Pith. sign in

REVIEW 3 major objections 4 minor 52 references

General Decentralized Stochastic Optimal Control via Change of Measure: Applications to the Witsenhausen Counterexample

T0 review · 3 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper derives two coupled nonlinear integral equations that any person-by-person optimal pair of strategies for the Witsenhausen counterexample must satisfy, and claims a unique fixed point in L2×L2.

desk verdict Real extension of the change-of-measure program with useful explicit integral equations and a fair numerical comparison, but the advertised fixed-point theorem is not established and the stated linearization fails at k2=1. read the letter →

arxiv 2509.11013 v1 pith:QUIDDKHZ submitted 2025-09-13 eess.SY cs.SY

Bhagyashri Telsang , Seddik Djouadi , Charalambos D. Charalambous This is my paper · ORCID
classification eess.SYcs.SY MSC 93E2093A1449K45
keywords decentralizedstochasticcontrolWitsenhausencounterexamplechangeofmeasureperson-by-personoptimalitynonlinearintegralequationsfixedpointtheoremGauss-Hermitequadraturenonclassicalinformation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper's aim is to give a unified derivation of optimality conditions for decentralized stochastic control problems with different information patterns, using a discrete-time Girsanov change of measure. For the Witsenhausen counterexample—the two-stage problem where the second controller does not see the first controller's observation—the paper shows that any person-by-person optimal strategy pair must solve two coupled nonlinear integral equations. It then claims a fixed point theorem that gives existence and uniqueness of solutions in L2×L2, turning a hard variational problem into an equation-solving one. Numerical solution by Gauss-Hermite quadrature produces the known phase behavior: in some parameter regions affine strategies are optimal, in others nonlinear signaling strategies win. A sympathetic reader would care because the counterexample is the canonical instance where classical dynamic programming fails, and the paper offers a concrete computational route to its optimal strategies.

What carries the argument

The engine is a discrete-time Girsanov / Radon-Nikodym change of measure: on a reference measure the state and observations are independent of the control actions, converting the decentralized control problem into a static team form. From this, the paper derives conditional stationary conditions and then specializes to the counterexample, where the Radon-Nikodym derivative is the Gaussian likelihood ratio exp(-(y1-x0-u1)^2/(2σ^2)) / exp(-y1^2/(2σ^2)). The resulting fixed point operator F: L2×L2 → L2×L2, defined by the right-hand sides of (4)-(5), is the object whose contraction (or invertibility at zero) is claimed to give existence and uniqueness.

What would settle it

Compute the spectrum of the linearized operator L_{0,0} (kernel ∇f at (γ1,γ2)=(0,0)) for a fixed (k^2, σ^2, σ_x^2). If the homogeneous equation z + ∫ ∇f(0,0) z dµ = 0 has a nonzero L2 solution, or the operator's spectral radius is ≥1, the claimed unique fixed point fails at those parameters. Alternatively, iterate F on (4)-(5) from two different starting pairs; convergence to different limits would disprove uniqueness.

Watch

Extended reading notes

Core claim

The central claim is that the person-by-person optimal strategies (γ1,γ2) of the Witsenhausen counterexample—with Gaussian noise and arbitrary law for the initial state—are exactly the solutions of the two coupled nonlinear integral equations (4) and (5). Equation (4) writes the first controller as the initial state minus a Gaussian-weighted correction that depends on how far the second controller's response sits from the first controller's signaling value; equation (5) says the second controller is the conditional expectation of the first controller's output under the Gaussian likelihood of the observed signal. Theorem III.4 asserts the corresponding nonlinear operator F has a unique fixed

Load-bearing premise

The load-bearing premise is that the linearized fixed-point operator at the zero strategy has no neutral direction—unity is not an eigenvalue—and that its Lipschitz constant can be brought below 1 by a reweighted norm; the text asserts both without computing them.

