The LR2 Lifetime Distribution: Rate Crossing, Reliability Properties, and Estimation

Abstract

We study the LR2 one-parameter lifetime distribution obtained by mixing an exponential component of rate $θ/2$ and a gamma component with shape four and rate $θ^2$. The exponential mixing probability is $θ^2/(1+θ^2)$ for $θ>0$. We derive the density, survival function, moments, transforms, truncated moments, quantiles, order-statistic distributions, mean residual life, and stress--strength reliability. Identifiability follows from the density at the origin, and strict monotonicity of the mean yields a unique moment estimator. A complete density-turning-point criterion and analytic initial and tail hazard expansions clarify the possible shapes. The component rates cross at one half. The hazard decreases initially for every positive parameter and eventually increases in the gamma-dominated regime; it eventually decreases in the exponential-dominated regime. We develop likelihood estimation, observed information, constrained expectation--maximization, spacing estimation, and inference under regularity conditions, and state the likelihood for independent right censoring. A reproducible Monte Carlo study with eight settings and 4,000 generated samples compares likelihood and moment estimation and evaluates log-scale confidence intervals. The analysis distinguishes mathematical properties from empirical validation and explains the restrictions associated with a fixed working time unit.