Skip to main content

On the density for sums of independent exponential, Erlang and gamma variates


This paper re-examines the density for sums of independent exponential, Erlang and gamma random variables. By using a divided difference perspective, the paper provides a unified approach to finding closed-form formulae for such convolutions. In particular, the divided difference perspective for sums of Erlang variates suggests a new approach to finding the density for sums of independent gamma variates using fractional calculus.

This is a preview of subscription content, access via your institution.


  1. Akkouchi M (2008) On the convolution of exponential distributions. J Chungcheong Math Soc 21:501–510

    Google Scholar 

  2. Amari SV, Misra RB (1997) Closed-form expressions for distribution of sum of exponential random variables. IEEE Trans Reliab 46:519–522

    Article  Google Scholar 

  3. Atkinson KE (1989) An introduction to numerical analysis, 2nd edn. Wiley, New York

    MATH  Google Scholar 

  4. Baxter BJC, Brummelhuis R (2011) Functionals of exponential Brownian motion and divided differences. J Comput Appl Math 236:424–433

    MathSciNet  Article  Google Scholar 

  5. Baxter BJC, Iserles A (1997) On approximation by exponentials. Ann Numer Math 4:39–54

    MathSciNet  MATH  Google Scholar 

  6. Coelho CA (1998) The generalized integer gamma distribution—a basis for distributions in multivariate statistics. J Multivar Anal 64:86–102

    MathSciNet  Article  Google Scholar 

  7. Di Salvo F (2008) A characterization of the distribution of a weighted sum of gamma variables through multiple hypergeometric functions. Integr Transforms Spec Funct 19:563–575

    MathSciNet  Article  Google Scholar 

  8. Harrison PG (1990) Laplace transform inversion and passage-time distributions in Markov processes. J Appl Probab 27:74–87

    MathSciNet  Article  Google Scholar 

  9. Hildebrand FB (1956) Introduction to numerical analysis. McGraw-Hill, New York

    MATH  Google Scholar 

  10. Jameson GJO (2016) The incomplete gamma functions. Math Gaz 100:298–306

    MathSciNet  Article  Google Scholar 

  11. Jasiulewicz H, Kordecki W (2003) Convolutions of Erlang and of Pascal distributions with applications to reliability. Demonstr Math 36:231–238

    MathSciNet  MATH  Google Scholar 

  12. Khuong HV, Kong H-Y (2006) General expression for pdf of a sum of independent exponential random variables. IEEE Commun Lett 10:159–161

    Article  Google Scholar 

  13. Li C, Qian D, Chen Y (2011) On Riemann-Liouville and Caputo derivatives. Discret Dyn Nat Soc 2011:562494

    MathSciNet  MATH  Google Scholar 

  14. Mathai AM (1982) Storage capacity of a dam with gamma type inputs. Ann Inst Stat Math 34:591–597

    MathSciNet  Article  Google Scholar 

  15. Miller KS, Ross B (1993) An introduction to the fractional calculus and fractional differential equations. Wiley, New York

    MATH  Google Scholar 

  16. Nadarajah S, Kotz S (2005) On the linear combination of exponential and gamma random variables. Entropy 7:161–171

    MathSciNet  Article  Google Scholar 

  17. Oldham KB, Spanier J (1974) The fractional calculus. Academic Press, New York

    MATH  Google Scholar 

  18. Olver FWJ, Lozier DW, Boisvert RF, Clark CW (2010) NIST handbook of mathematical functions. UIT Cambridge, Cambridge

    MATH  Google Scholar 

  19. Ostrowski AM (1966) Solution of equations and systems of equations, 2nd edn. Academic Press, New York

    MATH  Google Scholar 

  20. Podlubny I (1999) Fractional differential equations. Academic Press, San Diego

    MATH  Google Scholar 

  21. Ross S (1997) Introduction to probability models, 6th edn. Academic Press Inc, New York

    MATH  Google Scholar 

  22. Samko SG, Kilbas AA, Marichev OI (1993) Fractional integrals and derivatives: theory and applications. Gordon and Breach, New York

    MATH  Google Scholar 

  23. Schatzman M (2002) Numerical analysis: a mathematical introduction. Clarendon Press/Oxford University Press, New York

    MATH  Google Scholar 

  24. Sim CH (1992) Point processes with correlated gamma interarrival times. Stat Probab Lett 15:135–141

    MathSciNet  Article  Google Scholar 

  25. Srivastava HM, Karlsson PW (1985) Multiple Gaussian hypergeometric series. Dover, New York

    MATH  Google Scholar 

Download references


The author would like to thank Corina Constantinescu and Wei Zhu for helpful discussions. In addition, the author is pleased to acknowledge the valuable comments and suggestions of the two anonymous reviewers.


No financial support was provided for the conduct of the research and/or preparation of this manuscript.

Author information



Corresponding author

Correspondence to Edmond Levy.

Ethics declarations

Conflict of interest

The author has no conflicts of interest to declare that are relevant to the content of this manuscript.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.