Editorial extensions

If this is right

  • If the fixed-point theorem is correct, computing the PbP-optimal strategies is reduced to solving two deterministic integral equations, so numerical quadrature replaces search over function spaces.
  • The equations are derived for arbitrary P_x0, so the same conditions cover non-Gaussian initial states, not just the standard Gaussian case.
  • The stationary conditions reduce to conditional expectations on the original measure, recovering the known form γ2(y1)=E[γ1(x0)|y1] and supplementing it with a first-controller equation (1).
  • The numerical results confirm that for k^2=1, σ_x^2=1 the optimal first controller is affine, while for k=0.2, σ_x=5 a 7-step signaling law outperforms affine, matching Witsenhausen's dichotomy.
  • The change-of-measure construction extends PbP/global stationarity to general discrete-time decentralized problems, giving a common derivation for problems with nonclassical information patterns.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the fixed point theorem is valid, the same integral-equation reduction should apply to other decentralized problems with one-step delayed sharing or nonclassical information patterns; the change-of-measure machinery is general, so analogous coupled equations could be derived for multi-stage variants.
  • The GHQ collocation scheme gives a way to map the parameter plane: for each (k^2, σ_x^2), the number and shape of signaling levels produced by solving the discretized system can serve as a numerical indicator of where affine vs nonlinear strategies are PbP-optimal.
  • The numerical observation that the steps are slightly sloped, not perfectly flat, suggests that the optimal first controller in the nonlinear regime may be a smooth saturating curve rather than an ideal quantizer; the paper's equations imply this but it is not proven.
  • A reader who wants to test the theorem before relying on it should run fixed-point iteration from multiple starting pairs; convergences to different limits would indicate that the uniqueness claim needs additional conditions.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper develops a discrete-time Girsanov/change-of-measure framework for decentralized stochastic optimal control, derives person-by-person (PbP) stationary conditions, and applies them to the Witsenhausen counterexample. The main analytical results are the coupled nonlinear integral equations (4)-(5) for PbP optimal strategies, a claimed fixed-point theorem (Theorem III.4) giving existence and uniqueness of solutions in L2×L2, and a Gauss-Hermite quadrature/collocation numerical scheme that is compared with prior numerical studies. The derivation of the necessary conditions from first-order variations is coherent; the advertised existence/uniqueness theorem is the principal theoretical contribution.

Significance. If Theorem III.4 were correct, the paper would provide a rigorous functional-analytic basis for solving the Witsenhausen counterexample through integral equations, complementing a substantial body of numerical work. The change-of-measure reduction and the explicit integral equations (4)-(5) are genuinely useful and appear to be derived correctly from the stationary conditions. The numerical comparison with known results is a useful practical contribution. However, the existence/uniqueness theorem is not proven as written: both proof routes in Section III-D fail, and the claimed eigenvalue condition is actually false at a parameter value used later in the paper. Since the abstract and conclusions advertise the fixed-point theorem as a central contribution, this is a load-bearing gap.

major comments (3)
  1. [Section III-D, after Eq. (106)] The contraction-mapping argument is invalid. The sentence 'the Lipschitz constant can be made less than 1 ... by using a weighted L2×L2-norm by simply dividing by (ℓ+1)' is a uniform rescaling of the norm; it does not change the operator norm and does not yield a contraction. No non-equivalent norm is constructed, and ℓ is not shown to be finite, let alone <1. The bound in (106) is not a contraction estimate. Hence the contraction route does not establish Theorem III.4(ii).
  2. [Section III-D, after Eq. (107)] The inverse-function-theorem route has a false premise. Linearizing G(γ)=γ+∫f dμ at (0,0) using (101)-(104), one obtains (G'(0)h)_1 = h1 - h1/k2 + E_φ[h2]/k2 and (G'(0)h)_2 = h2 + E_{P_x0}[h1]. For k2=1, any h=(h1,0) with E_{P_x0}[h1]=0 is in the kernel, so I+L is singular. This is exactly the parameter value k=1 used in Section IV-B.1. Moreover, the inverse function theorem would only give a solution for x0 sufficiently small in norm, not the claimed global existence/uniqueness for arbitrary P_x0. The sign mismatch between (96)/(107) and (99)-(105) should also be resolved, but the singularity calculation stands with the correct sign.
  3. [Theorem III.4 statement] The function space L2×L2 is never specified: L2 with respect to which measure? The operator (94)-(97) and the norms in (98) and (106) are not well-defined without this choice. In particular, the finiteness of the Lipschitz constant ℓ depends on the underlying measure and is not established. As stated, the uniqueness claim lacks a precise mathematical meaning.
minor comments (4)
  1. [Throughout] There are numerous typos and undefined notations: "fuct." after (68), "Witsehausen" in the abstract, "R ∞ ∞" in (103), and "¯γ1" appears in (113) without definition. The paper would benefit from a careful editorial pass.
  2. [Section IV-B, Table II] The discussion of the second parameter set in [13] reports a discrepancy (the authors find linear optima while [13] reports a mix of linear and signaling forms). This discrepancy deserves a more explicit explanation rather than a one-sentence statement.
  3. [Section IV-B, Tables I-III] The numerical scheme solves the discretized necessary conditions via fsolve/lsqnonlin; no convergence or global-optimality guarantee is provided. The paper should state more clearly that these are candidate PbP stationary strategies for the discretized equations, not proven optimizers.
  4. [Section V] The claim that the observation "for some parameter values linear strategies are PbP optimal" is "not documented in previous numerical studies" is overstated; e.g., Refs. [14] and [7] already discuss parameter regions where affine laws are optimal.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central integral equations are derived from first-order optimality conditions rather than assumed, and the cited change-of-measure framework is reproved in the paper.

