Suppose square matrix $A$ is positive-definite and $B$ is similar to $A$. Is B positive-definite too?

The name of the pictureThe name of the pictureThe name of the pictureClash Royale CLAN TAG#URR8PPP











up vote
1
down vote

favorite












Suppose square matrix $A$ is positive-definite and $B$ is similar to $A$. Is B positive-definite too?



I know a square matrix $A$ is positive-definite equal that $A$ is congruent to identity matrix $E$, i.e there exists an invertible matrix $C$ such that $C'AC=E$. So every matrix congruent to a positive-definite matrix is positive-definite. But what it will behavior when it's similar to a positive-definite matrix. I suppose it may not be positive-definite. But I can't get a counterexample. I tried from the definition...







share|cite|improve this question
















  • 1




    Hint: Positive definiteness is connected to the eigenvalues of a matrix and eigenvalues are preserved under similarity.
    – Wraith1995
    Aug 12 at 3:00






  • 2




    If you conjugate a symmetric matrix by a non-orthogonal matrix, chances are that you're not going to get a symmetric matrix back.
    – Theo Bendit
    Aug 12 at 3:12










  • In some definitions, symmetry is implicitly assume when one talks about a positive definite matrix. In other definitions, it is now.
    – ulaff.net
    Aug 12 at 3:27














up vote
1
down vote

favorite












Suppose square matrix $A$ is positive-definite and $B$ is similar to $A$. Is B positive-definite too?



I know a square matrix $A$ is positive-definite equal that $A$ is congruent to identity matrix $E$, i.e there exists an invertible matrix $C$ such that $C'AC=E$. So every matrix congruent to a positive-definite matrix is positive-definite. But what it will behavior when it's similar to a positive-definite matrix. I suppose it may not be positive-definite. But I can't get a counterexample. I tried from the definition...







share|cite|improve this question
















  • 1




    Hint: Positive definiteness is connected to the eigenvalues of a matrix and eigenvalues are preserved under similarity.
    – Wraith1995
    Aug 12 at 3:00






  • 2




    If you conjugate a symmetric matrix by a non-orthogonal matrix, chances are that you're not going to get a symmetric matrix back.
    – Theo Bendit
    Aug 12 at 3:12










  • In some definitions, symmetry is implicitly assume when one talks about a positive definite matrix. In other definitions, it is now.
    – ulaff.net
    Aug 12 at 3:27












up vote
1
down vote

favorite









up vote
1
down vote

favorite











Suppose square matrix $A$ is positive-definite and $B$ is similar to $A$. Is B positive-definite too?



I know a square matrix $A$ is positive-definite equal that $A$ is congruent to identity matrix $E$, i.e there exists an invertible matrix $C$ such that $C'AC=E$. So every matrix congruent to a positive-definite matrix is positive-definite. But what it will behavior when it's similar to a positive-definite matrix. I suppose it may not be positive-definite. But I can't get a counterexample. I tried from the definition...







share|cite|improve this question












Suppose square matrix $A$ is positive-definite and $B$ is similar to $A$. Is B positive-definite too?



I know a square matrix $A$ is positive-definite equal that $A$ is congruent to identity matrix $E$, i.e there exists an invertible matrix $C$ such that $C'AC=E$. So every matrix congruent to a positive-definite matrix is positive-definite. But what it will behavior when it's similar to a positive-definite matrix. I suppose it may not be positive-definite. But I can't get a counterexample. I tried from the definition...









share|cite|improve this question











share|cite|improve this question




share|cite|improve this question










asked Aug 12 at 2:41









Jaqen Chou

3377




3377







  • 1




    Hint: Positive definiteness is connected to the eigenvalues of a matrix and eigenvalues are preserved under similarity.
    – Wraith1995
    Aug 12 at 3:00






  • 2




    If you conjugate a symmetric matrix by a non-orthogonal matrix, chances are that you're not going to get a symmetric matrix back.
    – Theo Bendit
    Aug 12 at 3:12










  • In some definitions, symmetry is implicitly assume when one talks about a positive definite matrix. In other definitions, it is now.
    – ulaff.net
    Aug 12 at 3:27












  • 1




    Hint: Positive definiteness is connected to the eigenvalues of a matrix and eigenvalues are preserved under similarity.
    – Wraith1995
    Aug 12 at 3:00






  • 2




    If you conjugate a symmetric matrix by a non-orthogonal matrix, chances are that you're not going to get a symmetric matrix back.
    – Theo Bendit
    Aug 12 at 3:12










  • In some definitions, symmetry is implicitly assume when one talks about a positive definite matrix. In other definitions, it is now.
    – ulaff.net
    Aug 12 at 3:27







