Derivative of a trace with Kronecker product
Say I have the following expression:
$f=mathbf{Tr} (S^T L S)$,where $S=(Bbigotimes A)H$ Then what's the derivative of $f$ with respect to $A$ and $B$ ?
derivatives chain-rule kronecker-product
asked Aug 31 '16 at 6:39
FuliXiong
12
add a comment |
Say I have the following expression:
$f=mathbf{Tr} (S^T L S)$,where $S=(Bbigotimes A)H$ Then what's the derivative of $f$ with respect to $A$ and $B$ ?
derivatives chain-rule kronecker-product
asked Aug 31 '16 at 6:39
FuliXiong
12
add a comment |
Say I have the following expression:
$f=mathbf{Tr} (S^T L S)$,where $S=(Bbigotimes A)H$ Then what's the derivative of $f$ with respect to $A$ and $B$ ?
derivatives chain-rule kronecker-product
asked Aug 31 '16 at 6:39
FuliXiong
12
Say I have the following expression:
$f=mathbf{Tr} (S^T L S)$,where $S=(Bbigotimes A)H$ Then what's the derivative of $f$ with respect to $A$ and $B$ ?
derivatives chain-rule kronecker-product
derivatives chain-rule kronecker-product
asked Aug 31 '16 at 6:39
FuliXiong
12
asked Aug 31 '16 at 6:39
FuliXiong
12
asked Aug 31 '16 at 6:39
FuliXiong
12
asked Aug 31 '16 at 6:39
FuliXiong
12
asked Aug 31 '16 at 6:39
FuliXiong
12
12
add a comment |
add a comment |
1 Answer
1
active
oldest
votes
First find the SVD of
$$H=sum_k(sigma_ku_k)v_k^T = sum_kw_kv_k^T$$
Then use the trace/Frobenius product $big(A!:!B={rm Tr}(A^TB)big)$ to rewrite the function before finding its differential and the gradients.
$$eqalign{
f &= L:SS^T crcr
df &= L:(dS,S^T+S,dS^T) cr
&= (L+L^T):dS,S^T cr
&= (L+L^T),S:dS cr
&= (L+L^T),S:(dBotimes A+Botimes dA),H cr
&= sum_k,,(L+L^T),S:(dBotimes A+Botimes dA),w_kv_k^T cr
&= sum_k,,(L+L^T),Sv_k:(dBotimes A+Botimes dA),w_k cr
&= sum_k,,q_k:{rm vec}(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_k:(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_kBW_k^T:dA + Q_k^TAW_k:dB crcr
frac{partial f}{partial A} &= sum_k,Q_kBW_k^T,,,,,,,,
frac{partial f}{partial B} = sum_k,Q_k^TAW_k cr
}$$
where
$$eqalign{
{rm vec}(Q_k) &= q_k &= (L+L^T),Sv_k cr
{rm vec}(W_k) &= w_k &= sigma_k v_k cr
}$$
answered Oct 15 '16 at 5:28
greg
7,5351821
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',
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
});
}
});
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%2f1909404%2fderivative-of-a-trace-with-kronecker-product%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
First find the SVD of
$$H=sum_k(sigma_ku_k)v_k^T = sum_kw_kv_k^T$$
Then use the trace/Frobenius product $big(A!:!B={rm Tr}(A^TB)big)$ to rewrite the function before finding its differential and the gradients.
$$eqalign{
f &= L:SS^T crcr
df &= L:(dS,S^T+S,dS^T) cr
&= (L+L^T):dS,S^T cr
&= (L+L^T),S:dS cr
&= (L+L^T),S:(dBotimes A+Botimes dA),H cr
&= sum_k,,(L+L^T),S:(dBotimes A+Botimes dA),w_kv_k^T cr
&= sum_k,,(L+L^T),Sv_k:(dBotimes A+Botimes dA),w_k cr
&= sum_k,,q_k:{rm vec}(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_k:(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_kBW_k^T:dA + Q_k^TAW_k:dB crcr
frac{partial f}{partial A} &= sum_k,Q_kBW_k^T,,,,,,,,
frac{partial f}{partial B} = sum_k,Q_k^TAW_k cr
}$$
where
$$eqalign{
{rm vec}(Q_k) &= q_k &= (L+L^T),Sv_k cr
{rm vec}(W_k) &= w_k &= sigma_k v_k cr
}$$
answered Oct 15 '16 at 5:28
greg
7,5351821
add a comment |
First find the SVD of
$$H=sum_k(sigma_ku_k)v_k^T = sum_kw_kv_k^T$$
Then use the trace/Frobenius product $big(A!:!B={rm Tr}(A^TB)big)$ to rewrite the function before finding its differential and the gradients.
