ERT -> ERD

[olafmersmann]

No description provided.

[nikohansen]

The ERT is already taken for "expected runtime" (meaning: average runtime as best estimator of its expectation = sum of runtimes divided by number of successes, I have seen it being read as "empirical runtime" too, which would not be wrong either), hence it's not an option to denote the runtime distribution.

We would have the choice between ERD (as used in the 2022 paper) or RTD or ERTD, both of which I kinda like better, also because they keep abbreviating runtime as RT. I can't remember why we ended up using ERD.

[olafmersmann]

We just discussed the issue some more and came up with runtime profile (RTP) or target runtime profiles (TRTP) with a slight bias towards RTP.

[nikohansen]

A disadvantage of "runtime distribution": it suggests data drawn uniformly and independently.