1




1




Hint: Positive definiteness is connected to the eigenvalues of a matrix and eigenvalues are preserved under similarity.
– Wraith1995
Aug 12 at 3:00




Hint: Positive definiteness is connected to the eigenvalues of a matrix and eigenvalues are preserved under similarity.
– Wraith1995
Aug 12 at 3:00




2




2




If you conjugate a symmetric matrix by a non-orthogonal matrix, chances are that you're not going to get a symmetric matrix back.
– Theo Bendit
Aug 12 at 3:12




If you conjugate a symmetric matrix by a non-orthogonal matrix, chances are that you're not going to get a symmetric matrix back.
– Theo Bendit
Aug 12 at 3:12












In some definitions, symmetry is implicitly assume when one talks about a positive definite matrix. In other definitions, it is now.
– ulaff.net
Aug 12 at 3:27




In some definitions, symmetry is implicitly assume when one talks about a positive definite matrix. In other definitions, it is now.
– ulaff.net
Aug 12 at 3:27










1 Answer
1






active

oldest

votes

















up vote
1
down vote













I assume $B$ is also symmetric, which is for most of the sources a precondition to talk about positive-definiteness, etc.



Try using the fact that an $ntimes n$ symmetric matrix $A$ is positive-definite iff for every $n$-dimensional column vector $vneq 0$ it is true that
$$v^T! cdot A cdot v >0$$
(which makes sense, since this is a $1times 1$ matrix.)



Now, if $B$ is similar to $A$, being $A$ and $B$ both symmetric, there exists a matrix $Q$ such that
$$B=Q^Tcdot A cdot Q.quad (*)$$



So what can you say about
$$v^T! cdot B cdot v$$
for any $vneq0$?




Solution:



Since
$$v^T! cdot B cdot v=v^T! cdot Q^Tcdot Acdot Q cdot v=(Qv)^T! cdot Acdot (Qv),$$
and being $(Qv)$ a vector of dimension $n$, say $w$, because $A$ is positive-definite we have
$$v^T! cdot B cdot v=w^T!cdot Acdot w>0,$$
which proves $B$ is also positive-definite.




(*) It is a theorem that if $A$ is a symmetric (real valued) matrix, then there exists an orthogonal matrix $P$ and a diagonal matrix $D$ such that
$$A=P^T D P.$$
Since the same is valid for $B$, say
$$B=tilde P^T tilde D tilde P^,$$
and being $P^T=P^-1$ and also $D=tilde D$ (because $A$ and $B$ are similar), we have
$$B=tilde P^T P A P^T tilde P=(P^Ttilde P)^T A (P^T tilde P),$$
and the desired result follows if we name $Q=P^Ttilde P$.






share|cite|improve this answer






















  • How do you know such a matrix $Q$ exists?
    – Robert Lewis
    Aug 12 at 3:09










  • Similarity means $exists S mid B = S^-1AS$.
    – Robert Lewis
    Aug 12 at 3:11










  • How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
    – Jaqen Chou
    Aug 12 at 3:12






  • 1




    I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
    – Robert Lewis
    Aug 12 at 3:44











  • Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
    – Alejandro Nasif Salum
    Aug 12 at 3:49










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: false,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
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%2f2879925%2fsuppose-square-matrix-a-is-positive-definite-and-b-is-similar-to-a-is-b-p%23new-answer', 'question_page');

);

Post as a guest






























1 Answer
1






active

oldest

votes








1 Answer
1






active

oldest

votes









active

oldest

votes






active

oldest

votes








up vote
1
down vote













I assume $B$ is also symmetric, which is for most of the sources a precondition to talk about positive-definiteness, etc.



Try using the fact that an $ntimes n$ symmetric matrix $A$ is positive-definite iff for every $n$-dimensional column vector $vneq 0$ it is true that
$$v^T! cdot A cdot v >0$$
(which makes sense, since this is a $1times 1$ matrix.)



Now, if $B$ is similar to $A$, being $A$ and $B$ both symmetric, there exists a matrix $Q$ such that
$$B=Q^Tcdot A cdot Q.quad (*)$$



So what can you say about
$$v^T! cdot B cdot v$$
for any $vneq0$?




Solution:



Since
$$v^T! cdot B cdot v=v^T! cdot Q^Tcdot Acdot Q cdot v=(Qv)^T! cdot Acdot (Qv),$$
and being $(Qv)$ a vector of dimension $n$, say $w$, because $A$ is positive-definite we have
$$v^T! cdot B cdot v=w^T!cdot Acdot w>0,$$
which proves $B$ is also positive-definite.