$$eqalign{
f &= L:SS^T crcr
df &= L:(dS,S^T+S,dS^T) cr
&= (L+L^T):dS,S^T cr
&= (L+L^T),S:dS cr
&= (L+L^T),S:(dBotimes A+Botimes dA),H cr
&= sum_k,,(L+L^T),S:(dBotimes A+Botimes dA),w_kv_k^T cr
&= sum_k,,(L+L^T),Sv_k:(dBotimes A+Botimes dA),w_k cr
&= sum_k,,q_k:{rm vec}(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_k:(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_kBW_k^T:dA + Q_k^TAW_k:dB crcr
frac{partial f}{partial A} &= sum_k,Q_kBW_k^T,,,,,,,,
frac{partial f}{partial B} = sum_k,Q_k^TAW_k cr
}$$
where
$$eqalign{
{rm vec}(Q_k) &= q_k &= (L+L^T),Sv_k cr
{rm vec}(W_k) &= w_k &= sigma_k v_k cr
}$$
answered Oct 15 '16 at 5:28
greg
7,5351821
add a comment |
First find the SVD of
$$H=sum_k(sigma_ku_k)v_k^T = sum_kw_kv_k^T$$
Then use the trace/Frobenius product $big(A!:!B={rm Tr}(A^TB)big)$ to rewrite the function before finding its differential and the gradients.
$$eqalign{
f &= L:SS^T crcr
df &= L:(dS,S^T+S,dS^T) cr
&= (L+L^T):dS,S^T cr
&= (L+L^T),S:dS cr
&= (L+L^T),S:(dBotimes A+Botimes dA),H cr
&= sum_k,,(L+L^T),S:(dBotimes A+Botimes dA),w_kv_k^T cr
&= sum_k,,(L+L^T),Sv_k:(dBotimes A+Botimes dA),w_k cr
&= sum_k,,q_k:{rm vec}(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_k:(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_kBW_k^T:dA + Q_k^TAW_k:dB crcr
frac{partial f}{partial A} &= sum_k,Q_kBW_k^T,,,,,,,,
frac{partial f}{partial B} = sum_k,Q_k^TAW_k cr
}$$
where
$$eqalign{
{rm vec}(Q_k) &= q_k &= (L+L^T),Sv_k cr
{rm vec}(W_k) &= w_k &= sigma_k v_k cr
}$$
answered Oct 15 '16 at 5:28
greg
7,5351821
First find the SVD of
$$H=sum_k(sigma_ku_k)v_k^T = sum_kw_kv_k^T$$
Then use the trace/Frobenius product $big(A!:!B={rm Tr}(A^TB)big)$ to rewrite the function before finding its differential and the gradients.
$$eqalign{
f &= L:SS^T crcr
df &= L:(dS,S^T+S,dS^T) cr
&= (L+L^T):dS,S^T cr
&= (L+L^T),S:dS cr
&= (L+L^T),S:(dBotimes A+Botimes dA),H cr
&= sum_k,,(L+L^T),S:(dBotimes A+Botimes dA),w_kv_k^T cr
&= sum_k,,(L+L^T),Sv_k:(dBotimes A+Botimes dA),w_k cr
&= sum_k,,q_k:{rm vec}(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_k:(AW_k,dB^T+dA,W_kB^T) cr
&= sum_k,,Q_kBW_k^T:dA + Q_k^TAW_k:dB crcr
frac{partial f}{partial A} &= sum_k,Q_kBW_k^T,,,,,,,,
frac{partial f}{partial B} = sum_k,Q_k^TAW_k cr
}$$
where
$$eqalign{
{rm vec}(Q_k) &= q_k &= (L+L^T),Sv_k cr
{rm vec}(W_k) &= w_k &= sigma_k v_k cr
}$$
answered Oct 15 '16 at 5:28
greg
7,5351821
edited Dec 2 '18 at 21:11
answered Oct 15 '16 at 5:28
greg
7,5351821
answered Oct 15 '16 at 5:28
greg
7,5351821
answered Oct 15 '16 at 5:28
greg
7,5351821
7,5351821
add a comment |
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%2f1909404%2fderivative-of-a-trace-with-kronecker-product%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
