I have two independent exponential RV's $X$ and $Y$ with parameters $\lambda_1$ and $\lambda_2$ respectively. I want to find the density of $X + Y$. I've tried using the convolution formula and got the following:

\begin{align*} f_{X+Y}(t) &= \int_{-\infty}^{\infty}f_{X,Y}(x,t-x)\, dx \\ &= \int_{0}^{t}\lambda_1\lambda_2e^{-\lambda_1x-\lambda_2(t-x)}\, dx \\ &= \left[\frac{\lambda_1\lambda_2}{\lambda_2 - \lambda_1}e^{x(\lambda_2 - \lambda_1)- \lambda_2t}\right]^t_0\\ &= \frac{\lambda_1\lambda_2}{\lambda_2 - \lambda_1}e^{-\lambda_1t} + \frac{\lambda_1\lambda_2}{\lambda_1 - \lambda_2}e^{-\lambda_2t} \end{align*}

Is this correct? If so, that begs the question of what happens when $\lambda_1 = \lambda_2$ ? That is, since the denominators are $0$.


Yes, when $\lambda_1\neq \lambda_2$ then $f_T(t)=\dfrac{\lambda_1\lambda_2(e^{-\lambda_2t}-e^{-\lambda_1t})}{\lambda_1-\lambda_2}\mathbf 1_{t\in[0;\infty)}$

As to what happens when $\lambda_1=\lambda_2$ try making the substitution before performing the integration.

$$f_T(t)=\int_0^t \lambda^2 e^{-\lambda x-\lambda(t-x)}\operatorname d x=\ldots$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.