Skip to content
Dynamic Corporate Transformation

Vol. II, Ch. 4 · Part 1. Enterprise Optimization Foundations · Week 9

Nonlinear Enterprise Optimization

Learning outcomes

After completing this chapter, the reader should be able to:

  1. Formulate nonlinear enterprise optimization problems in the canonical smooth form, identifying which declared blocks carry the nonlinearity and of what structural type.
  2. Distinguish convex from non-convex models precisely, and state which clauses of Chapter 3's contract each declared instance retains, weakens, or forfeits.
  3. Analyze nonlinear enterprise objective functions through gradients, Hessians, and Taylor models, and read curvature as economic structure (synergy, saturation, scale).
  4. Construct nonlinear enterprise constraint systems, including genuine nonlinear equalities, and characterize the geometry of the resulting feasible regions.
  5. Identify local and global optima, explain why the distinction now carries governance weight, and deploy the instruments (relaxation bounds, multistart, structure) that manage the gap between them.
  6. Apply first-order necessary and second-order sufficient optimality conditions, including the critical-cone curvature test.
  7. Interpret the Karush–Kuhn–Tucker conditions in the nonlinear setting: what they certify, under which constraint qualifications they are necessary, and what their multipliers price.
  8. Evaluate nonlinear enterprise trade-offs—synergy against budgets, price against volume, scale against resilience—from the geometry of the declared instance.
  9. Implement nonlinear enterprise models computationally with SQP, interior-point, and augmented Lagrangian methods, and audit candidate solutions with KKT and second-order checks.
  10. Prepare enterprise models for dynamic optimization: parametrize decisions as policies and recognize trajectory structure as the next source of nonlinearity.

Reading guide

Work through the chapter in section order; the full development, proofs, and worked examples are in the book — this page indexes them and does not replace them.

  1. Motivation for Nonlinear Enterprise Optimization

    Motivation for Nonlinear Enterprise Optimization

  2. Nonlinear Objective Functions

    Nonlinear Objective Functions

  3. Nonlinear Constraint Systems

    Nonlinear Constraint Systems

  4. Local and Global Optimality

    Local and Global Optimality

  5. First-Order Optimality Conditions

    First-Order Optimality Conditions

  6. Second-Order Optimality Conditions

    Second-Order Optimality Conditions

  7. Karush–Kuhn–Tucker Conditions

    Karush–Kuhn–Tucker Conditions

  8. Computational Solution Algorithms

    Computational Solution Algorithms

  9. Enterprise Applications

    Enterprise Applications

  10. Sensitivity and Parametric Analysis

    Sensitivity and Parametric Analysis

  11. Preparation for Dynamic Optimization

    Preparation for Dynamic Optimization

  12. Worked Examples

    Worked Examples

  13. Chapter Summary

    Chapter Summary

  14. Exercises

    Exercises

  15. Notes and Sources

    Notes and Sources

AXIOM

This chapter is instrumented by:

Launch the module, load the chapter model, modify inputs, run the optimization, and compare against the worked examples in the book.

Exercises

12 exercises, grouped A concept checks · B mathematical · C computational · D enterprise applications. Starred (★) exercises are on the advanced track. Full solutions appear in the Instructor's Manual, Chapter 4.

A. Concept checks

  1. 4.1
    Restate the guarantee ledger of Table (see book) from memory, and for each lost guarantee name the section of this chapter that prices, bounds, or locally restores it.
  2. 4.2
    Explain in business terms why the two local optima of Example (see book) are both honest recommendations, and why a gradient-following funding process systematically produces one of them.

B. Mathematical exercises

  1. 4.3
    Compute gradient and Hessian of f(x)=5.2ln⁡(1+x1)+4.3ln⁡(1+x2)+0.3x1x2f(\x) = 5.2\ln(1+x_1) + 4.3\ln(1+x_2) + 0.3 x_1 x_2, verify the eigenvalues (−0.596,+0.004)(-0.596, +0.004) at (3.189,2.811)(3.189, 2.811), and find the locus in the positive quadrant where the Hessian is exactly singular.
  2. 4.4
    Verify every clause of the KKT system (see book) at the solution of Example (see book) using the stated numbers, and confirm LICQ by computing ∇g(e∗)\nabla g(\mathbf{e}^{*}).
  3. 4.5
    In Example (see book), derive the closed form p∗=c+1/εp^{*} = c + 1/\varepsilon for demand D0e−εpD_0 e^{-\varepsilon p} and cost cc, prove unimodality of profit on [c,∞)[c, \infty), and compute the cap multiplier λ∗(pˉ)\lambda^{*}(\bar p) as a function of the cap for pˉ<p∗\bar p < p^{*}.

C. Computational exercises

  1. 4.6
    Prove the second-order necessary condition (SONC): if x∗\x^{*} is a local minimizer at which ACQ holds and, for the critical direction considered, an appropriate multiplier vector exists, then d⊤∇xx2L d≥0\mathbf{d}^{\top}\nabla^2_{\x\x}\Lag\,\mathbf{d} \ge 0 for all d∈C(x∗)\mathbf{d} \in \mathcal{C}(\x^{*}).
  2. 4.7
    Prove that a finitely generated cone is closed: reduce by Carath'eodory to simplicial subcones and show each is a closed set, completing the argument of Lemma (see book).
  3. 4.8
    Derive Gordan's theorem from Lemma (see book): exactly one of (a) Ad<0A\mathbf{d} < \mathbf{0} solvable, (b) A⊤λ=0A^{\top}\boldsymbol{\lambda} = \mathbf{0}, λ≥0\boldsymbol{\lambda} \ge \mathbf{0}, λ≠0\boldsymbol{\lambda} \neq \mathbf{0} solvable; extend to the mixed form with equality rows used in the proof of Theorem (see book)(ii).

D. Enterprise applications

  1. 4.9
    Implement the audit function of Table (see book): inputs (declared functions, candidate triple), outputs (KKT residuals, active-set report, CQ rank check, reduced-Hessian eigenvalues, verdict).
  2. 4.10
    Run Algorithm (see book) on Example (see book) with 2020 uniform seeds: report the basin inventory, the fraction of seeds terminating at abstention, and the concave-envelope bracket; then repeat with seeds biased below a=1a = 1 to reproduce the timid-pilot pathology.
  3. 4.11
    Re-declare Example (see book) with the resilience architecture priced in: add the declared expected-disruption cost 0.4 min⁡(x,100−x)0 ⁣⋅ ⁣10.4\, \min(x, 100-x)^{0}\!\cdot\!\mathbb{1}-free smooth proxy 0.4 (1−(x/100)(1−x/100)⋅4)−10.4\,(1 - (x/100)(1-x/100)\cdot 4)^{-1} or a declared alternative, re-solve, and report where the concentration verdict reverses as the resilience weight grows.
  4. 4.12 ★
    Develop second-order theory without strict complementarity: state and prove the strong second-order sufficient condition and Robinson's strong regularity for the NEO class, and characterize the enterprise instances (zero-priced binding covenants) where the distinction is operative.

Downloads

All three companions consume the same seeded engine (26204), so their numbers agree by construction — the MFMF convention, carried forward.