pith. sign in

arxiv: 2511.06156 · v4 · submitted 2025-11-08 · 🧮 math.OC

A Note on Optimal Product Pricing

Maximilian Schaller , Stephen Boyd This is my paper

Pith reviewed 2026-05-17 23:14 UTC · model grok-4.3

classification 🧮 math.OC
keywords product pricingprofit maximizationprice elasticitycross-elasticityconvex-concave procedureminorization-maximizationnonlinear programming
0
0 comments X

The pith

The product pricing problem with self- and cross-elasticities is solved by maximizing a sum of convex and concave functions using local optimization methods.

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

This paper examines how to set prices for multiple products to maximize total profit when demand depends on both a product's own price and the prices of other products. It reformulates the resulting optimization problem as maximizing the sum of a convex function and a concave function. The authors then test three different computational methods for finding good solutions and observe that they all reach the same point regardless of where they start. This suggests the solutions may be the best possible ones in practice.

Core claim

We consider the problem of choosing prices of a set of products so as to maximize profit, taking into account self-elasticity and cross-elasticity, subject to constraints on the prices. We show that this problem can be formulated as maximizing the sum of a convex and concave function. We compare three methods for finding a locally optimal approximate solution. In numerical examples all three converge reliably to the same local maximum, independent of the starting prices, leading us to believe that the prices found are likely globally optimal.

What carries the argument

The reformulation of the profit maximization objective as the sum of a convex function and a concave function, which enables the use of the convex-concave procedure, a minorization-maximization method, and general nonlinear programming to find local solutions.

If this is right

  • The convex-concave procedure solves the problem through a short sequence of convex optimization problems.
  • The custom minorization-maximization method reduces each step to solving a quadratic program.
  • General purpose nonlinear programming solvers can be applied directly to the formulated problem.
  • All three methods produce the same prices independent of the initial guess in the tested numerical examples.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The observed reliability across methods may indicate that the profit landscape for such pricing problems has few or no poor local maxima in practice.
  • Similar reformulations could be explored for other allocation problems where objectives combine convex and concave terms in decision variables.

Load-bearing premise

The assumption that consistent convergence to the same local maximum from varied starting points in numerical examples implies that the solution is globally optimal.

What would settle it

An instance where starting any of the three methods from a different initial price vector produces a feasible price set with strictly higher profit than the previously found maximum.

Figures

Figures reproduced from arXiv: 2511.06156 by Maximilian Schaller, Stephen Boyd.

Figure 1
Figure 1. Figure 1: Modeling and solving the PPP with CVXPY. The dimensions n, m and the data pi min, pi max, rnom, knom, E, C are given. 13 [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Profit and price changes for n = 1280 and m = 256. groups of related products that might be substitutes or complements. We sample the self￾elasticities Eii between −3.0 and −1.0, and the cross-elasticities Eij (within each block) between −0.05 and 0.05. We set the nominal revenue per product r nom i between 1.0 and 5.0, and the nominal cost to κ nom i = 0.9r nom i , i.e., a nominal profit margin of 10%. Co… view at source ↗
Figure 3
Figure 3. Figure 3: Solve times for different problem sizes, with CCP, QMM, and NLP. effect of dealing with a quadratic program at each iteration of QMM (and being able to use a specialized solver) appears to outweigh the effect of larger approximation errors. In fact, QMM took around 3–5 iterations for all problem sizes, just slightly more than the 3–4 iterations required by CCP. These scaling results were insensitive to the… view at source ↗
read the original abstract

We consider the problem of choosing prices of a set of products so as to maximize profit, taking into account self-elasticity and cross-elasticity, subject to constraints on the prices. We show that this problem can be formulated as maximizing the sum of a convex and concave function. We compare three methods for finding a locally optimal approximate solution. The first is based on the convex-concave procedure, and involves solving a short sequence of convex problems. Another one uses a custom minorization-maximization method, and involves solving a sequence of quadratic programs. The final method is to use a general purpose nonlinear programming method. In numerical examples all three converge reliably to the same local maximum, independent of the starting prices, leading us to believe that the prices found are likely globally optimal.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper formulates the profit-maximization pricing problem with self- and cross-elasticities as maximizing a sum of a convex function and a concave function (a DC program). It compares three local solvers: the convex-concave procedure (CCCP), a custom minorization-maximization algorithm that reduces to quadratic programs, and a general-purpose nonlinear programming solver. Numerical examples show that all three methods converge to the identical local maximum from varied initial prices, which the authors interpret as evidence that the solution is likely globally optimal.

