Fractal-like matrices (Q817032)

The authors define the set of \(n\)-ordered fractal-like matrices \(F_n\), for each \(n=2^k\), inductively: \(F_{2^0}:={\mathbf R}^{1\times 1}\) and for each \(k>0\), \(f_{2^k}=\left\{\left[\begin{smallmatrix} A & B\\ 0 & C \end{smallmatrix}\right]:A,B,C\in F_{2{k-1}}\right\}\). As \(n\to\infty\) the structure of \(E_n\) reminds us of the Sierpinski triangle which is a famous fractal. It is shown that \(F_n\) is an algebra over \({\mathbf R}\) containing \(E_n\). The number of arithmetic operations required to add, multiply or invert fractal-like matrices is obtained recursively by solving recurrence equations. A special storage scheme is proposed that saves memory and is convenient for arithmetic operations.