{x}p_{y}}{r_x s_y} ge 0.$$ Breaking the log immediately yields that the LHS is $D(p_X|r) + D(p_Y|s)$ and the inequality is then obvious, as is the equality condition.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 6:11










  • $begingroup$
    @stochasticboy321 I don't follow, how does this inequality yield the requested result?
    $endgroup$
    – H.Rappeport
    Dec 18 '18 at 10:18










  • $begingroup$
    $$D(p_{XY}|r_X s_Y) - D(p_{XY}|p_Xp_Y) = sum p_{xy} log frac{p_{xy}}{r_x s_y} - sum p_{xy} log frac{p_{xy}}{p_x p_y}$$ equals the expression in the comment. So, the inequality shows that $D(p_{XY}|r_X s_Y) ge I(X;Y)$ for every $(r,s).$ I suppose you do have to show that $I$ is the minimum, but this is trivial - just set $r,s$ to be the appropriate marginals.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 20:30










  • $begingroup$
    @stochasticboy321 I see, thanks! Do you want to post this as an answer? I'd be happy to accept.
    $endgroup$
    – H.Rappeport
    Dec 19 '18 at 20:13


















1












$begingroup$


I was asked to show that for two discrete RVs $X,Y$ with joint distribution $pleft(x,yright)$ and marginal distributions $pleft(xright)$ and $pleft(yright)$, it holds that for any two distributions $rleft(xright)$ and $sleft(yright)$ it holds that



$$Ileft(X;Yright)=underset{rleft(xright),sleft(yright)}{min}D_{KL}left[pleft(x,yright)midmid rleft(xright)cdot sleft(yright)right]$$



Now,



$$Ileft(X;Yright)=D_{KL}left[pleft(x,yright)midmid pleft(xright)cdot pleft(yright)right]=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



so this amounts to showing $$underset{rleft(xright),sleft(yright)}{min}sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{rleft(xright)sleft(yright)}=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



but how does one show this? Do I need to differentiate w.r.t. $rleft(xright)$ and $sleft(yright)$? Is that possible?










share|cite|improve this question











$endgroup$












  • $begingroup$
    This amounts to showing that for every pair of distributions $r,s$ it holds that $$ sum_{x,y} p_{xy} log frac{p_{x}p_{y}}{r_x s_y} ge 0.$$ Breaking the log immediately yields that the LHS is $D(p_X|r) + D(p_Y|s)$ and the inequality is then obvious, as is the equality condition.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 6:11










  • $begingroup$
    @stochasticboy321 I don't follow, how does this inequality yield the requested result?
    $endgroup$
    – H.Rappeport
    Dec 18 '18 at 10:18










  • $begingroup$
    $$D(p_{XY}|r_X s_Y) - D(p_{XY}|p_Xp_Y) = sum p_{xy} log frac{p_{xy}}{r_x s_y} - sum p_{xy} log frac{p_{xy}}{p_x p_y}$$ equals the expression in the comment. So, the inequality shows that $D(p_{XY}|r_X s_Y) ge I(X;Y)$ for every $(r,s).$ I suppose you do have to show that $I$ is the minimum, but this is trivial - just set $r,s$ to be the appropriate marginals.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 20:30










  • $begingroup$
    @stochasticboy321 I see, thanks! Do you want to post this as an answer? I'd be happy to accept.
    $endgroup$
    – H.Rappeport
    Dec 19 '18 at 20:13
















1












1








1





$begingroup$


I was asked to show that for two discrete RVs $X,Y$ with joint distribution $pleft(x,yright)$ and marginal distributions $pleft(xright)$ and $pleft(yright)$, it holds that for any two distributions $rleft(xright)$ and $sleft(yright)$ it holds that



$$Ileft(X;Yright)=underset{rleft(xright),sleft(yright)}{min}D_{KL}left[pleft(x,yright)midmid rleft(xright)cdot sleft(yright)right]$$



Now,



$$Ileft(X;Yright)=D_{KL}left[pleft(x,yright)midmid pleft(xright)cdot pleft(yright)right]=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



so this amounts to showing $$underset{rleft(xright),sleft(yright)}{min}sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{rleft(xright)sleft(yright)}=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



but how does one show this? Do I need to differentiate w.r.t. $rleft(xright)$ and $sleft(yright)$? Is that possible?










share|cite|improve this question











$endgroup$




I was asked to show that for two discrete RVs $X,Y$ with joint distribution $pleft(x,yright)$ and marginal distributions $pleft(xright)$ and $pleft(yright)$, it holds that for any two distributions $rleft(xright)$ and $sleft(yright)$ it holds that



$$Ileft(X;Yright)=underset{rleft(xright),sleft(yright)}{min}D_{KL}left[pleft(x,yright)midmid rleft(xright)cdot sleft(yright)right]$$



Now,



$$Ileft(X;Yright)=D_{KL}left[pleft(x,yright)midmid pleft(xright)cdot pleft(yright)right]=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



so this amounts to showing $$underset{rleft(xright),sleft(yright)}{min}sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{rleft(xright)sleft(yright)}=sum_{x,y}pleft(x,yright)logfrac{pleft(x,yright)}{pleft(xright)pleft(yright)}$$



but how does one show this? Do I need to differentiate w.r.t. $rleft(xright)$ and $sleft(yright)$? Is that possible?







probability-theory derivatives information-theory






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Dec 17 '18 at 21:32







H.Rappeport

















asked Dec 17 '18 at 13:46









H.RappeportH.Rappeport

6801510




6801510












  • $begingroup$
    This amounts to showing that for every pair of distributions $r,s$ it holds that $$ sum_{x,y} p_{xy} log frac{p_{x}p_{y}}{r_x s_y} ge 0.$$ Breaking the log immediately yields that the LHS is $D(p_X|r) + D(p_Y|s)$ and the inequality is then obvious, as is the equality condition.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 6:11