Significance. The DC reformulation is a direct application of standard convex-analysis techniques and the numerical experiments demonstrate reliable convergence behavior across the three methods. If the observed agreement holds more generally, the work supplies practical, easily implementable local solvers for this pricing model; however, the global-optimality conjecture rests entirely on empirical agreement rather than a proof or bound on the number of stationary points.

major comments (2)
  1. [Abstract] Abstract: the assertion that the prices found are 'likely globally optimal' is supported solely by the observation that CCCP, the custom MM method, and general NLP reach the same point from different starts. DC programs can possess multiple distinct local maxima, and the manuscript provides neither a proof of unimodality nor an upper bound on the number of stationary points; the numerical evidence therefore does not rigorously establish global optimality.
  2. [Numerical examples] Numerical examples section: while the three solvers agree on the reported instances, the paper does not report the number of distinct local maxima found across a broader set of random initializations or problem instances, nor does it supply a theoretical argument (e.g., via strict concavity of one term or a uniqueness result) that would guarantee a single global maximizer.
minor comments (2)
  1. [Introduction] The manuscript would benefit from a brief statement clarifying that the DC formulation is non-convex and that the reported solutions are local maxima whose global status is conjectural.
  2. [Problem formulation] Notation for the elasticity matrix and the profit function should be introduced with explicit definitions before the DC reformulation is stated.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We are grateful to the referee for the detailed review and valuable suggestions. We respond to the major comments point by point below. We agree that the global optimality claim requires qualification and will revise the manuscript accordingly to emphasize the empirical nature of our observations.

read point-by-point responses

  1. Referee: [Abstract] the assertion that the prices found are 'likely globally optimal' is supported solely by the observation that CCCP, the custom MM method, and general NLP reach the same point from different starts. DC programs can possess multiple distinct local maxima, and the manuscript provides neither a proof of unimodality nor an upper bound on the number of stationary points; the numerical evidence therefore does not rigorously establish global optimality.

    Authors: We acknowledge that our statement in the abstract relies on numerical evidence rather than a theoretical guarantee. DC programs indeed can have multiple local optima in general. Our intent was to report the observed behavior in the examples considered. We will revise the abstract to state that the methods converge to the same point from different initial prices in the numerical examples, which leads us to conjecture that the solution is globally optimal for the instances tested. We will also add a brief discussion noting the absence of a general proof. revision: yes

  2. Referee: [Numerical examples] while the three solvers agree on the reported instances, the paper does not report the number of distinct local maxima found across a broader set of random initializations or problem instances, nor does it supply a theoretical argument (e.g., via strict concavity of one term or a uniqueness result) that would guarantee a single global maximizer.

    Authors: We agree that reporting results from a larger number of random initializations and instances would provide stronger support. In the current manuscript, we tested several starting points and observed consistent convergence, but we did not exhaustively enumerate all possible local maxima. As this is a concise note focused on the DC formulation and solver comparisons, a comprehensive theoretical analysis of uniqueness is beyond its scope. We will expand the numerical examples section to include additional tests with random initializations and explicitly state that the agreement suggests but does not prove global optimality in general. revision: yes

