The current algo works by splitting the available sectors into contiguous blocks. It then steps through each block, creating a group starting with the first sector and, if necessary, adding subsequent sectors until the group has sufficient coverage (as evaluated on eclipse completeness metrics). It then moves forward, starting the next group from the next unused sector within the block and repeats the process.
This approach will tend to leave sectors at the end of a block which have insufficient coverage to create a group. Currently this is handled by extending the final group to include these additional sectors (see lines 127-131) which means the last group from a block tends to have better coverage than those preceding it.
An ideal outcome is that the sectors within the block are distributed within groups that are broadly similar coverage. This may involve extending an earlier group, rather than the last (while obviously still ensuring each group contains only contiguous sectors).
For an example of where this may be beneficial see TIC 0382067804. The first block covers sectors 3 to 7. The current algo splits this into two groups [3, 4] and [5, 6, 7]
Splitting the block as [3, 4, 5] and [6, 7] would ensure more even coverage.
The current algo works by splitting the available sectors into contiguous blocks. It then steps through each block, creating a group starting with the first sector and, if necessary, adding subsequent sectors until the group has sufficient coverage (as evaluated on eclipse completeness metrics). It then moves forward, starting the next group from the next unused sector within the block and repeats the process.
This approach will tend to leave sectors at the end of a block which have insufficient coverage to create a group. Currently this is handled by extending the final group to include these additional sectors (see lines 127-131) which means the last group from a block tends to have better coverage than those preceding it.
An ideal outcome is that the sectors within the block are distributed within groups that are broadly similar coverage. This may involve extending an earlier group, rather than the last (while obviously still ensuring each group contains only contiguous sectors).
For an example of where this may be beneficial see TIC 0382067804. The first block covers sectors 3 to 7. The current algo splits this into two groups [3, 4] and [5, 6, 7]
Splitting the block as [3, 4, 5] and [6, 7] would ensure more even coverage.