In gravitational collapse, it seems that there is no close or simple logical relationship between the formation of an event horizon (absolute horizon) and formation of a trapped surface (which implies an apparent horizon).
Modeling gravitational collapse is a specialized and highly technical field, and I don't know much about it. IIRC, simulations can most easily detect the formation of an apparent horizon, whereas the formation of an absolute horizon may not even be easy to pick out, since it's a global notion.
If we observe an object that has formed an event horizon, then I automatically imagine that it has also formed a trapped surface, and therefore must have a singularity due to the Penrose singularity theorem. But what basis do we have for this implication? Is it just a fact that arises from simulations, or is there some theorem that guarantees it?
Question: In realistic gravitational collapse, does the formation of an absolute horizon imply a trapped surface? If there are exceptions, are there strong reasons to believe these are unphysical or not generic?
The following, from Wald and Iyer, doi:10.1103/physrevd.44.r3719 , seems relevant:
...no general theorems require the presence of trapped surfaces in the collapse to a black hole. (The event horizon of the black hole must "settle down" to an outer marginally trapped surface at late times, but will normally have a positive expansion at any finite time. ) Nevertheless, the usual physical arguments concerning why black holes rather than naked singularities should be formed by collapse strongly suggests that outer trapped surfaces always should accompany black-hole formation.
From Hawking and Ellis, p. 321, it looks like "marginally" means the expansion scalar $\theta=0$. If I'm understanding the definition on p. 319 correctly, then "outer" means $\theta\ge0$. Their figure 59 on p. 321 shows an example of astrophysical collapse in which the apparent horizon forms later than the event horizon.
I'm not clear on how to interpret this and would appreciate further explanation. The gist of it seems to be that there are strong reasons to expect a trapped surface in realistic cases. The sentence in parens seems to be saying that due to no-hair theorems, the exterior spacetime has to approach a Kerr-Newman spacetime. I guess this would lead us to expect that the interior would also approach Kerr-Newman spacetime, but maybe this is not an absolute implication. Would the exception be unphysical, e.g., it has to be non-generic?
The final sentence from the Wald paper seems to be making a link with cosmic censorship, but that seems vague to me. Weak cosmic censorship just says there's an absolute horizon, but doesn't say there's a trapped surface.