Prove that $langlemathbf{A}, mathbf{C}rangle leq delta$ equals with $|mathbf{A}|_*leqdelta$












5












$begingroup$


Given an arbitrary matrix $mathbf{A}in R^{ntimes n}$ and the basis matrix set $mathbb{S}={mathbf{C}in R^{ntimes n}: mathbf{C}^Tmathbf{C}=mathbf{I}_n}$. Then how to prove:



1:If we have $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$ then it holds that $|mathbf{A}|_*leqdelta$.



2:Also, if we have $|mathbf{A}|_*leqdelta$, then we obtain $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$.



In the question, $|mathbf{A}|_*$ is the nuclear norm, and $langlemathbf{A}, mathbf{C}rangle = operatorname{Tr} (mathbf{A}^Tmathbf{C})$.










share|cite|improve this question











$endgroup$












  • $begingroup$
    @OmG, nuclear norm, see the last line of the question
    $endgroup$
    – pointguard0
    Jan 21 at 9:17


















5












$begingroup$


Given an arbitrary matrix $mathbf{A}in R^{ntimes n}$ and the basis matrix set $mathbb{S}={mathbf{C}in R^{ntimes n}: mathbf{C}^Tmathbf{C}=mathbf{I}_n}$. Then how to prove:



1:If we have $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$ then it holds that $|mathbf{A}|_*leqdelta$.



2:Also, if we have $|mathbf{A}|_*leqdelta$, then we obtain $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$.



In the question, $|mathbf{A}|_*$ is the nuclear norm, and $langlemathbf{A}, mathbf{C}rangle = operatorname{Tr} (mathbf{A}^Tmathbf{C})$.










share|cite|improve this question











$endgroup$












  • $begingroup$
    @OmG, nuclear norm, see the last line of the question
    $endgroup$
    – pointguard0
    Jan 21 at 9:17
















5












5








5


1



$begingroup$


Given an arbitrary matrix $mathbf{A}in R^{ntimes n}$ and the basis matrix set $mathbb{S}={mathbf{C}in R^{ntimes n}: mathbf{C}^Tmathbf{C}=mathbf{I}_n}$. Then how to prove:



1:If we have $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$ then it holds that $|mathbf{A}|_*leqdelta$.



2:Also, if we have $|mathbf{A}|_*leqdelta$, then we obtain $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$.



In the question, $|mathbf{A}|_*$ is the nuclear norm, and $langlemathbf{A}, mathbf{C}rangle = operatorname{Tr} (mathbf{A}^Tmathbf{C})$.










share|cite|improve this question











$endgroup$




Given an arbitrary matrix $mathbf{A}in R^{ntimes n}$ and the basis matrix set $mathbb{S}={mathbf{C}in R^{ntimes n}: mathbf{C}^Tmathbf{C}=mathbf{I}_n}$. Then how to prove:



1:If we have $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$ then it holds that $|mathbf{A}|_*leqdelta$.



2:Also, if we have $|mathbf{A}|_*leqdelta$, then we obtain $langlemathbf{A}, mathbf{C}rangle leq delta$ for all $mathbf{C}subsetmathbb{S}$.



In the question, $|mathbf{A}|_*$ is the nuclear norm, and $langlemathbf{A}, mathbf{C}rangle = operatorname{Tr} (mathbf{A}^Tmathbf{C})$.







calculus linear-algebra matrices






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Jan 22 at 12:17







olivia

















asked Jan 21 at 9:06









oliviaolivia

791616




791616












  • $begingroup$
    @OmG, nuclear norm, see the last line of the question
    $endgroup$
    – pointguard0
    Jan 21 at 9:17




















  • $begingroup$
    @OmG, nuclear norm, see the last line of the question
    $endgroup$
    – pointguard0
    Jan 21 at 9:17


















$begingroup$
@OmG, nuclear norm, see the last line of the question
$endgroup$
– pointguard0
Jan 21 at 9:17






$begingroup$
@OmG, nuclear norm, see the last line of the question
$endgroup$
– pointguard0
Jan 21 at 9:17












1 Answer
1






active

oldest

votes


















4












$begingroup$

We can use the polar decomposition
$$
A=Psqrt{A^*A}
$$
where $sqrt{A^*A}$ is positive semi-definite and $P$ is orthogonal, i.e. $PP^T=P^TP=I$. Note that $text{tr}(AB)=text{tr}(BA)$ and $text{tr}(A)=text{tr}(A^T)$ hold for any $ntimes n$ real matrices $A,B$.