Proof of Proposition 6.1

We prove this in two parts:

(a) Examine the differential of \(f[a_1,a_2, \dots , a_k]\) w.r.t. \(a_1\) as a limit as follows:

$$\begin{aligned} \begin{aligned} \frac{\partial }{\partial a_1}f[a_1,a_2, \dots , a_k]&= \lim _{h \rightarrow 0}\left( \frac{f[a_1+h, a_2, \dots , a_k]-f[a_1, a_2, \dots , a_k]}{h}\right) \\&= \lim _{h \rightarrow 0}\left( \frac{f[a_1+h, a_2, \dots , a_k]-f[a_1, a_2, \dots , a_k]}{(a_1+h)-a_1}\right) \\&=\lim _{h \rightarrow 0} f[a_1, a_2, \dots , a_k,a_1+h] \\&=f[a_1^{(2)}, a_2, \dots , a_k]. \end{aligned} \end{aligned}$$

(b) Consider the derivative of \(f[u_1^1, \dots ,u_n^1,a_2, \dots , a_k]\), \(n>1\), w.r.t. \(a_1\) and where each argument \(u_i^1\) is a function of \(a_1\):

$$\begin{aligned} \begin{aligned} \frac{\partial }{\partial a_1}f[u_1^1, \dots ,u_n^1,a_2, \dots , a_k]&= \sum _{i=1}^n \frac{\partial }{\partial u_i^1}f[u_1^1, \dots ,u_n^1,a_2, \dots , a_k]\frac{du_i^1}{da_1}\\&= \sum _{i=1}^n f[u_i^1, u_1^1, \dots ,u_n^1, a_2, \dots , a_k]\frac{du_i^1}{da_1}, \end{aligned} \end{aligned}$$

using (a). Now define \(u_i^1=a_1\) for \(i=1, \dots , n\). Hence we have

$$\begin{aligned} \frac{\partial }{\partial a_1}f[a_1^{(n)},a_2, \dots , a_k]= nf[a_1^{(n+1)}, a_2, \dots , a_k]. \end{aligned}$$

Using (a) and (b), we therefore have

$$\begin{aligned} \begin{aligned} \frac{\partial ^2}{\partial a_1^2}f[a_1, a_2, \dots , a_k]&=\frac{\partial }{\partial a_1}f[a_1^{(2)},a_2, \dots , a_k]\\&= 2f[a_1^{(3)}, a_2, \dots , a_k]. \end{aligned} \end{aligned}$$

It follows then that (b) taken with (a), the proposition is proved. See also Hildebrand (1956, Ch. 2). \(\square \)

Proof of Lemma 6.1

We recall the general Leibniz rule for m differentiable functions \(g_i(x)\), \(i=1, \dots , m\):

$$\begin{aligned} \frac{\partial ^n}{\partial x^n}\{g_1(x) \dots g_m(x)\} =\sum \limits _{\begin{array}{c} r_1+\dots +r_m=n\\ r_j \ge 0 \end{array}}\frac{n!}{r_1! \dots r_m!}g_1^{(r_1)} \dots g_m^{(r_m)}, \end{aligned}$$

and note that

$$\begin{aligned} \frac{d^{k}}{d a^{k}}\Big (\frac{1}{a^n}\Big ) =\frac{(-1)^k(n+k-1)!}{(n-1)!a^{n+k}}, \quad \text { for } n \in {\mathbb {N}}. \end{aligned}$$

The partial derivative for the divided difference \(f[a_1, \dots , a_m]\) in the lemma is then found by first invoking its Lagrange polynomial representation as follows:

$$\begin{aligned} \frac{\partial ^k}{\partial a_1^{k_1} \dots \partial a_m^{k_m}}f[a_1, \dots , a_m]= & {} \frac{\partial ^k}{\partial a_1^{k_1} \dots \partial a_m^{k_m}}\Big \{\sum _{i=1}^m f(a_i)\prod \limits _{\begin{array}{c} q=1 \\ q\ne i \end{array}}^m\frac{1}{(a_i-a_q)}\Big \} \\= & {} \sum _{i=1}^m \frac{\partial ^{k_i}}{\partial a_i^{k_i}}\Big \{f(a_i)\prod \limits _{\begin{array}{c} q=1 \\ q\ne i \end{array}}^m\frac{k_q!}{(a_i-a_q)^{k_q+1}}\Big \} \\= & {} \sum _{i=1}^m k_i! \sum \limits _{\begin{array}{c} r_1+\dots +r_m=k_i\\ r_j \ge 0 \end{array}} \frac{f^{r_i}(a_i)}{r_i!}\prod \limits _{\begin{array}{c} q=1 \\ q\ne i \end{array}}^m\frac{(-1)^{r_q}(k_q+r_q)!}{(a_i-a_q)^{k_q+r_q+1}r_q!} \\&\text { (using the general Leibniz rule)} \\= & {} \sum _{i=1}^m k_i! \sum \limits _{\begin{array}{c} r_1+\dots +r_m=k_i\\ r_j \ge 0 \end{array}} \frac{(-1)^{k_i-r_i}f^{r_i}(a_i)}{r_i!}\prod \limits _{\begin{array}{c} q=1 \\ q\ne i \end{array}}^m\frac{(k_q+r_q)!}{(a_i-a_q)^{k_q+r_q+1}r_q!}. \end{aligned}$$

\(\square \)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Levy, E. On the density for sums of independent exponential, Erlang and gamma variates. Stat Papers (2021).

Download citation


  • Convolutions
  • Exponential variables
  • Erlang density
  • Gamma density
  • Divided differences
  • Fractional calculus

Mathematics Subject Classification

  • 60E05
  • 62E10
  • 26A33