(*) It is a theorem that if $A$ is a symmetric (real valued) matrix, then there exists an orthogonal matrix $P$ and a diagonal matrix $D$ such that
$$A=P^T D P.$$
Since the same is valid for $B$, say
$$B=tilde P^T tilde D tilde P^,$$
and being $P^T=P^-1$ and also $D=tilde D$ (because $A$ and $B$ are similar), we have
$$B=tilde P^T P A P^T tilde P=(P^Ttilde P)^T A (P^T tilde P),$$
and the desired result follows if we name $Q=P^Ttilde P$.






share|cite|improve this answer






















  • How do you know such a matrix $Q$ exists?
    – Robert Lewis
    Aug 12 at 3:09










  • Similarity means $exists S mid B = S^-1AS$.
    – Robert Lewis
    Aug 12 at 3:11










  • How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
    – Jaqen Chou
    Aug 12 at 3:12






  • 1




    I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
    – Robert Lewis
    Aug 12 at 3:44











  • Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
    – Alejandro Nasif Salum
    Aug 12 at 3:49














up vote
1
down vote













I assume $B$ is also symmetric, which is for most of the sources a precondition to talk about positive-definiteness, etc.



Try using the fact that an $ntimes n$ symmetric matrix $A$ is positive-definite iff for every $n$-dimensional column vector $vneq 0$ it is true that
$$v^T! cdot A cdot v >0$$
(which makes sense, since this is a $1times 1$ matrix.)



Now, if $B$ is similar to $A$, being $A$ and $B$ both symmetric, there exists a matrix $Q$ such that
$$B=Q^Tcdot A cdot Q.quad (*)$$



So what can you say about
$$v^T! cdot B cdot v$$
for any $vneq0$?




Solution:



Since
$$v^T! cdot B cdot v=v^T! cdot Q^Tcdot Acdot Q cdot v=(Qv)^T! cdot Acdot (Qv),$$
and being $(Qv)$ a vector of dimension $n$, say $w$, because $A$ is positive-definite we have
$$v^T! cdot B cdot v=w^T!cdot Acdot w>0,$$
which proves $B$ is also positive-definite.




(*) It is a theorem that if $A$ is a symmetric (real valued) matrix, then there exists an orthogonal matrix $P$ and a diagonal matrix $D$ such that
$$A=P^T D P.$$
Since the same is valid for $B$, say
$$B=tilde P^T tilde D tilde P^,$$
and being $P^T=P^-1$ and also $D=tilde D$ (because $A$ and $B$ are similar), we have
$$B=tilde P^T P A P^T tilde P=(P^Ttilde P)^T A (P^T tilde P),$$
and the desired result follows if we name $Q=P^Ttilde P$.






share|cite|improve this answer






















  • How do you know such a matrix $Q$ exists?
    – Robert Lewis
    Aug 12 at 3:09










  • Similarity means $exists S mid B = S^-1AS$.
    – Robert Lewis
    Aug 12 at 3:11










  • How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
    – Jaqen Chou
    Aug 12 at 3:12






  • 1




    I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
    – Robert Lewis
    Aug 12 at 3:44











  • Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
    – Alejandro Nasif Salum
    Aug 12 at 3:49












up vote
1
down vote










up vote
1
down vote









I assume $B$ is also symmetric, which is for most of the sources a precondition to talk about positive-definiteness, etc.



Try using the fact that an $ntimes n$ symmetric matrix $A$ is positive-definite iff for every $n$-dimensional column vector $vneq 0$ it is true that
$$v^T! cdot A cdot v >0$$
(which makes sense, since this is a $1times 1$ matrix.)



Now, if $B$ is similar to $A$, being $A$ and $B$ both symmetric, there exists a matrix $Q$ such that
$$B=Q^Tcdot A cdot Q.quad (*)$$



So what can you say about
$$v^T! cdot B cdot v$$
for any $vneq0$?




Solution:



Since
$$v^T! cdot B cdot v=v^T! cdot Q^Tcdot Acdot Q cdot v=(Qv)^T! cdot Acdot (Qv),$$
and being $(Qv)$ a vector of dimension $n$, say $w$, because $A$ is positive-definite we have
$$v^T! cdot B cdot v=w^T!cdot Acdot w>0,$$
which proves $B$ is also positive-definite.




(*) It is a theorem that if $A$ is a symmetric (real valued) matrix, then there exists an orthogonal matrix $P$ and a diagonal matrix $D$ such that
$$A=P^T D P.$$
Since the same is valid for $B$, say
$$B=tilde P^T tilde D tilde P^,$$
and being $P^T=P^-1$ and also $D=tilde D$ (because $A$ and $B$ are similar), we have
$$B=tilde P^T P A P^T tilde P=(P^Ttilde P)^T A (P^T tilde P),$$
and the desired result follows if we name $Q=P^Ttilde P$.






share|cite|improve this answer














I assume $B$ is also symmetric, which is for most of the sources a precondition to talk about positive-definiteness, etc.



