Skip to main content
deleted 680 characters in body
Source Link
Christian Remling
  • 24.2k
  • 2
  • 48
  • 83

OK, let me try again, maybe I'll get it right this time. I'll show (I hope) that $P$ is positive definite. This will imply the claim because if $(P+iQ)(x+iy)=0$ with $x,y\in\mathbb R^n$, then $Px=Qy$, $Py=-Qx$, and by taking scalar products with $x$ and $y$, respectively, we see that $\langle x, Px \rangle = -\langle y, Py\rangle$, which implies that $x=y=0$. Here I use that $Q$ is symmetric.

Let me firstnow show that $\det P>0$. I'll use row operations (but I never exchange rows, so the determinant stays the same throughout)$P>0$. By subtracting multiples of the first row from all other rows Following math110's suggestion, we get to the matrix $$ \begin{pmatrix} 2 & 1 & 1 & 1 & 1 & \ldots & 1& 1 \\ -3 & 2 & 0 & 0 & 0 & \ldots & 0 & 0\\ -5 & -1 & 3 & 0 & 0 & \ldots & 0 & 0 \\ -7 & -2 & -1 & 4 & 0 & \ldots & 0 & 0 \\ &&&\ldots &&&\\ -(2n-1) & -(n-2) & -(n-3) & -(n-4) & -(n-5) & \ldots & -1 & n \end{pmatrix} $$ In fact, the precise values don't mattercan simplify my original calculation as follows: The key feature is that all entries in the lower triangular part are negative, and those on the diagonal and in the first row are positiveLet $ B=B_n = P -\textrm{diag}(1,2,\ldots , n)$.

This means that if we now subtract For example, for $1/n$ times the last row from the first row$n=5$, then the remaining entries inthis is the first row stay positive. So wematrix $$ B_ 5= \begin{pmatrix} 1 & 1 & 1 & 1 & 1\\ 1 & 2 & 2 & 2 & 2\\ 1 & 2 & 3 & 3 & 3\\ 1 & 2 & 3 & 4 & 4\\ 1 & 2 & 3 & 4 & 5 \end{pmatrix} . $$ I can now have to subtract (a positive multiple ofin general) subtract the $(n-1)$st row to get rid offrom the $(1,n-1)$ element, and againlast row, this only makes the remaining entries inthen the first$(n-2)$nd row bigger.

We continue in this way, always subtracting rows from the first row. When we're done with this, we are still looking at a positive $(1,1)$ element$(n-1)$st row etc. Thus indeedThis confirms that $\det P>0$$\det B_n=1$.

Finally Moreover, I can run the same computation to see that the determinants of all upper left $k\times k$ submatrices of $B_n$ are also positive, and henceof the same type; they equal $P>0$$B_k$. This shows that $B>0$, by Sylvester's criterion, and thus $P>0$ as well.

OK, let me try again, maybe I'll get it right this time. I'll show (I hope) that $P$ is positive definite. This will imply the claim because if $(P+iQ)(x+iy)=0$ with $x,y\in\mathbb R^n$, then $Px=Qy$, $Py=-Qx$, and by taking scalar products with $x$ and $y$, respectively, we see that $\langle x, Px \rangle = -\langle y, Py\rangle$, which implies that $x=y=0$. Here I use that $Q$ is symmetric.

Let me first show that $\det P>0$. I'll use row operations (but I never exchange rows, so the determinant stays the same throughout). By subtracting multiples of the first row from all other rows, we get to the matrix $$ \begin{pmatrix} 2 & 1 & 1 & 1 & 1 & \ldots & 1& 1 \\ -3 & 2 & 0 & 0 & 0 & \ldots & 0 & 0\\ -5 & -1 & 3 & 0 & 0 & \ldots & 0 & 0 \\ -7 & -2 & -1 & 4 & 0 & \ldots & 0 & 0 \\ &&&\ldots &&&\\ -(2n-1) & -(n-2) & -(n-3) & -(n-4) & -(n-5) & \ldots & -1 & n \end{pmatrix} $$ In fact, the precise values don't matter: The key feature is that all entries in the lower triangular part are negative, and those on the diagonal and in the first row are positive.

