interpreting dual norm of quadratic norm with change of basis
up vote
1
down vote
favorite
I'm trying to compute
$$
|z|_* := sup_{|x|_Aleq1} |z^T x|
$$
for $|x|_A := x^T A x$ with $A succ 0$, symmetric. I see that we're looking at $z^T in mathbb{R}^n$ as a linear functional, so this is really the operator norm with $|cdot|$ over $mathbb{R}$ and $|cdot|_A$ over $mathbb{R}^n$.
I make a change of coordinates $y = A^{1/2}x$ and substitute to get
$$
|z|_* := sup_{|y|_2leq1} |z^T A^{-1/2}y|.
$$
Based on @max_zorn's comment, I use symmetry of $A$ so that
$$
|z|_* = sup_{y^T y=1}|A^{-1/2}z^T y| = sup_{|y|_2=1}|A^{-1/2}z^T y| leq |A^{-1/2}z|_2 |y|_2
$$
with equality at $y = A^{-1/2}z / |A^{-1/2}z|_2$ so $|z|_* = |A^{-1/2}z|_2$.
What is the function analysis explanation for why the dual norm should be computable as the euclidean norm of the transformed vector $A^{-1/2}z$? How can I think about duality more intuitively in this case?
functional-analysis norm convex-optimization duality-theorems
add a comment |
up vote
1
down vote
favorite
I'm trying to compute
$$
|z|_* := sup_{|x|_Aleq1} |z^T x|
$$
for $|x|_A := x^T A x$ with $A succ 0$, symmetric. I see that we're looking at $z^T in mathbb{R}^n$ as a linear functional, so this is really the operator norm with $|cdot|$ over $mathbb{R}$ and $|cdot|_A$ over $mathbb{R}^n$.
I make a change of coordinates $y = A^{1/2}x$ and substitute to get
$$
|z|_* := sup_{|y|_2leq1} |z^T A^{-1/2}y|.
$$
Based on @max_zorn's comment, I use symmetry of $A$ so that
$$
|z|_* = sup_{y^T y=1}|A^{-1/2}z^T y| = sup_{|y|_2=1}|A^{-1/2}z^T y| leq |A^{-1/2}z|_2 |y|_2
$$
with equality at $y = A^{-1/2}z / |A^{-1/2}z|_2$ so $|z|_* = |A^{-1/2}z|_2$.
What is the function analysis explanation for why the dual norm should be computable as the euclidean norm of the transformed vector $A^{-1/2}z$? How can I think about duality more intuitively in this case?
functional-analysis norm convex-optimization duality-theorems
$langle z,A^{-1/2}yrangle = langle A^{-1/2}z,yrangle$
– max_zorn
Jan 6 at 19:24
So you have $|z|_*=sup_{y^Ty=1} |A^{-1/2}z^T y| $. Then what?
– jjjjjj
Jan 6 at 20:12
You are done. This computes the Euclidean norm of $A^{-1/2}{z}$.
– max_zorn
Jan 6 at 20:19
Sorry, I'm still not seeing how. Can you clarify?
– jjjjjj
Jan 6 at 21:22
1
$max_{|y|leq 1} langle w,yrangle = |w|$. To see this, Observe that Cauchy-Schwarz gives LHS $leq$ RHS. To get equality, consider $y=w/|w|$.
– max_zorn
Jan 6 at 21:23
add a comment |
up vote
1
down vote
favorite
up vote
1
down vote
favorite
I'm trying to compute
$$
|z|_* := sup_{|x|_Aleq1} |z^T x|
$$
for $|x|_A := x^T A x$ with $A succ 0$, symmetric. I see that we're looking at $z^T in mathbb{R}^n$ as a linear functional, so this is really the operator norm with $|cdot|$ over $mathbb{R}$ and $|cdot|_A$ over $mathbb{R}^n$.
I make a change of coordinates $y = A^{1/2}x$ and substitute to get
$$
|z|_* := sup_{|y|_2leq1} |z^T A^{-1/2}y|.
$$
Based on @max_zorn's comment, I use symmetry of $A$ so that
$$
|z|_* = sup_{y^T y=1}|A^{-1/2}z^T y| = sup_{|y|_2=1}|A^{-1/2}z^T y| leq |A^{-1/2}z|_2 |y|_2
$$
with equality at $y = A^{-1/2}z / |A^{-1/2}z|_2$ so $|z|_* = |A^{-1/2}z|_2$.
What is the function analysis explanation for why the dual norm should be computable as the euclidean norm of the transformed vector $A^{-1/2}z$? How can I think about duality more intuitively in this case?
functional-analysis norm convex-optimization duality-theorems
I'm trying to compute
$$
|z|_* := sup_{|x|_Aleq1} |z^T x|
$$
for $|x|_A := x^T A x$ with $A succ 0$, symmetric. I see that we're looking at $z^T in mathbb{R}^n$ as a linear functional, so this is really the operator norm with $|cdot|$ over $mathbb{R}$ and $|cdot|_A$ over $mathbb{R}^n$.
I make a change of coordinates $y = A^{1/2}x$ and substitute to get
$$
|z|_* := sup_{|y|_2leq1} |z^T A^{-1/2}y|.
$$
Based on @max_zorn's comment, I use symmetry of $A$ so that
$$
|z|_* = sup_{y^T y=1}|A^{-1/2}z^T y| = sup_{|y|_2=1}|A^{-1/2}z^T y| leq |A^{-1/2}z|_2 |y|_2
$$
with equality at $y = A^{-1/2}z / |A^{-1/2}z|_2$ so $|z|_* = |A^{-1/2}z|_2$.