1. We can take $B=P$ and it follows
$$
|A|_*=text{tr}(sqrt{A^*A})=text{tr}(B^TPsqrt{A^*A})=text{tr}(Psqrt{A^*A}B^T)=text{tr}(AB^T)=langle A,Brangle le delta.
$$



2. We have $$begin{eqnarray}langle A,Brangle &=&text{tr}(AB^T)\&=&text{tr}(Psqrt{A^*A}B^T)\&=&text{tr}(B^TPsqrt{A^*A})=text{tr}(tilde{B}sqrt{A^*A})end{eqnarray}$$ where $tilde{B}=B^TP$ is an orthogonal matrix. Since $(I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)$ is positive semi-definite, we have
$$
text{tr}left((I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)right)ge 0.
$$
This leads to
$$
text{tr}left(tilde{B}sqrt{A^*A}right)+text{tr}left(sqrt{A^*A}tilde{B}^Tright)letext{tr}left(sqrt{A^*A}right)+text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright).
$$
Since $$text{tr}left(sqrt{A^*A}tilde{B}^Tright)=text{tr}left((tilde{B}^T)^Tsqrt{A^*A}^Tright)=text{tr}left(tilde{B}sqrt{A^*A}right)$$ and $$text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright)=text{tr}left(tilde{B}^Ttilde{B}sqrt{A^*A}right)=text{tr}left(sqrt{A^*A}right),$$
it follows$$
langle A,Brangle=text{tr}left(sqrt{A^*A}tilde{B}^Tright)le text{tr}left(sqrt{A^*A}right)=|A|_*le delta.
$$






share|cite|improve this answer









$endgroup$













    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%2f3081655%2fprove-that-langle-mathbfa-mathbfc-rangle-leq-delta-equals-with-ma%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









    4












    $begingroup$

    We can use the polar decomposition
    $$
    A=Psqrt{A^*A}
    $$
    where $sqrt{A^*A}$ is positive semi-definite and $P$ is orthogonal, i.e. $PP^T=P^TP=I$. Note that $text{tr}(AB)=text{tr}(BA)$ and $text{tr}(A)=text{tr}(A^T)$ hold for any $ntimes n$ real matrices $A,B$.



    1. We can take $B=P$ and it follows
    $$
    |A|_*=text{tr}(sqrt{A^*A})=text{tr}(B^TPsqrt{A^*A})=text{tr}(Psqrt{A^*A}B^T)=text{tr}(AB^T)=langle A,Brangle le delta.
    $$



    2. We have $$begin{eqnarray}langle A,Brangle &=&text{tr}(AB^T)\&=&text{tr}(Psqrt{A^*A}B^T)\&=&text{tr}(B^TPsqrt{A^*A})=text{tr}(tilde{B}sqrt{A^*A})end{eqnarray}$$ where $tilde{B}=B^TP$ is an orthogonal matrix. Since $(I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)$ is positive semi-definite, we have
    $$
    text{tr}left((I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)right)ge 0.
    $$
    This leads to
    $$
    text{tr}left(tilde{B}sqrt{A^*A}right)+text{tr}left(sqrt{A^*A}tilde{B}^Tright)letext{tr}left(sqrt{A^*A}right)+text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright).
    $$
    Since $$text{tr}left(sqrt{A^*A}tilde{B}^Tright)=text{tr}left((tilde{B}^T)^Tsqrt{A^*A}^Tright)=text{tr}left(tilde{B}sqrt{A^*A}right)$$ and $$text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright)=text{tr}left(tilde{B}^Ttilde{B}sqrt{A^*A}right)=text{tr}left(sqrt{A^*A}right),$$
    it follows$$
    langle A,Brangle=text{tr}left(sqrt{A^*A}tilde{B}^Tright)le text{tr}left(sqrt{A^*A}right)=|A|_*le delta.
    $$






    share|cite|improve this answer









    $endgroup$


















      4












      $begingroup$

      We can use the polar decomposition
      $$
      A=Psqrt{A^*A}
      $$
      where $sqrt{A^*A}$ is positive semi-definite and $P$ is orthogonal, i.e. $PP^T=P^TP=I$. Note that $text{tr}(AB)=text{tr}(BA)$ and $text{tr}(A)=text{tr}(A^T)$ hold for any $ntimes n$ real matrices $A,B$.



      1. We can take $B=P$ and it follows
      $$
      |A|_*=text{tr}(sqrt{A^*A})=text{tr}(B^TPsqrt{A^*A})=text{tr}(Psqrt{A^*A}B^T)=text{tr}(AB^T)=langle A,Brangle le delta.
      $$



      2. We have $$begin{eqnarray}langle A,Brangle &=&text{tr}(AB^T)\&=&text{tr}(Psqrt{A^*A}B^T)\&=&text{tr}(B^TPsqrt{A^*A})=text{tr}(tilde{B}sqrt{A^*A})end{eqnarray}$$ where $tilde{B}=B^TP$ is an orthogonal matrix. Since $(I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)$ is positive semi-definite, we have
      $$
      text{tr}left((I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)right)ge 0.
      $$
      This leads to
      $$
      text{tr}left(tilde{B}sqrt{A^*A}right)+text{tr}left(sqrt{A^*A}tilde{B}^Tright)letext{tr}left(sqrt{A^*A}right)+text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright).
      $$
      Since $$text{tr}left(sqrt{A^*A}tilde{B}^Tright)=text{tr}left((tilde{B}^T)^Tsqrt{A^*A}^Tright)=text{tr}left(tilde{B}sqrt{A^*A}right)$$ and $$text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright)=text{tr}left(tilde{B}^Ttilde{B}sqrt{A^*A}right)=text{tr}left(sqrt{A^*A}right),$$
      it follows$$
      langle A,Brangle=text{tr}left(sqrt{A^*A}tilde{B}^Tright)le text{tr}left(sqrt{A^*A}right)=|A|_*le delta.
      $$






