Skip to content
Dynamic Corporate Transformation

Vol. II, Ch. 9 · Part 3. Enterprise Optimization Under Uncertainty · Week 11

Stochastic Enterprise Optimization

Learning outcomes

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

  1. Explain enterprise optimization under uncertainty and its departures from the deterministic doctrine.
  2. Formulate stochastic enterprise optimization problems on filtered probability spaces.
  3. Construct stochastic enterprise state equations as It^o diffusions.
  4. Model enterprise uncertainty using stochastic processes matched to its sources.
  5. Develop expectation-based enterprise objective functionals.
  6. Formulate stochastic control policies and distinguish adapted from anticipative rules.
  7. Interpret stochastic enterprise trajectories through moments, bands, and laws.
  8. Analyze enterprise optimization under probabilistic constraints.
  9. Implement stochastic optimization models computationally, with both error sources controlled.
  10. Prepare enterprise models for robust optimization under ambiguity.

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 Stochastic Enterprise Optimization

    Motivation for Stochastic Enterprise Optimization

  2. Enterprise Uncertainty

    Enterprise Uncertainty

  3. Stochastic Enterprise State Equations

    Stochastic Enterprise State Equations

  4. Stochastic Objective Functionals

    Stochastic Objective Functionals

  5. Stochastic Dynamic Programming

    Stochastic Dynamic Programming

  6. Stochastic Optimal Control

    Stochastic Optimal Control

  7. Risk-Neutral and Risk-Averse Enterprise Policies

    Risk-Neutral and Risk-Averse Enterprise Policies

  8. Chance-Constrained Enterprise Optimization

    Chance-Constrained Enterprise Optimization

  9. Computational Methods

    Computational Methods

  10. Enterprise Applications

    Enterprise Applications

  11. Preparation for Robust Optimization

    Preparation for Robust Optimization

  12. Chapter Summary

    Chapter Summary

  13. Worked Examples

    Worked Examples

  14. Exercises

    Exercises

  15. Notes and Sources

    Notes and Sources

On the map

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 9.

A. Concept checks

  1. 9.1
    For each row of Table (see book), propose the process model from Table (see book), defend the choice, and state one observable that would falsify it.
  2. 9.2
    Explain why certainty equivalence (Proposition (see book)(i)) is a property of the LQ structure and not of small noise: exhibit qualitatively how each departure—state-dependent σ\sigma, non-quadratic cost, risk aversion—breaks it, citing where the chapter proves each.

B. Mathematical exercises

  1. 9.3
    Derive the stationary variance σ2/(2κ)\sigma^{2}/(2\kappa) of the closed-loop OU process by applying It^o's formula to X2X^{2} and taking stationary expectations; recompute the Example (see book) thresholds.
  2. 9.4
    Solve the Lyapunov equation of Example (see book) for Σ∞\Sigma_{\infty} and size the ±2\pm 2-sigma workforce corridors; compare with the deterministic Chapter 6 corridors.
  3. 9.5
    Establish sharpness of the threshold in Proposition (see book): exhibit, for σˉ2\bar\sigma^{2} above it, a sequence of admissible policies along which the objective diverges to −∞-\infty slower than any quadratic bound, and conclude no quadratic verified value exists.
  4. 9.6
    Derive the weak-error expansion E[g(XTΔ)]−E[g(XT)]=c1Δ+O(Δ2)\E[g(X^{\Delta}_T)] - \E[g(X_T)] = c_1\Delta + O(\Delta^{2}) for Euler–Maruyama on the OU instance with quadratic gg by explicit computation of both sides, identifying c1c_1; reconcile with the measured bias ladder of Example (see book).

C. Computational exercises

  1. 9.7
    Prove the multidimensional certainty-equivalence and drag statement: for dX=(AX+Bu)dt+Σ dWdX = (AX + Bu)dt + \Sigma\,dW with quadratic cost, the optimal gain equals the deterministic CARE gain and the stationary value drag is tr(ΣΣ⊤P)/(2ρ)\mathrm{tr}(\Sigma\Sigma^{\top}P)/(2\rho), via the matrix completion-of-squares in the stochastic HJB equation.
  2. 9.8
    Prove the entropic–robust duality used in \S(see book): −1θlog⁡EP[e−θJ]=min⁡Q{EQ[J]+1θ KL(Q ∥ P)}-\tfrac1\theta\log\E_{\Prob} [e^{-\theta J}] = \min_{\mathbb{Q}}\{\E_{\mathbb{Q}}[J] + \tfrac1\theta\,\mathrm{KL}(\mathbb{Q}\,\Vert\,\Prob)\}, with the minimizing tilt identified.

D. Enterprise applications

  1. 9.9
    Reproduce Example (see book); add a control variate built from the deterministic value and report the variance-reduction factor; then add antithetic pairs and compare.
  2. 9.10
    Solve the stationary stochastic HJB equation for Example (see book) on a grid with the diffusion term included exactly (rather than the variance-penalty proxy), using a monotone centered/upwind hybrid; compare the computed policy with the closed-form proxy and quantify the proxy's error in gain and intercept.
  3. 9.11
    Meridian treasury mandate: combine Examples (see book) and (see book) into a two-constraint design—stationary breach $\le 1 $-1.
  4. 9.12 ★
    Extend the chapter's framework to jump–diffusion enterprise states dX=f dt+σ dW+γ dNdX = f\,dt + \sigma\,dW + \gamma\,dN (compensated Poisson NN): derive the integro-diffe…

Downloads

All three companions consume the same seeded engine (26209), so their numbers agree by construction — the MFMF convention, carried forward.