= Positive-definite quadratic optimization
For a real <positive-definite symmetric matrix> $V$ and vector $b$, the <quadratic function>
$$
f(x)=b^Tx-\tfrac12x^TVx
$$
has unique global maximizer $x_*=V^{-1}b$ and maximum $\tfrac12b^TV^{-1}b$. Completing the square proves both claims:
$$
f(x)=\tfrac12b^TV^{-1}b-\tfrac12(x-V^{-1}b)^TV(x-V^{-1}b).
$$
The subtracted <quadratic form> is nonnegative and vanishes only at $x_*$. Solving $Vx_*=b$ is therefore sufficient, rather than merely a necessary <first-order optimality condition>. Numerically one solves the <linear system> instead of explicitly constructing the inverse.
For $V=\operatorname{diag}(2,4)$ and $b=(2,4)^T$, the maximizer is $(1,1)^T$ and its value is three. If $V$ were only positive semidefinite, uniqueness could fail, and a linear component along its <kernel> could make the objective unbounded above. This is why strict positive definiteness carries both existence and uniqueness.
Back to article page