full rationale

The paper's central derivation—the PbP optimal strategies (γ°_1, γ°_2) satisfying the coupled integral equations (4)–(5)—is obtained from the first-order stationary conditions in Theorem III.3, which differentiate the equivalent reference-measure payoff (92)–(93). These equations are not assumed as an ansatz and contain no fitted parameters; k2, σ, and P_x0 are problem inputs, and the conditional-expectation form of γ°_2 is derived from the stationarity condition rather than imposed. Theorem II.1, which supplies the change-of-measure engine, is proved in the paper via the Radon–Nikodym derivative and Bayes' rule, so the related self-citations [40], [42]–[44] are contextual rather than load-bearing. The numerical section solves (4)–(5) and compares against external benchmarks (Witsenhausen's affine and nonlinear costs, [9], [10], [13]), so the validation is not circular. The main weakness—Theorem III.4's contraction argument via norm rescaling and the local-only inverse-function theorem—is a soundness gap in the proof of the existence/uniqueness claim, not a case of a conclusion being equivalent to its input by construction. Hence no circularity step meeting the required evidentiary standard is present.

Assumptions & free parameters 1 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new physical entities. The reference measure constructed via change of measure is a mathematical device from prior work, not an independently falsifiable new entity. The main load-bearing assumptions are regularity and existence conditions, plus an unproved eigenvalue assertion in the fixed point theorem.

free parameters (1)
  • Hermite quadrature order n = 7
    Number of Gauss-Hermite nodes used in the numerical solution; chosen by the authors and affects accuracy and the number of signaling levels.
assumptions (6)
  • domain assumption Absolute continuity conditions Q_s << Phi_s and S_s << Psi_s, plus integrability of the Radon-Nikodym derivative (A2)
    Needed for Theorem II.1 change of measure; stated as assumptions in Section II-B.
  • domain assumption Existence of a PbP optimal strategy (A3)
    Theorem II.3 assumes a PbP optimal strategy exists; used to derive stationary conditions and then (4)-(5).
  • domain assumption Density positivity and differentiability conditions (a.i)-(a.iii)
    Theorem III.2 assumes f_v>0 and L2 differentiability to justify Gateaux conditions.
  • domain assumption Witsenhausen's existence theorem for optimal strategies
    Used in Theorem III.3(ii) to assert existence of the optimal strategies; cited from [7, Theorem 1].
  • ad hoc to paper Unity is not an eigenvalue of the linearized kernel at (0,0)
    Asserted by 'straightforward computations' in Theorem III.4 without proof; needed for the inverse function theorem argument.
  • standard math Inverse function theorem and Lagrange theorem
    Used in Theorem III.4; standard results but invoked beyond the local validity of the inverse function theorem.

how reviews work

0 comments
Cite this review

Pith. "Pith review of General Decentralized Stochastic Optimal Control via Change of Measure: Applications to the Witsenhausen Counterexample." pith.science (2026). https://pith.science/paper/QUIDDKHZ

@misc{pith2026250911013,
  author       = {Pith},
  title        = {Pith review of: General Decentralized Stochastic Optimal Control via Change of Measure: Applications to the Witsenhausen Counterexample},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QUIDDKHZ}},
  note         = {Machine review of arXiv:2509.11013}
}
read the original abstract