      share|cite|improve this answer









      $endgroup$
















        4












        4








        4





        $begingroup$

        We can use the polar decomposition
        $$
        A=Psqrt{A^*A}
        $$
        where $sqrt{A^*A}$ is positive semi-definite and $P$ is orthogonal, i.e. $PP^T=P^TP=I$. Note that $text{tr}(AB)=text{tr}(BA)$ and $text{tr}(A)=text{tr}(A^T)$ hold for any $ntimes n$ real matrices $A,B$.



        1. We can take $B=P$ and it follows
        $$
        |A|_*=text{tr}(sqrt{A^*A})=text{tr}(B^TPsqrt{A^*A})=text{tr}(Psqrt{A^*A}B^T)=text{tr}(AB^T)=langle A,Brangle le delta.
        $$



        2. We have $$begin{eqnarray}langle A,Brangle &=&text{tr}(AB^T)\&=&text{tr}(Psqrt{A^*A}B^T)\&=&text{tr}(B^TPsqrt{A^*A})=text{tr}(tilde{B}sqrt{A^*A})end{eqnarray}$$ where $tilde{B}=B^TP$ is an orthogonal matrix. Since $(I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)$ is positive semi-definite, we have
        $$
        text{tr}left((I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)right)ge 0.
        $$
        This leads to
        $$
        text{tr}left(tilde{B}sqrt{A^*A}right)+text{tr}left(sqrt{A^*A}tilde{B}^Tright)letext{tr}left(sqrt{A^*A}right)+text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright).
        $$
        Since $$text{tr}left(sqrt{A^*A}tilde{B}^Tright)=text{tr}left((tilde{B}^T)^Tsqrt{A^*A}^Tright)=text{tr}left(tilde{B}sqrt{A^*A}right)$$ and $$text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright)=text{tr}left(tilde{B}^Ttilde{B}sqrt{A^*A}right)=text{tr}left(sqrt{A^*A}right),$$
        it follows$$
        langle A,Brangle=text{tr}left(sqrt{A^*A}tilde{B}^Tright)le text{tr}left(sqrt{A^*A}right)=|A|_*le delta.
        $$






        share|cite|improve this answer









        $endgroup$



        We can use the polar decomposition
        $$
        A=Psqrt{A^*A}
        $$
        where $sqrt{A^*A}$ is positive semi-definite and $P$ is orthogonal, i.e. $PP^T=P^TP=I$. Note that $text{tr}(AB)=text{tr}(BA)$ and $text{tr}(A)=text{tr}(A^T)$ hold for any $ntimes n$ real matrices $A,B$.



        1. We can take $B=P$ and it follows
        $$
        |A|_*=text{tr}(sqrt{A^*A})=text{tr}(B^TPsqrt{A^*A})=text{tr}(Psqrt{A^*A}B^T)=text{tr}(AB^T)=langle A,Brangle le delta.
        $$



        2. We have $$begin{eqnarray}langle A,Brangle &=&text{tr}(AB^T)\&=&text{tr}(Psqrt{A^*A}B^T)\&=&text{tr}(B^TPsqrt{A^*A})=text{tr}(tilde{B}sqrt{A^*A})end{eqnarray}$$ where $tilde{B}=B^TP$ is an orthogonal matrix. Since $(I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)$ is positive semi-definite, we have
        $$
        text{tr}left((I-tilde{B})sqrt{A^*A}(I-tilde{B}^T)right)ge 0.
        $$
        This leads to
        $$
        text{tr}left(tilde{B}sqrt{A^*A}right)+text{tr}left(sqrt{A^*A}tilde{B}^Tright)letext{tr}left(sqrt{A^*A}right)+text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright).
        $$
        Since $$text{tr}left(sqrt{A^*A}tilde{B}^Tright)=text{tr}left((tilde{B}^T)^Tsqrt{A^*A}^Tright)=text{tr}left(tilde{B}sqrt{A^*A}right)$$ and $$text{tr}left(tilde{B}sqrt{A^*A}tilde{B}^Tright)=text{tr}left(tilde{B}^Ttilde{B}sqrt{A^*A}right)=text{tr}left(sqrt{A^*A}right),$$
        it follows$$
        langle A,Brangle=text{tr}left(sqrt{A^*A}tilde{B}^Tright)le text{tr}left(sqrt{A^*A}right)=|A|_*le delta.
        $$







        share|cite|improve this answer












        share|cite|improve this answer



        share|cite|improve this answer










        answered Jan 21 at 9:54









        SongSong

        14.4k1635




        14.4k1635






























            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%2f3081655%2fprove-that-langle-mathbfa-mathbfc-rangle-leq-delta-equals-with-ma%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

            Mario Kart Wii

            What does “Dominus providebit” mean?

            File:Tiny Toon Adventures Wacky Sports JP Title.png