Appendix D Equivalence between LQT and LQR with augmented state space

The equivalence between the original LQT problem and the corresponding LQR problem with an augmented state space can be shown by using the block matrices composition rule

\bm{M}=\footnotesize\begin{bmatrix}\bm{M}_{11}&\bm{M}_{12}\\ \bm{M}_{21}&\bm{M}_{22}\end{bmatrix}\quad\iff\quad\bm{M}^{-1}=\footnotesize\begin{bmatrix}\bm{I}&\bm{0}\\ -\bm{M}_{22}^{-1}\bm{M}_{21}&\bm{I}\end{bmatrix}\begin{bmatrix}\bm{S}^{-1}&\bm{0}\\ \bm{0}&\bm{M}_{22}^{-1}\end{bmatrix}\begin{bmatrix}\bm{I}&-\bm{M}_{12}\bm{M}_{22}^{-1}\\ \bm{0}&\bm{I}\end{bmatrix},

where \bm{M}_{11}\!\in\!\mathbb{R}^{d\times d} , \bm{M}_{12}\!\in\!\mathbb{R}^{d\times D} , \bm{M}_{21}\!\in\!\mathbb{R}^{D\times d} , \bm{M}_{22}\!\in\!\mathbb{R}^{D\times D} , and \bm{S}=\bm{M}_{11}-\bm{M}_{12}\bm{M}_{22}^{-1}\bm{M}_{21} the Schur complement of the matrix \bm{M} .

In our case, we have

\begin{bmatrix}\bm{Q}^{-1}\!+\!\bm{\mu}\bm{\mu}^{\scriptscriptstyle\top}&\bm{\mu}\\ \bm{\mu}^{\scriptscriptstyle\top}&1\end{bmatrix}^{-1}=\begin{bmatrix}\bm{I}&\bm{0}\\ -\bm{\mu}^{\scriptscriptstyle\top}&1\end{bmatrix}\begin{bmatrix}\bm{Q}&\bm{0}\\ \bm{0}&1\end{bmatrix}\begin{bmatrix}\bm{I}&-\bm{\mu}\\ \bm{0}&1\end{bmatrix},

and it is easy to see that

{(\bm{x}-\bm{\mu})}^{\scriptscriptstyle\top}\bm{Q}(\bm{x}-\bm{\mu})=\begin{bmatrix}\bm{x}\\ 1\end{bmatrix}^{\scriptscriptstyle\top}\begin{bmatrix}\bm{Q}^{-1}\!+\!\bm{\mu}\bm{\mu}^{\scriptscriptstyle\top}&\bm{\mu}\\ \bm{\mu}^{\scriptscriptstyle\top}&1\end{bmatrix}^{-1}\begin{bmatrix}\bm{x}\\ 1\end{bmatrix}-1.