In this paper we present global and person-by-person (PbP) optimality conditions for general decentralized stochastic dynamic optimal control problems, using a discrete-time version of Girsanov's change of measure. The PbP optimality conditions are applied to the Witsenhausen counterexample to show that the two strategies satisfy two coupled nonlinear integral equations. Further, we prove a fixed point theorem in a function space, establishing existence and uniqueness of solutions to the integral equations. We also provide numerical solutions of the two integral equations using the Gauss Hermite Quadrature scheme, and include a detail comparison to other numerical methods of the literature. The numerical solutions confirm Witsehausen's observation that, for certain choices of parameters, linear or affine strategies are optimal, while for other choices of parameters nonlinear strategies outperformed affine strategies.

Figures

Figures reproduced from arXiv: 2509.11013 by the authors.

Figure 1
Figure 1. Witsenhausen’s counterexample: x0 : Ω → R is the initial state random variable (RV), v : Ω → R is a noise RV, x0 and v are independent, and x1 = x0 + γ1(y0), x2 = x1 − γ2(y1), y0 = x0, y1 = x1 + v. The objec￾tive is to minimize over (γ1, γ2) the payoff, J(γ1, γ2) △ = Eγ1,γ2 n k 2 (γ1(y0))2 + (x2) 2 o , k2 > 0. The counterexample is called linear-quadratic-Gaussian (LQG) if x0 ∈ G(0, σ2 x ) and v ∈ G(0, σ2 ), where G… view at source ↗
Figure 2
Figure 2. Comparison of the optimal control laws and the special [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 4
Figure 4. Comparison of the strategy of the first controller for [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Observation of the slight slope in quantizers – magni [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 3
Figure 3. Figure 3: Optimal first control law γ1 for parameters in [13] [PITH_FULL_IMAGE:figures/full_fig_p015_3.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references

  1. [1]

    Separation of estimation and control for discrete time systems,

    H. S. Witsenhausen, “Separation of estimation and control for discrete time systems,” inProceedings of the IEEE, pp. 1557–1566, 1971

    1971

  2. [2]

    Elliott, L

    R. Elliott, L. Aggoun, and J. Moore,Hidden Markov Models: estimation and Control. Springer, 1995

    1995

  3. [3]

    Bensoussan,Lecture on Stochastic Control, Lecture Notes in Math- ematics

    A. Bensoussan,Lecture on Stochastic Control, Lecture Notes in Math- ematics. Springer-Verlag, Berlin, 1982

    1982

  4. [4]

    Bensoussan,Stochastic Control of Partially Observable Systems

    A. Bensoussan,Stochastic Control of Partially Observable Systems. Cambridge University Press, 1992

    1992

  5. [5]

    Control of partially observed diffusions,

    R. J. Elliott and H. Yang, “Control of partially observed diffusions,” Journal on Optimization Theory and Applications, vol. 71, no. 3, pp. 485–501, 1991

    1991

  6. [6]

    Certain classes of nonlinear partially observable stochastic optimal control problems with explicit optimal control laws equivalent to LEQG/LQG problems,

    C. D. Charalambous and R. Elliott, “Certain classes of nonlinear partially observable stochastic optimal control problems with explicit optimal control laws equivalent to LEQG/LQG problems,”IEEE Transactions on Automatic Control, vol. 42, no. 4, pp. 482–497, 1997

    1997

  7. [7]

    A counterexample in stochastic optimal control,

    H.S.Witsenhausen, “A counterexample in stochastic optimal control,” SIAM Journal on Control, vol. 6, no. 1, pp. 131–147, 1968

    1968

  8. [8]

    Intractable problems in control theory,

    C. H. Papadimitriou and J. Tsitsiklis, “Intractable problems in control theory,”SIAM Journal on Control and Optimization, vol. 24, no. 4, pp. 639–654, 1986

    1986

Show all 52 references
  1. [9]

    The Witsenhausen counterexample: a hierarchical search approach for nonconvex optimization problems,

    J. T. Lee, E. Lau, and Y .-C. Ho, “The Witsenhausen counterexample: a hierarchical search approach for nonconvex optimization problems,” IEEE Transactons on Automatic Control, vol. 46, pp. 382–397, 2001

    2001

  2. [10]

