How does error propagate during the operations of floating$-$point multiplication and division?

Relative errors multiply $($i$.$e$.,$ the relative error is roughly bounded by the product of the relative errors

of the inputs$).$

Absolute errors multiply $($i$.$e$.,$ the absolute error is roughly bounded by the product of the absolute

errors of the inputs$).$

Absolute errors add $($i$.$e$.,$ the absolute error is roughly bounded by the sum of the absolute errors of

the inputs$).$

Relative errors add $($i$.$e$.,$ the relative error is roughly bounded by the sum of the relative errors of the

inputs$).$