This means that if we now subtract $1/n$ times the last row from the first row, then the remaining entries in the first row stay positive. So we now have to subtract (a positive multiple of) the $(n-1)$st row to get rid of the $(1,n-1)$ element, and again, this only makes the remaining entries in the first row bigger.

We continue in this way, always subtracting rows from the first row. When we're done with this, we are still looking at a positive $(1,1)$ element. Thus indeed $\det P>0$.

Finally, I can run the same computation to see that the determinants of all upper left $k\times k$ submatrices are also positive, and hence $P>0$ by Sylvester's criterion.

OK, let me try again, maybe I'll get it right this time. I'll show that $P$ is positive definite. This will imply the claim because if $(P+iQ)(x+iy)=0$ with $x,y\in\mathbb R^n$, then $Px=Qy$, $Py=-Qx$, and by taking scalar products with $x$ and $y$, respectively, we see that $\langle x, Px \rangle = -\langle y, Py\rangle$, which implies that $x=y=0$. Here I use that $Q$ is symmetric.

Let me now show that $P>0$. Following math110's suggestion, we can simplify my original calculation as follows: Let $ B=B_n = P -\textrm{diag}(1,2,\ldots , n)$. For example, for $n=5$, this is the matrix $$ B_ 5= \begin{pmatrix} 1 & 1 & 1 & 1 & 1\\ 1 & 2 & 2 & 2 & 2\\ 1 & 2 & 3 & 3 & 3\\ 1 & 2 & 3 & 4 & 4\\ 1 & 2 & 3 & 4 & 5 \end{pmatrix} . $$ I can now (in general) subtract the $(n-1)$st row from the last row, then the $(n-2)$nd row from the $(n-1)$st row etc. This confirms that $\det B_n=1$. Moreover, the upper left $k\times k$ submatrices of $B_n$ are of the same type; they equal $B_k$. This shows that $B>0$, by Sylvester's criterion, and thus $P>0$ as well.

Post Undeleted by Christian Remling
added 572 characters in body
Source Link
Christian Remling
  • 24.2k
  • 2
  • 48
  • 83

It suffices toOK, let me try again, maybe I'll get it right this time. I'll show (I hope) that either $P$ or $Q$ is non-singularpositive definite. This will imply the claim because if $(P+iQ)(x+iy)=0$ with $x,y\in\mathbb R^n$, then $Px=Qy$, $Py=-Qx$, and by taking scalar products with $x$ and $y$, respectively, we see that $\langle x, Px \rangle = -\langle y, Py\rangle$, which implies that $x=y=0$. Here I use that $Q$ is symmetric.

Let me first show that $\det P>0$. I'll do it foruse row operations $P$(but I never exchange rows, so the determinant stays the same throughout). By By subtracting multiples of the first row from all other rows, we get to the matrix $$ \begin{pmatrix} 2 & 1 & 1 & 1 & 1 & \ldots & 1& 1 \\ -3 & 2 & 0 & 0 & 0 & \ldots & 0 & 0\\ -5 & -1 & 3 & 0 & 0 & \ldots & 0 & 0 \\ -7 & -2 & -1 & 4 & 0 & \ldots & 0 & 0 \\ &&&\ldots &&&\\ -(2n-1) & -(n-2) & -(n-3) & -(n-4) & -(n-5) & \ldots & -1 & n \end{pmatrix} $$ In fact, the precise values don't matter: The key feature is that all entries in the lower triangular part are negative, and those on the diagonal and in the first row are positive.

This means that if we now subtract $1/n$ times the last row from the first row, then the remaining entries in the first row stay positive. So we now have to subtract (a positive multiple of) the $(n-1)$st row to get rid of the $(1,n-1)$ element, and again, this only makes the remaining entries in the first row bigger.

We continue in this way, always subtracting rows from the first row. When we're done with this, we are still looking at a positive $(1,1)$ element. In particularThus indeed $\det P>0$.

Finally, it is not zeroI can run the same computation to see that the determinants of all upper left $k\times k$ submatrices are also positive, and the transformed matrix has full rankhence $P>0$ by Sylvester's criterion.