Circularity Check

0 steps flagged

No significant circularity; DC formulation and solver comparisons are independently derived

full rationale

The paper starts from a standard profit-maximization objective with linear elasticity demand and algebraically rewrites it as the sum of a convex function and a concave function, which is a direct mathematical decomposition rather than a self-referential definition. The three solution methods (convex-concave procedure, custom minorization-maximization, and general NLP) are then applied as off-the-shelf algorithms to this DC program; none of the methods or the convergence observation in the numerical examples is obtained by fitting parameters to the target result or by renaming an input. No load-bearing self-citation chain is invoked to justify uniqueness or global optimality; the authors simply report that the three local solvers agree across random starts and therefore conjecture global optimality on heuristic grounds. Because the core derivation remains self-contained and externally verifiable against the original profit expression, the analysis contains no circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work relies on standard convexity properties of the profit function under linear elasticity models; elasticities are treated as given inputs rather than fitted parameters.

axioms (1)
  • domain assumption Profit can be expressed as the sum of a convex function and a concave function based on the elasticity demand model.
    Invoked in the initial problem formulation to enable the convex-concave procedure.

pith-pipeline@v0.9.0 · 5418 in / 1155 out tokens · 51722 ms · 2026-05-17T23:14:32.363538+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Estimating Price Elasticity Matrices

    math.OC 2026-04 accept novelty 6.0

    Presents three methods to estimate price elasticity matrices from price and demand data by fitting a diagonal-plus-low-rank factor model via bi-convex optimization.

Reference graph

Works this paper leans on

54 extracted references · 54 canonical work pages · cited by 1 Pith paper

  1. [1]

    Allen and J

    R. Allen and J. Hicks. A reconsideration of the theory of value, II . Economica , 1(2):196--219, 1934

    work page 1934

  2. [2]

    R. Amir, J. Jin, G. Pech, and M. Tr \"o ge. Prices and deadweight loss in multiproduct monopoly. Journal of Public Economic Theory , 18(3):346--362, 2016

    work page 2016

  3. [3]

    Armstrong and J

    M. Armstrong and J. Vickers. Multiproduct pricing made simple. Journal of Political Economy , 126(4):1444--1471, 2018

    work page 2018

  4. [4]

    Agrawal, R

    A. Agrawal, R. Verschueren, S. Diamond, and S. Boyd. A rewriting system for convex optimization problems. Journal of Control and Decision , 5(1):42--60, 2018

    work page 2018

  5. [5]

    Bertsekas

    D. Bertsekas. Nonlinear programming. Journal of the Operational Research Society , 48(3):334--334, 1997

    work page 1997

  6. [6]

    Bajwa, B

    N. Bajwa, B. Fontem, and C. Sox. Optimal product pricing and lot sizing decisions for multiple products with nonlinear demands. Journal of Management Analytics , 3(1):43--58, 2016

    work page 2016

  7. [7]

    Ban and N

    G. Ban and N. Keskin. Personalized dynamic pricing with machine learning: H igh-dimensional features and heterogeneous elasticity. Management Science , 67(9):5549--5568, 2021

    work page 2021

  8. [8]

    Baron and R

    D. Baron and R. Myerson. Regulating a monopolist with unknown costs. Econometrica: Journal of the Econometric Society , pages 911--930, 1982

    work page 1982

  9. [9]

    R. Byrd, J. Nocedal, and R. Waltz. KNITRO : A n integrated package for nonlinear optimization. In Large-scale nonlinear optimization , pages 35--59. Springer, 2006

    work page 2006

  10. [10]

    M. Boiteux. Sur la gestion des monopoles publics astreints \`a l' \'e quilibre budg \'e taire. Econometrica, Journal of the Econometric Society , pages 22--40, 1956

    work page 1956

  11. [11]

    Boyd and L

    S. Boyd and L. Vandenberghe. Convex Optimization . Cambridge University Press, 2004

    work page 2004

  12. [12]

    Besbes and A

    O. Besbes and A. Zeevi. Dynamic pricing without knowing the demand function: Risk bounds and near-optimal algorithms. Operations research , 57(6):1407--1420, 2009

    work page 2009

  13. [13]

    A. Cournot. Principes de la T h \'e orie des R ichesses . Hachette, 1863

    work page

  14. [14]

    Diamond and S

    S. Diamond and S. Boyd. CVXPY : A P ython-embedded modeling language for convex optimization. Journal of Machine Learning Research , 17(83):1--5, 2016

    work page 2016

  15. [15]

