First, I only have a vague idea about what I mean by "discovery rate."
I'm defining discovery rate as the rate at which new parts of the program under test are discovered - and leaving out any further details.
The intuition is as follows:
If certain branches have a tendency to discover new parts of the program, they have a "rate" at which they do this. If certain branches contain higher rates of discovery, then should they be preferred for future exploration (given other heuristics are approx. equal)?
Said another way, "Where there's smoke, there's fire. Where there's more smoke, there's more fire." Our chances of finding new parts of the program may be higher for parts that tend to have higher incidence of discovery.
First, I only have a vague idea about what I mean by "discovery rate."
I'm defining discovery rate as the rate at which new parts of the program under test are discovered - and leaving out any further details.
The intuition is as follows:
If certain branches have a tendency to discover new parts of the program, they have a "rate" at which they do this. If certain branches contain higher rates of discovery, then should they be preferred for future exploration (given other heuristics are approx. equal)?
Said another way, "Where there's smoke, there's fire. Where there's more smoke, there's more fire." Our chances of finding new parts of the program may be higher for parts that tend to have higher incidence of discovery.