Let A∈RN×N represent a square matrix,
A−1AmA−1(A1A2…AM)=em=(e1e2…eM)
A−1(n=1∑NxnAn)=x,x∈RN
This is because
Ax=n=1∑NxnAnA−1Ax=A−1(n=1∑NxnAn)x=A−1(n=1∑NxnAn)
If A−1 exists, then any y∈RN can be expressed as a linear combination y=x1A1+⋯+xNAN=Ax.
The weights x={xn}n=1N can be obtained using A−1y.
This is because
y=Ax,A−1y=A−1Ax,A−1y=x.
Discussion