    Douaioui, R

    K. Douaioui, R. Oucheikh, O. Benmoussa, and C. Mabrouki. Machine learning and deep learning models for demand forecasting in supply chain management: A critical review. Applied System Innovation (ASI) , 7(5), 2024

    work page 2024

  16. [16]

    Deng and C

    S. Deng and C. Yano. Joint production and pricing decisions with setup costs and capacity constraints. Management Science , 52(5):741--756, 2006

    work page 2006

  17. [17]

    C. Dye. Optimal joint dynamic pricing, advertising and inventory control model for perishable items with psychic stock effect. European Journal of Operational Research , 283(2):576--587, 2020

    work page 2020

  18. [18]

    Ferreira, B

    K. Ferreira, B. Lee, and D. Simchi-Levi. Analytics for an online retailer: D emand forecasting and price optimization. Manufacturing & service operations management , 18(1):69--88, 2016

    work page 2016

  19. [19]

    Guerreiro and J

    R. Guerreiro and J. Amaral. Cost-based price and value-based price: A re they conflicting approaches? Journal of Business & Industrial Marketing , 33(3):390--404, 2018

    work page 2018

  20. [20]

    arXiv preprint arXiv:2405.12762 (2 024)

    P. Goulart and Y. Chen. Clarabel: A n interior-point solver for conic programs with quadratic objectives. arXiv preprint arXiv:2405.12762 , 2024

    work page arXiv 2024

  21. [21]

    Gallego and G

    G. Gallego and G. Van Ryzin. Optimal dynamic pricing of inventories with stochastic demand over finite horizons. Management science , 40(8):999--1020, 1994

    work page 1994

  22. [22]

    T. Hastie. Generalized additive models. Statistical models in S , pages 249--307, 2017

    work page 2017

  23. [23]

    Herbon and E

    A. Herbon and E. Khmelnitsky. Optimal dynamic pricing and ordering of a perishable product under additive effects of price and time on demand. European Journal of Operational Research , 260(2):546--556, 2017

    work page 2017

  24. [24]

    Hartford, G

    J. Hartford, G. Lewis, K. Leyton-Brown, and M. Taddy. Deep IV : A flexible approach for counterfactual prediction. In International Conference on Machine Learning , pages 1414--1423. PMLR, 2017

    work page 2017

  25. [25]

    Houthakker

    H. Houthakker. The capacity method of quadratic programming. Econometrica, Journal of the Econometric Society , pages 62--87, 1960

    work page 1960

  26. [26]

    Hastie and R

    T. Hastie and R. Tibshirani. Generalized additive models. Statistical science , 1(3):297--310, 1986

    work page 1986

  27. [27]

    Johnson and D

    J. Johnson and D. Myatt. Multiproduct C ournot oligopoly. The RAND Journal of Economics , 37(3):583--601, 2006

    work page 2006

  28. [28]

    Kuhn and A

    H. Kuhn and A. Tucker. Nonlinear Programming . Springer, 2013

    work page 2013

  29. [29]

    Lipp and S

    T. Lipp and S. Boyd. Variations and extension of the convex--concave procedure. Optimization and Engineering , 17(2):263--287, 2016

    work page 2016

  30. [30]

    A. Lerner. The concept of monopoly and the measurement of monopoly power. The review of economic studies , 1(3):157--175, 1934

    work page 1934

  31. [31]

    Liu and J

    D. Liu and J. Nocedal. On the limited memory BFGS method for large scale optimization. Mathematical programming , 45(1):503--528, 1989

    work page 1989

  32. [32]

    Lewbel and K

    A. Lewbel and K. Pendakur. Tricks with H icks: T he EASI demand system. American Economic Review , 99(3):827--863, 2009

    work page 2009

  33. [33]

    Laffont and J

    J. Laffont and J. Tirole. A Theory of Incentives in Procurement and Regulation . MIT press, 1993

    work page 1993

  34. [34]

