MASTER: R abort session from infinite Rejections of end conditions

[gtng92]

Stochastically, usually when nparticle > 3, printout from MASTER call becomes infinite:

Rejection end condition met at time 0.31206950547771256
Rejection end condition met at time 0.31378176369729716
Rejection end condition met at time 0.2785179708429142
Rejection end condition met at time 0.33230080829353104
Rejection end condition met at time 0.238955047263293
...

In the current XML template:

<lineageEndCondition spec='LineageEndCondition' population='@I' nLineages='0' isRejection='true'/>

Is this because the ancestors never coalesce?

[ArtPoon]

That statement halts the simulation when the infected subpopulation goes extinct. There is no coalescent process in these simulations -- they are run forward in time. My guess is that the parameter settings are such that the transmission rate is not sufficiently high to sustain the epidemic, at least to the point that you reach the required number of sampled infections.

[gtng92]

Ok thanks, that seems to work for higher numbers of nparticles now.

[ArtPoon]

With SMC, there is always a chance that a particle will end up in a silly region of parameter space, so we should have some hard limit to simulation time that we can catch as an exception and return a big distance to spike that particle out of contention.

[ArtPoon]

In the MASTER XML InheritanceTrajectory specification, use the argument maxConditionRejects to set a limit to the number of attempts for a particle to generate a tree

[gtng92]

Currently setting maxConditionRejects to 50 before exiting.