Truncated multivariate normal likelihood

Right, but here it’s one extra variable per observed censored data-point