\( \newcommand{\matr}[1] {\mathbf{#1}} \newcommand{\vertbar} {\rule[-1ex]{0.5pt}{2.5ex}} \newcommand{\horzbar} {\rule[.5ex]{2.5ex}{0.5pt}} \newcommand{\E} {\mathrm{E}} \)
deepdream of
          a sidewalk
Show Question
\( \newcommand{\cat}[1] {\mathrm{#1}} \newcommand{\catobj}[1] {\operatorname{Obj}(\mathrm{#1})} \newcommand{\cathom}[1] {\operatorname{Hom}_{\cat{#1}}} \newcommand{\multiBetaReduction}[0] {\twoheadrightarrow_{\beta}} \newcommand{\betaReduction}[0] {\rightarrow_{\beta}} \newcommand{\betaEq}[0] {=_{\beta}} \newcommand{\string}[1] {\texttt{"}\mathtt{#1}\texttt{"}} \newcommand{\symbolq}[1] {\texttt{`}\mathtt{#1}\texttt{'}} \newcommand{\groupMul}[1] { \cdot_{\small{#1}}} \newcommand{\groupAdd}[1] { +_{\small{#1}}} \newcommand{\inv}[1] {#1^{-1} } \newcommand{\bm}[1] { \boldsymbol{#1} } \newcommand{\qed} { {\scriptstyle \Box} } \require{physics} \require{ams} \require{mathtools} \)
Math and science::INF ML AI

Ax=b, in the steepest descent method

When searching for an \( x \) such that \( Ax = b \), a fundamental equation used in optimization is:

\[ \frac{\partial}{\partial x} \left( \frac{1}{2} x^T A x \right) = A x \]

Sometimes this appears by defining a function \( f \):

\[ f(x) = \frac{1}{2} x^T A x - b^T x \]

and stating that the goal is to minimize:

\[ \nabla f(x) = \frac{\partial f}{\partial x}. \]

When \( \nabla f(x) = 0 \), we must have \( Ax = b \).

Geometric interpretation

\( f(x) \) defines a paraboloid in the same space as \( x \). The minimum of this paraboloid is the solution to \( Ax = b \). The gradient of \( f(x) \) is the direction of steepest ascent. When the gradient is zero, we are at the minimum.