I need to recheck a proof for minimizing the error by finding optimum step-size. I re-checked the proof many times but still can't find a mistake although the number I am getting in Matlab is not expected, so please if any one is good with linear algebra can tell me If I have an obvious mistake.
here is my notation:
- $\hat{X}_t$ is function of $\mu$ and matrices of size $M \times N$
- $\mathbf{X_t}$, $\mathbf{Y_t}$, $\boldsymbol{\Omega_t}$ are constant matrices of size $M \times N$
- $\mathbf{A}$ is constant matrix $M \times R$, $\mathbf{B}$ is constant matrix of size $N \times R$ and $\mathbf{D}$ is constant matrix $R \times R$
- $\mu$, $\lambda$, $t$ are constant scalar values
- $\|\cdot\|_F$ denotes the Frobenius norm
