I am currently reading the book ``Symmetries, Lie Algebras and Representations: A graduate course for physicists'', by Jurgen Fuchs & Christoph Schweigert. On page 75 (unfortunately, the google book does not cover this page), I read:
Another simple subalgebra of $gl(2n)$ consists of those matrices of $M$ which fulfill the relation $M^tK+KM=0$, where $K$ is the $2n\times 2n$ -matrix $$K:=\begin{pmatrix} 0_n & 1_n \\ 1_n & 0_n \end{pmatrix}.$$
My question is how can such a definition be consistent with the normal definition of the lie algebra $so(2n)$ which consist of all $2n\times 2n$ antisymmetric matrices?
The authors also note after some sentences that
Note that, here the symbols $sp(n)$ and $so(n)$ refer to Lie algebras over complex numbers.
I am not sure is this remark relevant to my question.
Let's take $n=1$ for example. Let $M=\begin{pmatrix} a & b \\ c & d \end{pmatrix}$, then $M^tK+KM=0$ implies that $$\begin{pmatrix} a & c \\ b & d \end{pmatrix}\equiv M^t=-KMK^{-1}=-KMK=-\begin{pmatrix} d & c \\ b & a \end{pmatrix}.$$ That is, $a=-d$, $c=b=0$, meaning that $M=\begin{pmatrix} a & 0 \\ 0 & -a \end{pmatrix}$ which is inconsistent with the normal one $M=\begin{pmatrix} 0 & b \\ -b & 0 \end{pmatrix}$.