I'm trying to do a few proofs about rings: $(1)$ I want to show that for a ring $R$ and the ring of matrices $M_n(R)$, if $I\lhd R\implies M_n(I)\lhd M_n(R)$ and that all ideals of $M_n(R)$ are of the form $M_n(I)$. $(2)$ Prove that if $R$ is a ring with division then $M_n(R)$ is a simple ring.

The relation $I\lhd R$ implies that with $r\in R$ and $x\in I$ we have $rx\in I, xr\in I$. But how can I prove the statement for matrices?, here's what I would do: Consider $\psi \in M_n(R)$, if is given any matrix $\phi\in M_n(I)$ it must be $\psi \phi \in M_n(I),\phi \psi \in M_n(I)$, if we denote the element $(i,j)$ of each matrix by $\phi_{ij}$ and $\psi_{ij}$, then $(\psi \phi)_{ij}=\sum_{k=1}^n \psi_{ik}\phi_{kj}$ and $(\phi \psi)_{ij}=\sum_{k=1}^n \phi_{ik}\psi_{kj}$.

Since $I\lhd R$ is $\psi_{ik}\phi_{kj}\in I$ and $\phi_{ik}\psi_{kj}\in I$, which means that $I\lhd R\implies M_n(I)\lhd M_n(R)$.

To show that every ideal of $M_n(R)$ are of the form $M_n(I)$ with $I$ ideal: Suppose $I$ is not an ideal of $R$, then exists $r\in R$ such that $rx\notin I$ or $xr\notin I$, then with taking a matrix $\phi\in M_n(I)$ such that $x$ is an entry of $\phi$ and a matrix $\psi$ such that $r$ is an entry of $\psi$ we get that the matrices $\phi$ and $\psi$ have an entry $xr$ or $rx$, in either case this is not an element of $I$ then the matrix is not in $M_n(I)$. This concludes my proof (is it?) for $(1)$.

For $(2)$ I'd like to proceed by absurdum . Let's say that $M_n(R)$ is not a simple ring, then it has at least one ideal. Denote by $\varphi$ this ideal in $I$, then for any matrix $\psi$ we have $\varphi\psi \in M_n(I)$ and $\psi \varphi \in M_n(I)$.

Now, can $R$ be a ring with division?.I'm not sure how to prove that it can't. I believe I could use this theorem: "If $I\lhd R$ and $1\in I$ then $I=R$. The case $I\neq R$ implies $1\neq R$ and since $\varphi \psi$ and $\psi \varphi$ are in $M_n(I)$ then there are at entries in the product matrix which belong to $I$ (using $I \lhd R$) but are not $1$ (1\notin I); then there are elements in $I$ (and therefore in $R$ by $I\subset R$) that are not invertible, follows that $R$ is not a ring with division. But if $I=R$?, here I can't see a why $R$ can't be a ring with division.


Your proof of the second part of $(1)$ shows that if $I$ is not an ideal of $R$, then $M_n(I)$ is not an ideal of $M_n(R)$. Although this is true, it isn't quite what we're hoping to prove. We want to show that if $I$ is any ideal of $M_n(R)$, then there must be some ideal $J\subset R$ such that $I=M_n(J)$.

A brief hint on how to do this: let $S$ be the subset of $R$ consisting of all elements which appear as an entry in some matrix in $I$, and let $J$ be the ideal generated by $S$. I claim that $I=M_n(J)$. First, by definition of $J$ we have that $I\subset M_n(J)$. To show the reverse containment, notice that if $A=(a_{ij})\in I$, then multiplying $A$ by appropriate matrices on the left and right, we can show that for any $a_{ij}$, the matrix whose only nonzero entry is $a_{ij}$ in position $(i,j)$ is also in $I$. Then, we can multiply by an appropriate permutation matrix to move the nonzero entry to any position. In essence, we are using the rather strong condition that $I$ is a two-sided ideal to show that any matrix in $I$ leads to the existence of a lot of other, simpler matrices in $I$. These simpler matrices can then be added appropriately to show that $M_n(J)\subset I$.

I left out a lot of details for you to work through, but this is the basic idea.

For $(2)$, you are definitely on the right track. Suppose $I$ is an ideal in $M_n(R)$, so that there is some ideal $J\subset R$ such that $I=M_n(J)$ by what we just proved. However, division rings are simple (why?), so it follows that... (see if you can finish the argument).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.