  • $begingroup$
    @stochasticboy321 I don't follow, how does this inequality yield the requested result?
    $endgroup$
    – H.Rappeport
    Dec 18 '18 at 10:18










  • $begingroup$
    $$D(p_{XY}|r_X s_Y) - D(p_{XY}|p_Xp_Y) = sum p_{xy} log frac{p_{xy}}{r_x s_y} - sum p_{xy} log frac{p_{xy}}{p_x p_y}$$ equals the expression in the comment. So, the inequality shows that $D(p_{XY}|r_X s_Y) ge I(X;Y)$ for every $(r,s).$ I suppose you do have to show that $I$ is the minimum, but this is trivial - just set $r,s$ to be the appropriate marginals.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 20:30










  • $begingroup$
    @stochasticboy321 I see, thanks! Do you want to post this as an answer? I'd be happy to accept.
    $endgroup$
    – H.Rappeport
    Dec 19 '18 at 20:13




















  • $begingroup$
    This amounts to showing that for every pair of distributions $r,s$ it holds that $$ sum_{x,y} p_{xy} log frac{p_{x}p_{y}}{r_x s_y} ge 0.$$ Breaking the log immediately yields that the LHS is $D(p_X|r) + D(p_Y|s)$ and the inequality is then obvious, as is the equality condition.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 6:11










  • $begingroup$
    @stochasticboy321 I don't follow, how does this inequality yield the requested result?
    $endgroup$
    – H.Rappeport
    Dec 18 '18 at 10:18










  • $begingroup$
    $$D(p_{XY}|r_X s_Y) - D(p_{XY}|p_Xp_Y) = sum p_{xy} log frac{p_{xy}}{r_x s_y} - sum p_{xy} log frac{p_{xy}}{p_x p_y}$$ equals the expression in the comment. So, the inequality shows that $D(p_{XY}|r_X s_Y) ge I(X;Y)$ for every $(r,s).$ I suppose you do have to show that $I$ is the minimum, but this is trivial - just set $r,s$ to be the appropriate marginals.
    $endgroup$
    – stochasticboy321
    Dec 18 '18 at 20:30










  • $begingroup$
    @stochasticboy321 I see, thanks! Do you want to post this as an answer? I'd be happy to accept.
    $endgroup$
    – H.Rappeport
    Dec 19 '18 at 20:13


















$begingroup$
This amounts to showing that for every pair of distributions $r,s$ it holds that $$ sum_{x,y} p_{xy} log frac{p_{x}p_{y}}{r_x s_y} ge 0.$$ Breaking the log immediately yields that the LHS is $D(p_X|r) + D(p_Y|s)$ and the inequality is then obvious, as is the equality condition.
$endgroup$
– stochasticboy321
Dec 18 '18 at 6:11




$begingroup$
This amounts to showing that for every pair of distributions $r,s$ it holds that $$ sum_{x,y} p_{xy} log frac{p_{x}p_{y}}{r_x s_y} ge 0.$$ Breaking the log immediately yields that the LHS is $D(p_X|r) + D(p_Y|s)$ and the inequality is then obvious, as is the equality condition.
$endgroup$
– stochasticboy321
Dec 18 '18 at 6:11












$begingroup$
@stochasticboy321 I don't follow, how does this inequality yield the requested result?
$endgroup$
– H.Rappeport
Dec 18 '18 at 10:18




$begingroup$
@stochasticboy321 I don't follow, how does this inequality yield the requested result?
$endgroup$
– H.Rappeport
Dec 18 '18 at 10:18












$begingroup$
$$D(p_{XY}|r_X s_Y) - D(p_{XY}|p_Xp_Y) = sum p_{xy} log frac{p_{xy}}{r_x s_y} - sum p_{xy} log frac{p_{xy}}{p_x p_y}$$ equals the expression in the comment. So, the inequality shows that $D(p_{XY}|r_X s_Y) ge I(X;Y)$ for every $(r,s).$ I suppose you do have to show that $I$ is the minimum, but this is trivial - just set $r,s$ to be the appropriate marginals.
$endgroup$
– stochasticboy321
Dec 18 '18 at 20:30




$begingroup$
$$D(p_{XY}|r_X s_Y) - D(p_{XY}|p_Xp_Y) = sum p_{xy} log frac{p_{xy}}{r_x s_y} - sum p_{xy} log frac{p_{xy}}{p_x p_y}$$ equals the expression in the comment. So, the inequality shows that $D(p_{XY}|r_X s_Y) ge I(X;Y)$ for every $(r,s).$ I suppose you do have to show that $I$ is the minimum, but this is trivial - just set $r,s$ to be the appropriate marginals.
$endgroup$
– stochasticboy321
Dec 18 '18 at 20:30












$begingroup$
@stochasticboy321 I see, thanks! Do you want to post this as an answer? I'd be happy to accept.
$endgroup$
– H.Rappeport
Dec 19 '18 at 20:13






$begingroup$
@stochasticboy321 I see, thanks! Do you want to post this as an answer? I'd be happy to accept.
$endgroup$
– H.Rappeport
Dec 19 '18 at 20:13












0






active

oldest

votes











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%2f3043954%2fmutual-information-proving-ixy-undersetrx-sy-mind-klpx-y-mid%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























0






active

oldest

votes








0






active

oldest

votes









active

oldest

votes






active

oldest

votes
















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%2f3043954%2fmutual-information-proving-ixy-undersetrx-sy-mind-klpx-y-mid%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