It suffices to show that either $P$ or $Q$ is non-singular, and I'll do it for $P$. By subtracting multiples of the first row from all other rows, we get to the matrix $$ \begin{pmatrix} 2 & 1 & 1 & 1 & 1 & \ldots & 1& 1 \\ -3 & 2 & 0 & 0 & 0 & \ldots & 0 & 0\\ -5 & -1 & 3 & 0 & 0 & \ldots & 0 & 0 \\ -7 & -2 & -1 & 4 & 0 & \ldots & 0 & 0 \\ &&&\ldots &&&\\ -(2n-1) & -(n-2) & -(n-3) & -(n-4) & -(n-5) & \ldots & -1 & n \end{pmatrix} $$ In fact, the precise values don't matter: The key feature is that all entries in the lower triangular part are negative, and those on the diagonal and in the first row are positive.

This means that if we now subtract $1/n$ times the last row from the first row, then the remaining entries in the first row stay positive. So we now have to subtract the $(n-1)$st row to get rid of the $(1,n-1)$ element, and again, this only makes the remaining entries in the first row bigger.

We continue in this way, always subtracting rows from the first row. When we're done with this, we are still looking at a positive $(1,1)$ element. In particular, it is not zero, and the transformed matrix has full rank.

OK, let me try again, maybe I'll get it right this time. I'll show (I hope) that $P$ is positive definite. This will imply the claim because if $(P+iQ)(x+iy)=0$ with $x,y\in\mathbb R^n$, then $Px=Qy$, $Py=-Qx$, and by taking scalar products with $x$ and $y$, respectively, we see that $\langle x, Px \rangle = -\langle y, Py\rangle$, which implies that $x=y=0$. Here I use that $Q$ is symmetric.

Let me first show that $\det P>0$. I'll use row operations (but I never exchange rows, so the determinant stays the same throughout). By subtracting multiples of the first row from all other rows, we get to the matrix $$ \begin{pmatrix} 2 & 1 & 1 & 1 & 1 & \ldots & 1& 1 \\ -3 & 2 & 0 & 0 & 0 & \ldots & 0 & 0\\ -5 & -1 & 3 & 0 & 0 & \ldots & 0 & 0 \\ -7 & -2 & -1 & 4 & 0 & \ldots & 0 & 0 \\ &&&\ldots &&&\\ -(2n-1) & -(n-2) & -(n-3) & -(n-4) & -(n-5) & \ldots & -1 & n \end{pmatrix} $$ In fact, the precise values don't matter: The key feature is that all entries in the lower triangular part are negative, and those on the diagonal and in the first row are positive.

This means that if we now subtract $1/n$ times the last row from the first row, then the remaining entries in the first row stay positive. So we now have to subtract (a positive multiple of) the $(n-1)$st row to get rid of the $(1,n-1)$ element, and again, this only makes the remaining entries in the first row bigger.

We continue in this way, always subtracting rows from the first row. When we're done with this, we are still looking at a positive $(1,1)$ element. Thus indeed $\det P>0$.

Finally, I can run the same computation to see that the determinants of all upper left $k\times k$ submatrices are also positive, and hence $P>0$ by Sylvester's criterion.

Post Deleted by Christian Remling
Source Link
Christian Remling
  • 24.2k
  • 2
  • 48
  • 83

It suffices to show that either $P$ or $Q$ is non-singular, and I'll do it for $P$. By subtracting multiples of the first row from all other rows, we get to the matrix $$ \begin{pmatrix} 2 & 1 & 1 & 1 & 1 & \ldots & 1& 1 \\ -3 & 2 & 0 & 0 & 0 & \ldots & 0 & 0\\ -5 & -1 & 3 & 0 & 0 & \ldots & 0 & 0 \\ -7 & -2 & -1 & 4 & 0 & \ldots & 0 & 0 \\ &&&\ldots &&&\\ -(2n-1) & -(n-2) & -(n-3) & -(n-4) & -(n-5) & \ldots & -1 & n \end{pmatrix} $$ In fact, the precise values don't matter: The key feature is that all entries in the lower triangular part are negative, and those on the diagonal and in the first row are positive.

This means that if we now subtract $1/n$ times the last row from the first row, then the remaining entries in the first row stay positive. So we now have to subtract the $(n-1)$st row to get rid of the $(1,n-1)$ element, and again, this only makes the remaining entries in the first row bigger.

We continue in this way, always subtracting rows from the first row. When we're done with this, we are still looking at a positive $(1,1)$ element. In particular, it is not zero, and the transformed matrix has full rank.