Minimal sufficient and complete statistic for the generalized inverse Gaussin disributoin.

Σ weights[j] * sample[j] / Σ weights[j]

Σ weights[j] / sample[j] / Σ weights[j]

Σ weights[j] * log(sample[j]) / Σ weights[j]

See Implementation

Minimal sufficient and complete statistic for the generalized inverse Gaussin disributoin.