I think the low mumu efficiency for d4 and ttRR is probably coming from what we were talking about in a different email, your requirement that a true jet must not match an electron or muon. This requirement should not be imposed for muons. The reason it is different for muons and electrons is that muons are not included in ATLAS default truth jets, while electrons are included.

This has reminded me that we had to alter the default ATLAS true jet definitions in our last paper. I apologize that I didn't comment on this earlier (I see now it was mentioned in previous emails but I didn't pick up on it). I've put the full explanation and details below; if we want to use true jets in our true fiducial region definition, I think we will need to discuss this issue further (and it is actually more important for electrons than for muons). But I am wondering if we can avoid this discussion for the results we want to send out soon (and maybe longer-term) by removing the dR(jet, lepton) requirements altogether (for both electrons and muons). In other words, use only the true ptcone and etcone requirements. How is the model dependence if you do that?

Here are more details about the truth jet stuff:

The collection of true particles used to build default truth jets in ATLAS includes all stable true hadrons and electrons. Muons and neutrinos are not included in the collection used to make true jets. So (as you guys know) every electron from DCH decay results in a true jet; then it doesn't really make sense to require that this electron is separated from true jets.

A better theoretically motivated definition would modify that collection as follows:
- Include muons and neutrinos that are part of the hadronic shower in the true jet.
- Remove stable electrons, muons, and neutrinos from prompt decays. In the Standard Model, this is only particles coming from W or Z decays. In exotics models, where there can be leptonic decays from other true particles, one would also need to remove these leptons from the collection.

Many Standard Model analyses have already had to deal with this issue. As a result, there is a collection in the SMWZ D3PD called jet_AntiKt4TruthWZ*, which modifies the input particle collection as described above. These truth jets include non-interacting particles (muons and neutrinos) and remove leptons from W/Z decays. They do not take into account possible exotics models that result in prompt leptons from other decays. We could use these jets in the d4 and ttRR models, where the prompt leptons come from W/Z decays, but not in DCH or LRSM models.

In our previous analysis, we built our own collection of truth jets, which were the same as the jet_AntiKt4TruthWZ collection, except that prompt leptons from exotic decays were also excluded. We did this by modifying the code that builds the jet_AntiKt4TruthWZ* code.

So probably it is not possible to figure out a fully correct approach in the next day; maybe we could all discuss this on Skype sometime soon? But if the model independence using just the true etcone and ptcone is already good, then we might not have to worry about this.