Skip to main content
Log in

A representation theorem for entropies with the branching property

  • Research Papers
  • Published:
aequationes mathematicae Aims and scope Submit manuscript

Summary

Using a recent result by B. Ebanks on the functional equation

$$h(x,y) + h(x + y,z) = h(x,y + z) + h(y,z)$$

we derive a representation theorem for a large class of entropy functionals that exhibit the “branching property”.

LetV(Ω, F,m) be the set of probability densities on a non-atomic measure space {Ω,F,m} and\(\bar V\)(Ω,F,m) be the set of all simple probability densities. A functional Ф: (Ω,F,m) →R ∪ { − ∞, ∞} will be said to have thebranching property, if, given any setA ∈ F and any two functionsf, g ∈ V such that at least one of Ф(f) or Ф(g) is finite andf(ω) = g(ω) whenever ω ∈ Ω/A, then

$$\Phi (f) - \Phi (g) = \Psi (f_A ,g_A ),$$

wheref A is the restriction off to the setA and Ψ:L 1(A, F,m) ×L 1(A, F,m) →R ∪ {− ∞, ∞}.

Theorem 1.Given Ф: V(Ω,F,m) →R ∪ {−∞, ∞},\(\bar V\)(Ω,F,m) →R,If

  1. (i)

    Фhas the branching property

  2. (ii)

    Фis invariant under all metric endomorphisms

  3. (iii)

    (continuity) for any sequence of simple functions {si}, with si ↑ f we have (with ∥ · ∥ the L1 norm)

    $$\Phi \left( {\frac{{s_i }}{{\parallel s_i \parallel }}} \right) \to \Phi (f)$$

    then there exists h:[0, ∞) →R continuous on (0, ∞)with h(0) = 0such that Ф(f) = ∫ Ω h(f) d m.

Фis said to be “recursive” if, for any set A ∈ F and any two functions, f, g ∈ V such that f(ω) = g(ω) at each ω ∈ Ω/A and p:=∫ A f d m = A g d m >0,

$$\Phi (f) - \Phi (g) = p\left[ {\Phi \left( {\frac{{f\chi _A }}{p}} \right) - \Phi \left( {\frac{{g\chi _A }}{p}} \right)} \right],$$

where ϰ A is the characteristic function of the set A.

By strengthening (i) in Theorem 1 to “Ф is recursive” we obtain a new characterization of the Boltzmann—Shannon entropy.

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. Campbell, L. L.,Characterization of entropy of probability distributions on the real line. Inform. and Control21 (1972), 329–338.

    Google Scholar 

  2. Ebanks, G.,Measures of inset information on open domains—V: Branching measures. Manuscript, 1987.

  3. Forte, B. andBortone, C.,Non-symmetric entropies with the branching property. Utilitas Math.12 (1977), 3–24.

    Google Scholar 

  4. Forte, B. andNg, C. T.,Entropies with the branching property. Ann. Mat. Pura Appl.101 (1974), 355–373.

    Google Scholar 

  5. Forte, B. andSastri, C. C. A.,Is something missing in the Boltzmann entropy? J. Math. Phys.16 (1975), 1453–1456.

    Google Scholar 

  6. Halmos, P.,Lectures on ergodic theory. The Mathematical Society of Japan, Tokyo, 1956.

    Google Scholar 

  7. Martin, N. andEngland, J.,Mathematical theory of entropy. (Encyclopedia of Mathematics and Its Applications, Vol. 12). Addison-Wesley, London, 1981.

    Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Rights and permissions

Reprints and permissions

About this article

Cite this article

Forte, B., Hughes, W. A representation theorem for entropies with the branching property. Aeq. Math. 37, 219–232 (1989). https://doi.org/10.1007/BF01836445

Download citation

  • Received:

  • Accepted:

  • Issue Date:

  • DOI: https://doi.org/10.1007/BF01836445

AMS (1980) subject classification

Navigation