Abstract
Convolutions of independent gamma variables are encountered in many applications such as insurance, reliability, and network engineering. Accurate and fast evaluations of their density and distribution functions are critical for such applications, but no open source, user-friendly software implementation has been available. We review several numerical evaluations of the density and distribution of convolution of independent gamma variables and compare them with respect to their accuracy and speed. The methods that are based on the Kummer confluent hypergeometric function are computationally most efficient. The benefit of employing the confluent hypergeometric function is further demonstrated by a renewal process application, where the gamma variables in the convolution are Erlang variables. We derive a new computationally efficient formula for the probability mass function of the number of renewals by a given time. An R package coga provides efficient C++ based implementations of the discussed methods and are available in CRAN.
Similar content being viewed by others
References
Abramowitz M, Stegun IA (1972) Handbook of mathematical functions with formulas, graphs, and mathematical tables, vol 9. Dover, New York
Akkouchi M (2005) On the convolution of gamma distributions. Soochow J Math 31(2):205–211
Barnabani M (2017) An approximation to the convolution of gamma distributions. Commun Stat Simul Comput 46(1):331–343
Bon JL, Pãltãnea E (1999) Ordering properties of convolutions of exponential random variables. Lifetime Data Anal 5(2):185–192
Crimaldi I, Di Crescenzo A, Iuliano A, Martinucci B (2013) A generalized telegraph process with velocity driven by random trials. Adv Appl Probab 45(4):1111–1136
Di Salvo F (2008) A characterization of the distribution of a weighted sum of gamma variables through multiple hypergeometric functions. Integral Transforms Spec Funct 19(8):563–575. https://doi.org/10.1080/10652460802045258
Furman E, Landsman Z (2005) Risk capital decomposition for a multivariate dependent gamma portfolio. Insurance Math Econ 37(3):635–649
Galassi M, Davies J, Theiler J, Gough B, Jungman G (2009) GNU Scientific Library: reference manual, 3rd edn. Network Theory Ltd, Bristol
Gupta AK, Nagar DK (2018) Matrix variate distributions. Chapman and Hall, London
Hu C, Pozdnyakov V, Yan J (2017) coga: convolution of gamma distributions. https://CRAN.R-project.org/package=coga. R package version 0.2.1
Iranmanesh A, Arashi M, Nagar D, Tabatabaey S (2013) On inverted matrix variate gamma distribution. Commun Stat Theory Methods 42(1):28–41
Jain GC, Consul PC (1971) A generalized negative binomial distribution. SIAM J Appl Math 21(4):501–513
Jasiulewicz H, Kordecki W (2003) Convolutions of Erlang and of Pascal distributions with applications to reliability. Demonstr Math 36(1):231–238
Kaas R, Goovaerts M, Dhaene J, Denuit M (2008) Modern actuarial risk theory: using R, vol 128. Springer, Berlin
Kadri T, Smaili K, Kadry S (2015) Markov modeling for reliability analysis using hypoexponential distribution. In: Kadry S, El Hami A (eds) Numerical methods for reliability and safety assessment: multiscale and multiphysics systems. Springer, Cham, pp 599–620
Khaledi BE, Kochar S (2013) A review on convolutions of gamma random variables. In: Li H, Li X (eds) Stochastic orders in reliability and risk. Springer, Berlin, pp 199–217
Mathai AM (1982) Storage capacity of a dam with gamma type inputs. Ann Inst Stat Math 34(1):591–597
Mathai A (2005) A pathway to matrix-variate gamma and normal densities. Linear Algebra Appl 396:317–328
Mathai AM, Saxena RK (1978) The H function with applications in statistics and other disciplines. Wiley, New Delhi. http://cds.cern.ch/record/101905
Moschopoulos PG (1985) The distribution of the sum of independent gamma random variables. Ann Inst Stat Math 37(1):541–544
Nadarajah S, Kotz S (2007) On the convolution of pareto and gamma distributions. Comput Netw 51(12):3650–3654
Perry D, Stadje W, Zacks S (1999) First-exit times for increasing compound processes. Commun Stat Stoch Models 15(5):977–992
Pozdnyakov V, Hu C, Meyer T, Yan J (2017) On discretely observed Brownian motion governed by a continuous time Markov chain. University of Connecticut, Department of Statistics, Tech. rep
R Core Team (2018) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. https://www.R-project.org/
Sen A, Balakrishnan N (1999) Convolution of geometrics and a reliability problem. Stat Probab Lett 43(4):421–426
Sim C (1992) Point processes with correlated gamma interarrival times. Stat Probab Lett 15(2):135–141
Srivastava HM, Karlsson PW (1985) Multiple Gaussian hypergeometric series. Ellis Horwood series in mathematics and its applications. E. Horwood, Billingham
Vellaisamy P, Upadhye N (2009) On the sums of compound negative binomial and gamma random variables. J Appl Probab 46(1):272–283
Zacks S (2004) Generalized integrated telegraph processes and the distribution of related stopping times. J Appl Probab 41(2):497–507
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Proofs
1.1 Proof of Proposition 1
First, note that
where \(E_1\) and \(E_2\) are independent of \(\{ U_k \}_{k \ge 1}\) with \(\mathcal {E}xp(\beta _1)\) and \(\mathcal {E}xp(\beta _2)\) distributions, respectively. Next, using mixture randomization we get
Similarly,
That is
Equation (6) of Lemma 1 completes the proof. \(\square \)
1.2 Proof of Lemma 1
Equation (4) gives us that
where the second equation follows from
Because of the identities
and
we obtain that
\(\square \)
1.3 Proof of Corollary 1
Note first that
where \(E_s\) are independent of \(\{ U_k \}_{k \ge 1}\) with an \(\mathcal {E}xp(\beta _s)\) distribution. Then we get that
\(\square \)
An alternative expression for confluent hypergeometric function
For completeness, we illustrate the derivation given by Di Salvo (2008) here. Equation (2) can be expressed as a multiple integral over \(n-1\)-dimensional standard simplex (Srivastava and Karlsson 1985), that is,
where \(E_{n-1}=\{(t_2,\dots ,t_n):t_2>0, \dots , t_n>0, t_2+\dots +t_n\le 1\}\). Di Salvo (2008) demonstrated that, with transformation
Equation (8) can be written as a multiple integral over the standard hypercube, that is,
where \(h_2 = 1\), \(h_i(\mathbf {u})=\prod _{j=2}^{i-1}(1-u_j)\), \(i=3,\dots , n\), and \(\bar{\alpha _i}=\sum _{j=i+1}^n \alpha _j + \alpha _1 - 1\), \(i=2, \dots , n-1\), \(\bar{\alpha _n}=\alpha _1-1\). Because of good analytical properties of integrand in Eq. (9), the multiple integral can be efficiently computed.
Rights and permissions
About this article
Cite this article
Hu, C., Pozdnyakov, V. & Yan, J. Density and distribution evaluation for convolution of independent gamma variables. Comput Stat 35, 327–342 (2020). https://doi.org/10.1007/s00180-019-00924-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00180-019-00924-9