Assignment quality is count normalization before it is a prediction input. Exact arithmetic here means exact results for the stated model inputs; measured inputs still carry uncertainty and significant-figure limits.
highlighted = computed this step
A symmetric assignment scan starts with count pairs
Each row uses the same total shots for prepared zero and prepared one. The scan changes the correct count and error count together.
Ncorrect+Nerror=10
Three matrix shapes show diagonal strength
The table compresses the symmetric matrix to correct and error columns. The displayed matrix still contains both prepared rows.
Nc1086Ne024Ac15453Ae05152
The soft row is normalized but less decisive
The soft row has correct probability 3/5 and error probability 2/5. The row still sums to one.
53+52=1
Matrix quality is count normalization here
The scan does not claim a microscopic fidelity model. It only turns prepared-state counts into normalized readout rows.