Skip to main content
Log in

The Hotelling bi-matrix game

  • Original Paper
  • Published:
Optimization Letters Aims and scope Submit manuscript

Abstract

We study the pure equilibrium set for a specific symmetric finite game in strategic form, referred to as the Hotelling bi-matrix game. General results that guarantee non-emptiness of this set (for all parametric values) do not seem to exist. We prove non-emptiness by determining the pure equilibrium set. In this proof so-called demi-modality properties of the conditional payoff functions play an important role.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

References

  1. Abudaldah, N., Heijman, W., Heringa, P., von Mouche, P.H.M.: Return of the ice cream men. A discrete Hotelling game. Wass working paper no. 11, Wageningen University, Wageningen (2015)

  2. Chamberlin, E.: The Theory of Monopolistic Competition, 3 edn. Harvard University Press (1938)

  3. Gabszewicz, J.J., Thisse, J.F.: Location. In: Aumann, R.J., Hart, S. (eds.) Handbook of Game Theory with Economic Applications, pp. 281–304. Elsevier, Amsterdam (1992)

    Chapter  Google Scholar 

  4. Hotelling, H.: Stability in competition. Econ. J. 39(153), 41–57 (1929)

    Article  Google Scholar 

  5. Huang, Z.: The mixed strategy equilibrium of the three-firm location game with discrete location choices. Econ. Bull. 31(3), 2109–2116 (2011)

    Google Scholar 

  6. Iimura, T., Watanabe, T.: Existence of a pure strategy equilibrium in finite symmetric games where payoff functions are integrally concave. Discret. Appl. Math. 166, 26–33 (2014)

    Article  MathSciNet  MATH  Google Scholar 

  7. Prisner, E.: Best response digraphs for two location games on graphs. In: Petrosjan, L.A., Zenkevich N.A. (eds.) Contributions to Game Theory and Management, vol. IV, pp. 378–388. Saint Petersburg (2011)

Download references

Acknowledgments

The article benefited from the comments of two anonymous referees and from a discussion Pierre von Mouche had with Takahiro Watanabe and Takuya Iimura.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Pierre von Mouche.

Calculations

Calculations

Lemma 6

  1. 1.

    Suppose \(w \ne 1\).

    1. (a)

      If \(x \ge 1\), then \( f_1^{(x)}(x-1) = \frac{1-w^x}{1-w}\) and \(f_1^{(x-1)}(x) = \frac{1- w^{n-x+1} }{1-w}\). If \(x \ge 0\), then \( f_1^{(x)}(x) = \frac{1 +w - w^{x+1} - w^{n-x+1} }{2(1-w)}\).

    2. (b)

      If \(x \ge 1\), then \(\Delta f_1^{(x)}(x) = \frac{-1 + w + 2 w^x - w^{x+1} - w^{n-x+1} }{2(1-w)}\).

  2. 2.
    1. (a)

      If \(w \ne 1\) and \(\frac{n}{2} < x \le n\), then \( \Delta f_1^{(x)}(x) < 0\).

    2. (b)

      If \(w =1\), then \(\Delta f_1^{(x)}(x) = \frac{n+1}{2} - x\).

    3. (c)

      If \(w \ne 1\) and \(x \ge 1\), then \( \Delta f_1^{(x-1)}(x) = \frac{ 1 - w + w^x + w^{n-x+2} - 2 w^{n-x+1} }{ 2(1-w) }\).

  3. 3.

    If \( \frac{n}{2} < x <n\), then \(f_1^{(x)}(x+s) - f_1^{(x)}(x-s) < 0 \; (1 \le s \le n-x)\).

Proof

1a. Apply the formula for \(f_1\) in Sect. 2.

  1. 1b.

    By part 1a.

  2. 2a.

    Use part 1b and, using \(w \in ] {0},{1} \, [ \) and \( x > \frac{n}{2}\), note that \( 1- w - 2 w^x + w^{x+1} + w^{n-x+1} > 1-w - 2 w^x + w^{x+1} + w^{\frac{n}{2}+1} = (w^{\frac{n}{2}+1} - w^x) + (1-w^x) (1-w) > 0 \,+\, 0 = 0\).

  3. 2b.

    As by the formula for \(f_1\) we have \(\Delta f_1^{(x)}(x) = f_1(x,x) - f_1(x-1,x) = \frac{n+1}{2} -x\).

  4. 2c.

    A direct consequence of part 1a.

  5. 3.

    The locations that contribute to \(f_1^{(x)}(x+s)\) are those in \(V_+ := \{ x + \lfloor \frac{s+1}{2} \rfloor , \ldots , x+s-1, x+s, x+s+1, \ldots , n \}\). The locations that contribute to \(f_2^{(x)}(x-s)\) are those in \(V_- := V_{-+} \cup V_{--}\) where \( V_{-+} = \{ x - \lfloor \frac{s+1}{2} \rfloor , \ldots , x-s+1, x-s, x-s-1, \ldots , n - 2\lfloor \frac{s+1}{2}\rfloor \}\) and \(V_{--} = \{ n - 2 \lfloor \frac{s+1}{2} \rfloor -1, \ldots , 1, 0 \}\). The contribution of \(V_+\) to \(f_1^{(x)}(x+s)\) is the same as that of \(V_{-+}\) to \(f_1^{(x)}(x-s)\). As \(V_{--} \ne \emptyset \) the desired result follows.\(\square \)

Lemma 7

  1. 1.

    If \(x_1, x_2 \in H\) with \(x_2 \ge 1\), then \( \Delta f_1^{(x_1)}(x_2) = - \Delta f_1^{(n-x_1)}(n-x_2+1)\).

