how do I introduce a custom three parameter life distribution into JMP Analyze-->Reliability and Survival--> Life Distribution

[#1]

want to be able to perform standard set of life distribution analyses on a custom three parameter life distribution ... how do I configure platform to have an additional life distribution to show up along with standard list of Weibull, Exponential, Lognormal, DS Weibull, etc

the CDF for this life distribution is given by

F(t|p,a,y) = 1 - exp(p*exp(-t/a)-yt-p)

or expressed equivalently

F(t|a,y,x) = 1 - exp(ax*exp(-t/a)-yt-ax)

and corresponds to a hazard function starting at failure rate (y+x) which decays exponentially with time constant a to constant failure rate y .... here, the DS proportion p=ax in the limit that y is zero

have found that for some datasets, the BIC figure of merit for this life distribution is lower than that of other early/mid bathtub curve options

would like to have life distribution platform compute the standard set of estimates, covariance matrix etc for this three-parameter life distribution

[#2]

See custom life distribution reliability model 

The Fan.jmp sample data set under "Reliablity/Survival" folder is an example of how to set up the loss function. The three columns "Exponential", "Weibull", and "Extreme Value" define loss functions for three different distributions. To use them, launch Nonlinear, and put one of those three columns to "Loss", then click "Go" in the launch dialog. Then click "Go" in the Control Panel within the report.

What you need to do is to craft the loss function for your distribution. Notice the Fan.jmp only have failures and right censored observations. If you have other censoring types, e.g. left censored, you need to make additional changes accordingly.

[#3]

doesn't answer my question ... how can I augment the set of life distributions in the Life Distribution platform ... I want to be able to analyze datasets apples-to-apples using this custom life distribution and pre-existing ones ... all with standard Life Distribution platform figures of merit, estimates, co-variance matrices etc