    Numerical solutions to the Witsenhausen counterexample by approximating networks,

    M. Baglietto, T. Parisini, and R. Zoppoli, “Numerical solutions to the Witsenhausen counterexample by approximating networks,”IEEE Transactions on Automatic Control, vol. 46, pp. 1471–1477, Sep. 2001

    2001

  3. [11]

    Iterative source-channel coding approach to Witsenhausen’s counterexample,

    J. Karlsson, A. Gattami, T. J. Oechtering, and M. Skoglund, “Iterative source-channel coding approach to Witsenhausen’s counterexample,” in Proceedings of the American Control Conference, pp. 5348–5353, June 2011

    2011

  4. [12]

    Variations on the theme of the Witsenhausen counterexample,

    T. Basar, “Variations on the theme of the Witsenhausen counterexample,” inProceedings of the 47th IEEE Conference on Decision and Control, pp. 1614–1619, Dec 2008

    2008

  5. [13]

    An optimization approach to the Witsenhausen counterexample,

    W. M. McEneaney, S. H. Han, and A. Liu, “An optimization approach to the Witsenhausen counterexample,” inProceedings of the 50th IEEE Conference on Decision and Control and European Control Conference, pp. 5023–5028, Dec 2011

    2011

  6. [14]

    Witsenhausen’s counterexample: A view from optimal transport theory,

    Y . Wu and S. Verd ´u, “Witsenhausen’s counterexample: A view from optimal transport theory,” inProceedings of the 50th IEEE Conference on Decision and Control and European Control Conference, pp. 5732– 5737, Dec 2011. (a) γ1 is linear fork= 0.05, σ= 5, σx = 2 (b) γ1 is7−step ...

    2011

  7. [15]

    Optimization formulation and monotonic solution method for the Witsenhausen problem,

    W. M. McEneaney and S. H. Han, “Optimization formulation and monotonic solution method for the Witsenhausen problem,”Automatica, vol. 55, pp. 55 – 65, 2015

    2015

  8. [16]

    Some new numeric results concerning the Witsenhausen counterexample,

    V . Subramanian, L. Brink, N. Jain, K. V odrahalli, A. Jalan, N. Shinde, and A. Sahai, “Some new numeric results concerning the Witsenhausen counterexample,” inProceedings of the 56th Annual Allerton Conference on Communication, Control, and Computing (Allerton), pp. 413–420, 2018

    2018

  9. [17]

    Elements for a theory of teams,

    J. Marschak, “Elements for a theory of teams,”Management Science, vol. 1, no. 2, 1955

    1955

  10. [18]

    Team decision problems,

    R. Radner, “Team decision problems,”The Annals of Mathematical Statistics, vol. 33, no. 3, pp. 857–881, 1962

    1962

  11. [19]

    Marschak and R

    J. Marschak and R. Radner,Economic Theory of Teams. New Haven: Yale University Pres, 1972

    1972

  12. [20]

    Static team problems- part I: Sufficient conditions and the exponential cost criterion,

    J. Krainak, J. L. Speyer, and S. I. Marcus, “Static team problems- part I: Sufficient conditions and the exponential cost criterion,”IEEE Transactions on Automatic Control, vol. 27, no. 4, pp. 839–848, 1982

    1982

  13. [21]

    Static team problems-part II: Affine control laws, projections, algorithms, and the LEGT problem,

    J. Krainak, J. L. Speyer, and S. I. Marcus, “Static team problems-part II: Affine control laws, projections, algorithms, and the LEGT problem,” IEEE Transactions on Automatic Control, vol. 27, no. 4, pp. 848–859, 1982

    1982

  14. [22]

    Linear-Quadratic-Gaussian control with one-step-delay sharing pattern,

    B.-Z. Kurtaran and R. Sivan, “Linear-Quadratic-Gaussian control with one-step-delay sharing pattern,”IEEE Transactions on Automatic Con- trol, vol. 19, no. 5, pp. 571–574, 1974

    1974

  15. [23]

    Dynamic programming approach to decentralized stochastic control problems,

    T. Yoshikawa, “Dynamic programming approach to decentralized stochastic control problems,”IEEE Transactions on Automatic Control, vol. 20, no. 6, pp. 796–797, 1975

    1975

  16. [24]

