=

The passage around this formula

The central mathematical test is simple to state. Apply a common frame transformation h to the physical description. Covariance sends g and $\hat g$ to hg and h$\hat g$ , or to the appropriate left/right action fixed by convention. A left-invariant distance satisfies $d(h\hat g,hg)=d(\hat g,g)$. The Haar measure is invariant, and the transformed POVM produces the correspondingly transformed outcome probabilities. Therefore the integration region defining success is merely relabeled. The value of $p_\epsilon(F)$ , and hence $R_{\epsilon,\delta}$ , should not change.

Read this part in the article →