Logarithmic barrier ($f_{\mu }(x)=f(x)-\mu {\sum }_i\log c_i(x)$)

For optimization constraints $c_i(x)\ge 0$ and $\mu >0$, the logarithmic barrier objective is $f_{\mu }(x)=f(x)-\mu {\sum }_i\log c_i(x)$ on $D=\{x:c_i(x)>0\forall i\}$. Its derivatives are defined only on this strictly feasible open subset. One may extend its value to $+\infty$ off $D$ to formulate an extended-real convex function. The penalty alone is $-{\sum }_i\log c_i$, while $f_{\mu }$ includes the original objective function.

When $f$ is convex and each $c_i$ is concave, $D$ is a convex set andEvery term of this Hessian matrix is a positive semidefinite matrix, so $f_{\mu }$ is convex. Its critical point satisfies the stationarity part of the Karush-Kuhn-Tucker conditions with estimates ${\lambda }_i=\mu /c_i(x)>0$ and perturbed complementary slackness ${\lambda }_i{c}_i(x)=\mu$.

For symmetric $A$, every nonzero stationary point of this barrier lies in a negative-eigenvalue eigenspace: $Ax=\lambda x$, $\lambda <0$, $\Vert x{\Vert }^2=1+\mu /\lambda$. The origin is also stationary for every positive $\mu$ and must be considered separately. On a fixed nonzero branch, the Hessian has radial eigenvalue $4{\lambda }^2\Vert x{\Vert }^2/\mu$ and transverse eigenvalues $2({\lambda }_i-\lambda )$. A simple lowest negative eigenvalue gives a minimizing branch with a condition number growing as $1/\mu$; repeated lowest eigenvalues produce zero transverse curvature instead. Nevertheless the minimizing radius tends to one and the limiting vectors solve the constrained quadratic problem. The condition number measures numerical difficulty, not failure of the approximation principle.

Suppose strictly feasible stationary points of $f(x)-{\mu }_k{\sum }_j\log c_j(x)$ form a bounded sequence. Every cluster point satisfying the linear independence constraint qualification is a KKT point. Inactive multipliers tend to zero. The active gradients near the limit have uniformly bounded dual vectors, so pairing barrier stationarity with those vectors bounds the active multipliers. A convergent multiplier subsequence then gives stationarity, nonnegative multipliers and complementary slackness, since ${\lambda }_{k,j}{c}_j(x_k)={\mu }_k\to 0$. A stationarity residual tending to zero is also allowed.

This conclusion is stationarity, not global optimality. If the subproblems are globally minimized and strictly feasible points are dense in the feasible set, any cluster point is instead a global minimizer. Compare each barrier minimum with a fixed strictly feasible point; boundedness near the cluster point bounds $\log c_j(x_k)$ above. Taking the limit gives the objective comparison on strictly feasible points, then continuity and density extend it to all feasible points.

Let $c_i$ be concave and continuously differentiable, and let a strict feasible point $z$ satisfy $\delta ={\min }_i{c}_i(z)>0$. Suppose $x_k\to x_{\ast }$, $c_i(x_k)>0$, ${\mu }_k\to 0$, and the logarithmic barrier residuals $r_k=\nabla f(x_k)-{\sum }_i{\lambda }_{k,i}\nabla c_i(x_k)$ tend to zero, where ${\lambda }_{k,i}={\mu }_k/c_i(x_k)$. By concavity,The right-hand side is bounded when $\nabla f$ is continuous near $x_{\ast }$. Thus the nonnegative Lagrange multiplier estimates are bounded. A convergent subsequence gives ${\lambda }_{\ast }\ge 0$; passing to the limit in stationarity and ${\lambda }_{k,i}{c}_i(x_k)={\mu }_k$ proves the Karush-Kuhn-Tucker conditions at $x_{\ast }$. An inactive optimization constraint has $c_i(x_{\ast })>0$ and hence ${\lambda }_{k,i}\to 0$ even without selecting a subsequence. With a convex objective function, these conditions certify a global minimum. This result is conditional on the stated residual and primal convergence assumptions; it does not assert that every barrier subproblem has a minimizer.

Near an active constraint with limiting Lagrange multiplier ${\lambda }_i^{\ast }>0$, a convergent barrier sequence has $c_i(x_k)\sim {\mu }_k/{\lambda }_i^{\ast }$. Its Hessian matrix therefore contains the termFor a nonzero constraint gradient, this gives large curvature in a direction normal to the boundary. If a nonzero common tangent direction retains positive curvature of order one, while the remaining terms stay bounded, the condition number typically grows like $1/{\mu }_k$. This is a mechanism rather than a universal rate: there may be no tangent direction, all curvatures may grow together, or an active constraint may have zero limiting Lagrange multiplier.