 Variance of a corrected heterodyne overlap estimate

The variance of a corrected heterodyne overlap estimate quantifies uncertainty in $\hat{\gamma }=B/(2\sqrt{I_1{I}_2})$, where $B$ is corrected peak beat current and $I_1,I_2$ are separately measured arm currents. The consistency bound for separately measured heterodyne arm powers requires the physical overlap not to exceed one, but measured estimates fluctuate.

For small errors, let $\mathbf{e}=(\delta B/B,\delta I_1/I_1,\delta I_2/I_2{)}^T$ have covariance $R$, and put $\mathbf{v}=(1,-1/2,-1/2{)}^T$. Linearization givesThus common influences can cancel rather than contribute three independent uncertainty terms. This is the covariance form of the primary measurement-propagation rule.^[ref]

For an original model, all three current readings share a $2\%$ multiplicative calibration error, with additional independent $0.5\%$ errors on each. The common contribution to ${\mathbf{v}}^T\mathbf{e}$ is zero. The relative standard uncertainty is only $\sqrt{3/2}\times 0.5\%\approx 0.6124\%$, or $0.00490$ at overlap $0.80$. Treating the common calibration error as independent in all three readings would instead give about $2.525\%$ relative uncertainty.

The common-gain cancellation in heterodyne overlap uncertainty is exact for a single positive gain $g$: replacing $(B,I_1,I_2)$ by $(gB,g{I}_1,g{I}_2)$ leaves the ratio unchanged. A separate radio-frequency gain uncertainty affecting only $B$ does not cancel, nor does an additive dark-current offset generally have this property.

NIST's heterodyne receiver procedure corrected electrical loss and impedance effects separately from matching optical powers and modes.^[ref] Build the covariance from those actual shared and separate calibration paths. This first-order variance model also needs reconsideration when an arm current is close to zero or background subtraction makes its uncertainty comparable to its value.