    Solution of some nonclassical LQG stochastic decision problems,

    N. R. Sandell and M. Athans, “Solution of some nonclassical LQG stochastic decision problems,”IEEE Transactions on Automatic Control, vol. 19, no. 2, pp. 108–116, 1974

    1974

  17. [25]

    On delay sharing patterns,

    P. Varaiya and J. Walrand, “On delay sharing patterns,”IEEE Transac- tions on Automatic Control, vol. 23, no. 3, pp. 443–445, 1978

    1978

  18. [26]

    Decentralized optimal control of markov chains with a common past information,

    M. Aicardi, F. Davoli, and R. Minciardi, “Decentralized optimal control of markov chains with a common past information,”IEEE Transactions on Automatic Control, vol. 32, no. 11, pp. 1028–1031, 1987

    1987

  19. [27]

    Stochastic teams with nonclassical informa- tion revisited: when is an affine law optimal,

    R. Bansal and T. Basar, “Stochastic teams with nonclassical informa- tion revisited: when is an affine law optimal,”IEEE Transactions on Automatic Control, vol. 32, no. 6, pp. 554–559, 1987

    1987

  20. [28]

    A class of team problems with discrete action spaces: Optimality conditions based on multimodularity,

    P. R. de Waal and J. H. van Schuppen, “A class of team problems with discrete action spaces: Optimality conditions based on multimodularity,” SIAM Journal on Control and Optimization, vol. 38, no. 3, pp. 875–892, 2000

    2000

  21. [29]

    A convex characterization of distributed control problems in spatially invariant systems with communication constraints,

    B. Bamieh and P. V oulgaris, “A convex characterization of distributed control problems in spatially invariant systems with communication constraints,”Systems and Control Letters, vol. 54, no. 6, pp. 575–583, 2005

    2005

  22. [30]

    Optimal H2 decentralized control of cone causal spatially invariant systems,

    M. E. Raoufat and S. Djouadi, “Optimal H2 decentralized control of cone causal spatially invariant systems,” inProceedings of the American Control Conference, pp. 961–966, 2018

    2018

  23. [31]

    A state-space solution to the two-player optimal control problems,

    L. Lessard and S. Lall, “A state-space solution to the two-player optimal control problems,” inProceedings of 49th Annual Allerton Conference on Communication, Control and Computing, 2011

    2011

  24. [32]

    Control of distributed stochastic systems- introduction, problems, and approaches,

    J. H. van Schuppen, “Control of distributed stochastic systems- introduction, problems, and approaches,” inInternational Proceedings of the IFAC World Congress, 2011

    2011

  25. [33]

    Control of distributed systems: Tutorial and overview,

    J. H. van Schuppen, O. Boutin, P. L. Kempker, J. Komenda, T. Masopust, N. Pambakian, and A. C. M. Ran, “Control of distributed systems: Tutorial and overview,”European Journal on Control, pp. 579–602, 2012

    2012

  26. [34]

    Decentralized stochastic control with partial history sharing: A common information approach,

    A. Nayyar, A. Mahajan, and D. Teneketzis, “Decentralized stochastic control with partial history sharing: A common information approach,” IEEE Transactions on Automatic Control, vol. 58, no. 7, pp. 1644–1658, 2013

    2013

  27. [35]

    Network topology and communication-computation tradeoffs in decentralized optimization,

    A. Nedic, A. Olshevsky, and M. G. Rabbat, “Network topology and communication-computation tradeoffs in decentralized optimization,” Proceedings of the IEEE, vol. 106, no. 5, pp. 953–976, 2017

    2017

  28. [36]

    Team decision theory and information structures in optimal control problems-part I,

    Y .-C. Ho and K.-C. Chu, “Team decision theory and information structures in optimal control problems-part I,”IEEE Transactions on Automatic Control, vol. 17, no. 1, pp. 15–22, 1972

    1972

  29. [37]

    On the equivalence of information structures in static and dynamic teams,

    Y .-C. Ho and K.-C. Chu, “On the equivalence of information structures in static and dynamic teams,”IEEE Transactions on Automatic Control, vol. 18, no. 2, pp. 187–188, 1973

    1973

  30. [38]

    Coordination coding with causal decoder for vector-valued Witsenhausen counterexample setups,