Try using the fact that an $ntimes n$ symmetric matrix $A$ is positive-definite iff for every $n$-dimensional column vector $vneq 0$ it is true that
$$v^T! cdot A cdot v >0$$
(which makes sense, since this is a $1times 1$ matrix.)



Now, if $B$ is similar to $A$, being $A$ and $B$ both symmetric, there exists a matrix $Q$ such that
$$B=Q^Tcdot A cdot Q.quad (*)$$



So what can you say about
$$v^T! cdot B cdot v$$
for any $vneq0$?




Solution:



Since
$$v^T! cdot B cdot v=v^T! cdot Q^Tcdot Acdot Q cdot v=(Qv)^T! cdot Acdot (Qv),$$
and being $(Qv)$ a vector of dimension $n$, say $w$, because $A$ is positive-definite we have
$$v^T! cdot B cdot v=w^T!cdot Acdot w>0,$$
which proves $B$ is also positive-definite.




(*) It is a theorem that if $A$ is a symmetric (real valued) matrix, then there exists an orthogonal matrix $P$ and a diagonal matrix $D$ such that
$$A=P^T D P.$$
Since the same is valid for $B$, say
$$B=tilde P^T tilde D tilde P^,$$
and being $P^T=P^-1$ and also $D=tilde D$ (because $A$ and $B$ are similar), we have
$$B=tilde P^T P A P^T tilde P=(P^Ttilde P)^T A (P^T tilde P),$$
and the desired result follows if we name $Q=P^Ttilde P$.







share|cite|improve this answer














share|cite|improve this answer



share|cite|improve this answer








edited Aug 12 at 3:39

























answered Aug 12 at 3:02









Alejandro Nasif Salum

3,10617




3,10617











  • How do you know such a matrix $Q$ exists?
    – Robert Lewis
    Aug 12 at 3:09










  • Similarity means $exists S mid B = S^-1AS$.
    – Robert Lewis
    Aug 12 at 3:11










  • How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
    – Jaqen Chou
    Aug 12 at 3:12






  • 1




    I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
    – Robert Lewis
    Aug 12 at 3:44











  • Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
    – Alejandro Nasif Salum
    Aug 12 at 3:49
















  • How do you know such a matrix $Q$ exists?
    – Robert Lewis
    Aug 12 at 3:09










  • Similarity means $exists S mid B = S^-1AS$.
    – Robert Lewis
    Aug 12 at 3:11










  • How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
    – Jaqen Chou
    Aug 12 at 3:12






  • 1




    I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
    – Robert Lewis
    Aug 12 at 3:44











  • Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
    – Alejandro Nasif Salum
    Aug 12 at 3:49















How do you know such a matrix $Q$ exists?
– Robert Lewis
Aug 12 at 3:09




How do you know such a matrix $Q$ exists?
– Robert Lewis
Aug 12 at 3:09












Similarity means $exists S mid B = S^-1AS$.
– Robert Lewis
Aug 12 at 3:11




Similarity means $exists S mid B = S^-1AS$.
– Robert Lewis
Aug 12 at 3:11












How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
– Jaqen Chou
Aug 12 at 3:12




How can we get $B=Q^TAQ$ by $B$ is similar to $A$. Shouldn't it is be $B=Q^-1AQ$?
– Jaqen Chou
Aug 12 at 3:12




1




1




I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
– Robert Lewis
Aug 12 at 3:44





I grant your argument if we assume $B^T = B$, so endorsed, +1; but I think you've added a assumption the OP may not have intended. Nevertheless, thanks for the explanation! Cheers!
– Robert Lewis
Aug 12 at 3:44













Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
– Alejandro Nasif Salum
Aug 12 at 3:49




Maybe, but it turns out that symmetry is usually assumed to study definiteness of a squared matrix/quadratic form. While it is possible to define the usual positive/negative/indefinite notions even if the matrix is not symmetric, most of the theory needs symmetry: just to say something, if a matrix is not symmetric, you can't even be sure that it will be diagonalizable, or that its eigenvalues will be real numbers at all. =S
– Alejandro Nasif Salum
Aug 12 at 3:49












 

draft saved


draft discarded


























 


draft saved


draft discarded














StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2879925%2fsuppose-square-matrix-a-is-positive-definite-and-b-is-similar-to-a-is-b-p%23new-answer', 'question_page');

);

Post as a guest













































































這個網誌中的熱門文章

How to combine Bézier curves to a surface?

Carbon dioxide

Why am i infinitely getting the same tweet with the Twitter Search API?