What is the function analysis explanation for why the dual norm should be computable as the euclidean norm of the transformed vector $A^{-1/2}z$? How can I think about duality more intuitively in this case?
functional-analysis norm convex-optimization duality-theorems
functional-analysis norm convex-optimization duality-theorems
edited Jan 7 at 18:11
asked Jan 6 at 19:21
jjjjjj
1,100515
1,100515
$langle z,A^{-1/2}yrangle = langle A^{-1/2}z,yrangle$
– max_zorn
Jan 6 at 19:24
So you have $|z|_*=sup_{y^Ty=1} |A^{-1/2}z^T y| $. Then what?
– jjjjjj
Jan 6 at 20:12
You are done. This computes the Euclidean norm of $A^{-1/2}{z}$.
– max_zorn
Jan 6 at 20:19
Sorry, I'm still not seeing how. Can you clarify?
– jjjjjj
Jan 6 at 21:22
1
$max_{|y|leq 1} langle w,yrangle = |w|$. To see this, Observe that Cauchy-Schwarz gives LHS $leq$ RHS. To get equality, consider $y=w/|w|$.
– max_zorn
Jan 6 at 21:23
add a comment |
$langle z,A^{-1/2}yrangle = langle A^{-1/2}z,yrangle$
– max_zorn
Jan 6 at 19:24
So you have $|z|_*=sup_{y^Ty=1} |A^{-1/2}z^T y| $. Then what?
– jjjjjj
Jan 6 at 20:12
You are done. This computes the Euclidean norm of $A^{-1/2}{z}$.
– max_zorn
Jan 6 at 20:19
Sorry, I'm still not seeing how. Can you clarify?
– jjjjjj
Jan 6 at 21:22
1
$max_{|y|leq 1} langle w,yrangle = |w|$. To see this, Observe that Cauchy-Schwarz gives LHS $leq$ RHS. To get equality, consider $y=w/|w|$.
– max_zorn
Jan 6 at 21:23
$langle z,A^{-1/2}yrangle = langle A^{-1/2}z,yrangle$
– max_zorn
Jan 6 at 19:24
$langle z,A^{-1/2}yrangle = langle A^{-1/2}z,yrangle$
– max_zorn
Jan 6 at 19:24
So you have $|z|_*=sup_{y^Ty=1} |A^{-1/2}z^T y| $. Then what?
– jjjjjj
Jan 6 at 20:12
So you have $|z|_*=sup_{y^Ty=1} |A^{-1/2}z^T y| $. Then what?
– jjjjjj
Jan 6 at 20:12
You are done. This computes the Euclidean norm of $A^{-1/2}{z}$.
– max_zorn
Jan 6 at 20:19
You are done. This computes the Euclidean norm of $A^{-1/2}{z}$.
– max_zorn
Jan 6 at 20:19
Sorry, I'm still not seeing how. Can you clarify?
– jjjjjj
Jan 6 at 21:22
Sorry, I'm still not seeing how. Can you clarify?
– jjjjjj
Jan 6 at 21:22
1
1
$max_{|y|leq 1} langle w,yrangle = |w|$. To see this, Observe that Cauchy-Schwarz gives LHS $leq$ RHS. To get equality, consider $y=w/|w|$.
– max_zorn
Jan 6 at 21:23
$max_{|y|leq 1} langle w,yrangle = |w|$. To see this, Observe that Cauchy-Schwarz gives LHS $leq$ RHS. To get equality, consider $y=w/|w|$.
– max_zorn
Jan 6 at 21:23
add a comment |
1 Answer
1
active
oldest
votes
up vote
0
down vote
I suppose that you drop a power exponent of 1/2 in the term $||x||_A = (x^TAx)^{1/2}$
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
add a comment |
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',
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
});
}
});
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
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2594694%2finterpreting-dual-norm-of-quadratic-norm-with-change-of-basis%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
up vote
0
down vote
I suppose that you drop a power exponent of 1/2 in the term $||x||_A = (x^TAx)^{1/2}$
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
add a comment |
up vote
0
down vote
I suppose that you drop a power exponent of 1/2 in the term $||x||_A = (x^TAx)^{1/2}$
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
add a comment |
up vote
0
down vote
up vote
0
down vote
I suppose that you drop a power exponent of 1/2 in the term $||x||_A = (x^TAx)^{1/2}$
I suppose that you drop a power exponent of 1/2 in the term $||x||_A = (x^TAx)^{1/2}$
edited Nov 24 at 4:35
answered Nov 23 at 14:08
Caldera
11
11
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
add a comment |
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
I'm not sure this actually answers the question
– MRobinson
Nov 23 at 14:32
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
Well, I didn't put it clearly, I mean that the question do help me a lot, and my point is in fact the latter. Feel sorry to reedit my answer.
– Caldera
Nov 24 at 11:26
add a comment |
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.
Some of your past answers have not been well-received, and you're in danger of being blocked from answering.
Please pay close attention to the following guidance:
- 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.
To learn more, see our tips on writing great answers.
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
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2594694%2finterpreting-dual-norm-of-quadratic-norm-with-change-of-basis%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
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
Required, but never shown
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
Required, but never shown
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
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
$langle z,A^{-1/2}yrangle = langle A^{-1/2}z,yrangle$
– max_zorn
Jan 6 at 19:24
So you have $|z|_*=sup_{y^Ty=1} |A^{-1/2}z^T y| $. Then what?
– jjjjjj
Jan 6 at 20:12
You are done. This computes the Euclidean norm of $A^{-1/2}{z}$.
– max_zorn
Jan 6 at 20:19
Sorry, I'm still not seeing how. Can you clarify?
– jjjjjj
Jan 6 at 21:22
1
$max_{|y|leq 1} langle w,yrangle = |w|$. To see this, Observe that Cauchy-Schwarz gives LHS $leq$ RHS. To get equality, consider $y=w/|w|$.
– max_zorn
Jan 6 at 21:23