  2. 2.

    If \(x_1,x_2 \in H\) with \(x_2-x_1 \ge 2\), then \( \Delta f_1^{(x_1)}(x_2) = - q_{n-x_2+1;n-x_1}(w)\).

Proof

  1. 1.

    \( \Delta f_1^{(x_1)}(x_2) = f_1(x_2,x_1) - f_1(x_2-1,x_1) = f_1(n-x_2,n-x_1) - f_1(n-x_2+1,n-x_1) = - \Delta f_1^{(n-x_1)}(n-x_2+1)\).

  2. 2.

    By part 1, \( \Delta f_1^{(x_1)}(x_2) = -\Delta f_1^{(n-x_1)}(n-x_2+1)\). As \(n-x_2+1 < n -x_1\), we obtain \( \Delta f_1^{(x_1)}(x_2) = - q_{n-x_2+1;n-x_1}(w)\). \(\square \)

Lemma 8

Consider \((x_1,x_2) \in H \times H\) with \(0< x_1 < x_2 < n\) and \(x_2-x_1 \ge 2\).

  1. 1.

    Suppose \(x_2-x_1\) is even.

    1. (a)

      \( 0 \ge \Delta f_1^{(x_2)}(x_1+1) \; \Leftrightarrow \; w^{\frac{3x_1-x_2+2}{2}} \le \frac{1}{2}\);

    2. (b)

      \( \Delta f_1^{(x_2)}(x_1) \ge 0 \; \Leftrightarrow \; w^{\frac{3x_1-x_2}{2}} \ge \frac{1}{2}\);

    3. (c)

      \( 0 \ge \Delta f_1^{(x_1)}(x_2+1) \Leftrightarrow w^{\frac{2n-3x_2+x_1}{2}} \ge \frac{1}{2}\);

    4. (d)

      \( \Delta f_1^{(x_1)}(x_2) \ge 0 \Leftrightarrow w^{\frac{2n-3x_2+x_1+2}{2}} \le \frac{1}{2}\).

  2. 2.

    Suppose \(x_2-x_1\) is odd.

    1. (a)

      \(0 \ge \Delta f_1^{(x_2)}(x_1+1) \; \Leftrightarrow \; w^{\frac{3x_1-x_2+3}{2}} \le \frac{1}{2}\);

    2. (b)

      \( \Delta f_1^{(x_2)}(x_1) \ge 0 \; \Leftrightarrow \; w^{\frac{3x_1-x_2-1}{2}} \ge \frac{1}{2}\);

    3. (c)

      \( 0 \ge \Delta f_1^{(x_1)}(x_2+1) \Leftrightarrow w^{\frac{2n-3x_2+x_1-1}{2}} \ge \frac{1}{2}\);

    4. (d)

      \( \Delta f_1^{(x_1)}(x_2) \ge 0 \Leftrightarrow w^{\frac{2n-3x_2+x_1+3}{2}} \le \frac{1}{2}\).

Proof

Proposition 2, (4) and Lemma 7(2) imply \(\Delta f_1^{(x_2)}(x_1+1) \le 0 \; \Leftrightarrow \; w^{x_1+1} \le \frac{1}{2} w^{ \lfloor \frac{x_2-x_1}{2} \rfloor }; 0 \le \Delta f_1^{(x_2)}(x_1) \; \Leftrightarrow \; w^{x_1} \ge \frac{1}{2} w^{ \lfloor \frac{x_2-x_1+1}{2} \rfloor }; \Delta f_1^{(x_1)}(x_2+1) \le 0 \; \Leftrightarrow \; w^{n-x_2} \ge \frac{1}{2} w^{ \lfloor \frac{x_2-x_1+1}{2} \rfloor }; 0 \le \Delta f_1^{(x_1)}(x_2) \; \Leftrightarrow \; w^{n-x_2+1} \le \frac{1}{2} w^{ \lfloor \frac{x_2-x_1}{2} \rfloor }\). This leads to the desired results. \(\square \)

Lemma 9

  1. 1.

    Suppose n is even.

    1. (a)

      \(\Delta f_1^{(p)} (p) = - \Delta f_1^{( p )} ( p + 1 ) = w^p -\frac{1}{2}\).

    2. (b)

      If \(p \ge 2\), then \(\Delta f_1^{(p)} (p-1) = - \Delta f_1^{(p )} ( p + 2 ) = w ( w^{p-2} - \frac{1}{2} ) \).

    3. (c)

      \(\Delta f_1^{(p+1)} (p) = - \Delta f_1^{( p-1 )} ( p + 1 ) = w^p - \frac{w}{2}\).

    4. (d)

      \(\Delta f_1^{(p-1)} (p) = - \Delta f_1^{(p+1)}(p+1) = - w^{p+1} + \frac{1}{2}\).

  2. 2.

    Suppose n is odd.

    1. (a)

      If \(p \ge 1\), then \(\Delta f_1^{(p+1)} (p) = - \Delta f_1^{( p )} ( p + 2 ) = w( w^{p-1} - \frac{1}{2} )\).

    2. (b)

      \(\Delta f_1^{(p+1)} (p+1) = - \Delta f_1^{( p )} ( p + 1 ) = \frac{w^{p+1}-1}{2}\).

Proof

It is sufficient to prove the statements for \(w \ne 1\). Well, use Proposition 2 and Lemmas 6(1b) and 7. \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

von Mouche, P., Pijnappel, W. The Hotelling bi-matrix game. Optim Lett 12, 187–202 (2018). https://doi.org/10.1007/s11590-015-0964-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11590-015-0964-6

Keywords

Navigation