    O. J. Tobias and M. L. Treust, “Coordination coding with causal decoder for vector-valued Witsenhausen counterexample setups,” inProceedings of the 2019 IEEE Information Theory Workshop (ITW), August 2024

    2019

  31. [39]

    Equivalent stochastic control problems,

    H. Witsenhausen, “Equivalent stochastic control problems,”Mathematics of Control Signals and Systems, vol. 1, pp. 3–11, 1988

    1988

  32. [40]

    Equivalence of decentralized stochastic dynamic decision systems via Girsanov’s measure transfor- mation,

    C. D. Charalambous and N. U. Ahmed, “Equivalence of decentralized stochastic dynamic decision systems via Girsanov’s measure transfor- mation,” inProceedings of the 53rd IEEE Conference on Decision and Control, pp. 439–444, 2014

    2014

  33. [41]

    Liptser and A

    R. Liptser and A. Shiryayev,Statistics of Random Processes Vol.1. Springer-Verlag New York, 1977

    1977

  34. [42]

    Centralized versus decentralized optimization of distributed stochastic differential decision systems with different information structures-Part I: A general theory,

    C. D. Charalambous and A. N. U., “Centralized versus decentralized optimization of distributed stochastic differential decision systems with different information structures-Part I: A general theory,”IEEE Trans- actions on Automatic Control, vol. 62, no. 3, pp. 1194–1209, 2017

    2017

  35. [43]

    Centralized versus decen- tralized optimization of distributed stochastic differential decision sys- tems with different information structures—part II: Applications,

    C. D. Charalambous and N. U. Ahmed, “Centralized versus decen- tralized optimization of distributed stochastic differential decision sys- tems with different information structures—part II: Applications,”IEEE Transactions on Automatic Control, vol. 63, no. 7, pp. 1913–1928, 2018

    1913

  36. [44]

    Decentralized optimality conditions of stochastic differential decision problems via Girsanov’s measure transformation,

    C. D. Charalambous, “Decentralized optimality conditions of stochastic differential decision problems via Girsanov’s measure transformation,” Mathematics of Control, Signals and Systems (MCSS), vol. 28, no. 3, pp. 1–55, 2016

    2016

  37. [45]

    Stochastic teams with nonclassical information revisited: When is an affine law optimal?,

    R. Bansal and T. Basar, “Stochastic teams with nonclassical information revisited: When is an affine law optimal?,”IEEE Transactions on Automatic Control, vol. 32, pp. 554–559, June 1987

    1987

  38. [46]

    Grover,Information Structures, the Witsenhausen Counterexample, and Communicating Using Actions, pp

    P. Grover,Information Structures, the Witsenhausen Counterexample, and Communicating Using Actions, pp. 1–6. London: Springer London, 2013

    2013

  39. [47]

    Pr ´ecis d’analyse fonctionnelle,

    L. Lusternik and V . Sobolev, “Pr ´ecis d’analyse fonctionnelle,”Editions Mir, 1989

    1989

  40. [48]

    Zeros of the hermite polynomials and weights for gauss mechanical quadrature formula,

    R. Greenwood and J. Miller, “Zeros of the hermite polynomials and weights for gauss mechanical quadrature formula,”Bulletin of the American Mathematical Society, vol. 54, no. 8, pp. 765–769, 1947

    1947

  41. [49]

    Hermite polynomials and gauss quadrature,

    J. Pimbley, “Hermite polynomials and gauss quadrature,” 2017

    2017

  42. [50]

    Calculation of gauss quadrature rules,

    G. H. Golub and J. H. Welsch, “Calculation of gauss quadrature rules,” Mathematics of Computation, vol. 23, no. 106, pp. 221–s10, 1969

    1969

  43. [51]

    Piecewise continuous collocation for integral equations,

    K. Atkinson, I. Graham, and I. Sloan, “Piecewise continuous collocation for integral equations,”SIAM Journal on Numerical Analysis, vol. 20, no. 1, pp. 172–186, 1983

    1983

  44. [52]

    R. J. Elliott,Stochastic Calculus and Applications. Springer-Verlag, 1982. PLACE PHOTO HERE Bhagyashri Telsangreceived her Bachelor of Engi- neering in Instrumentation and Control from Mani- pal Institute of Technology in 2014, Master of Sci- ence in Systems and Control from D...

    1982

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.