=

The passage around this formula

Now define $R_{\epsilon,\delta}(g)= \max\left\{M:\begin{array}{l} F_1,\ldots,F_M\ \text{are disjoint},\\ p_{\epsilon}(F_k)\geq1-\delta\ \text{for every }k \end{array}\right\}$. subject to 1-$\delta$>$p_{\mathrm{blind}}(\epsilon)$ and to a fixed rule for which fragment partitions are admissible. The displayed g records the physical parameter of interest; under the Haar-average convention the resulting scalar is independent of its coordinate label. A worst-case definition would retain explicit dependence until the infimum is taken.

Read this part in the article →