    Q. Li, P. Yu, and X. Wu. Managing perishable inventories in retailing: R eplenishment, clearance sales, and segregation. Operations Research , 64(6):1270--1284, 2016

    work page 2016

  35. [35]

    Mas-Colell, M

    A. Mas-Colell, M. Whinston, and J. Green. Microeconomic Theory , volume 1. Oxford University Press New York, 1995

    work page 1995

  36. [36]

    Mahato and G

    C. Mahato and G. Mahata. Optimal pricing and inventory decisions for perishable products with multivariate demand function under trade credit. Operations Research Forum , 4(2): A rticle 24, 2023

    work page 2023

  37. [37]

    Nocke and N

    V. Nocke and N. Schutz. Multiproduct-firm oligopoly: A n aggregative games approach. Econometrica , 86(2):523--557, 2018

    work page 2018

  38. [38]

    Nocedal and S

    J. Nocedal and S. Wright. Numerical Optimization . Springer, 2006

    work page 2006

  39. [39]

    O'Donoghue, E

    B. O'Donoghue, E. Chu, N. Parikh, and S. Boyd. Conic optimization via operator splitting and homogeneous self-dual embedding. Journal of Optimization Theory and Applications , 169(3):1042--1068, 2016

    work page 2016

  40. [40]

    W. Oi. A D isneyland dilemma: T wo-part tariffs for a M ickey M ouse monopoly. The Quarterly Journal of Economics , 85(1):77--96, 1971

    work page 1971

  41. [41]

    J. Perloff. Microeconomics . Pearson Education, 2009

    work page 2009

  42. [42]

    Samuelson

    P. Samuelson. Spatial price equilibrium and linear programming. The American economic review , 42(3):283--303, 1952

    work page 1952

  43. [43]

    Stellato, G

    B. Stellato, G. Banjac, P. Goulart, A. Bemporad, and S. Boyd. OSQP : A n operator splitting solver for quadratic programs. Mathematical Programming Computation , 12(4):637--672, 2020

    work page 2020

  44. [44]

    Y. Sun, P. Babu, and D. Palomar. Majorization-minimization algorithms in signal processing, communications, and machine learning. IEEE Transactions on Signal Processing , 65(3):794--816, 2016

    work page 2016

  45. [45]

    X. Shen, S. Diamond, Y. Gu, and S. Boyd. Disciplined convex-concave programming. In 2016 IEEE 55th conference on decision and control (CDC) , pages 1009--1014. IEEE, 2016

    work page 2016

  46. [46]

    Sirmans, D

    S. Sirmans, D. Macpherson, and E. Zietz. The composition of hedonic pricing models. Journal of real estate literature , 13(1):1--44, 2005

    work page 2005

  47. [47]

    Upasani and R

    A. Upasani and R. Uzsoy. Integrated production planning and pricing decisions in congestion-prone capacitated production systems. In Essays in Production, Project Planning and Scheduling: A Festschrift in Honor of Salah Elmaghraby , pages 29--68. Springer, 2013

    work page 2013

  48. [48]

    H. Uzawa. Iterative methods for concave programming. Studies in linear and nonlinear programming , 6:154--165, 1958

    work page 1958

  49. [49]

    H. Varian. Microeconomic Analysis , volume 3. Norton New York, 1992

    work page 1992

  50. [50]

    H. Varian. Intermediate Microeconomics with Calculus: A Modern Approach . WW Norton & Company, 2014

    work page 2014

  51. [51]

    X. Vives. Oligopoly Pricing: Old Ideas and New Tools . MIT Press (MA), 1999

    work page 1999

  52. [52]

    W \"a chter and L

    A. W \"a chter and L. Biegler. On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Mathematical programming , 106(1):25--57, 2006

    work page 2006

  53. [53]

    R. Wilson. Nonlinear Pricing . Oxford University Press, 1993

    work page 1993

  54. [54]

    Zellner, J

    A. Zellner, J. Kmenta, and J. Dreze. Specification and estimation of C obb- D ouglas production function models. Econometrica: Journal of the Econometric Society , pages 784--795, 1966

    work page 1966