(Proposed by Shachar Carmeli, Weizmann Institute of Science)
Solution.
Answer: The maximum dimension of such a space is 2n(n+1).
The number 2n(n+1) can be achieved, for example the symmetric matrices are obviously t-normal and they form a linear space with dimension 2n(n+1). We shall show that this is the maximal possible dimension.
Let Mn denote the space of n×n complex matrices, let Sn⊂Mn be the subspace of all symmetric matrices and let An⊂Mn be the subspace of all anti-symmetric matrices, i.e. matrices A for which At=−A.
Let V⊂Mn be a linear subspace consisting of t-normal matrices. We have to show that dim(V)≤dim(Sn). Let π:V→Sn denote the linear map π(A)=A+At. We have
dim(V)=dim(Ker(π))+dim(Im(π))
so we have to prove that
dim(Ker(π))+dim(Im(π))≤dim(Sn). Notice that
Ker(π)⊆An.
We claim that for every A∈Ker(π) and B∈V, Aπ(B)=π(B)A. In other words, Ker(π) and Im(π) commute. Indeed, if A,B∈V and A=−At then
(A+B)(A+B)t=(A+B)t(A+B)⇔⇔AAt+ABt+BAt+BBt=AtA+AtB+BtA+BtB⇔⇔ABt−BA=−AB+BtA⇔A(B+Bt)=(B+Bt)A⇔⇔Aπ(B)=π(B)A.
Our bound on the dimension on V follows from the following lemma:
Lemma. Let X⊆Sn and Y⊆An be linear subspaces such that every element of X commutes with every element of Y. Then
dim(X)+dim(Y)≤dim(Sn)
Proof. Without loss of generality we may assume X=ZSn(Y):={x∈Sn:xy=yx∀y∈Y}. Define the bilinear map B:Sn×An→C by B(x,y)=tr(d[x,y]) where [x,y]=xy−yx and d=diag(1,…,n) is the matrix with diagonal elements 1,…,n and zeros off the diagonal. Clearly B(X,Y)={0}. Furthermore, if y∈Y satisfies that B(x,y)=0 for all x∈Sn then tr(d[x,y])=−tr([d,x],y])=0 for every x∈Sn.
We claim that {[d,x]:x∈Sn}=An. Let Eij denote the matrix with 1 in the entry (i,j) and 0 in all other entries. Then a direct computation shows that [d,Eij]=(j−i)Eij and therefore [d,Eij+Eji]=(j−i)(Eij−Eji) and the collection {(j−i)(Eij−Eji)}1≤i<j≤n span An for i=j.
It follows that if B(x,y)=0 for all x∈Sn then tr(yz)=0 for every z∈An. But then, taking z=yˉ, where yˉ is the entry-wise complex conjugate of y, we get 0=tr(yyˉ)=−tr(yyˉt) which is the sum of squares of all the entries of y. This means that y=0.
It follows that if y1,…,yk∈Y are linearly independent then the equations
B(x,y1)=0,…,B(x,yk)=0
are linearly independent as linear equations in
x, otherwise there are
a1,…,ak such that
B(x,a1y1+…+akyk)=0 for every
x∈Sn, a contradiction to the observation above. Since the solution of
k linearly independent linear equations is of codimension
k,
dim({x∈Sn:[x,yi]=0, for i=1,..,k})≤≤dim(x∈Sn:B(x,yi)=0 for i=1,…,k)=dim(Sn)−k.
The lemma follows by taking
y1,…,yk to be a basis of
Y.
Since Ker(π) and Im(π) commute, by the lemma we deduce that
dim(V)=dim(Ker(π))+dim(Im(π))≤dim(Sn)=2n(n+1).
Μεταγραφή από το επίσημο κείμενο λύσεων, σελίδες 3–4. Η μεταγραφή δεν έχει ελεγχθεί δεύτερη φορά· αν έχεις αμφιβολία, κοίτα το πρωτότυπο.