The $D^{*+}$ veto is used to minimize the overlap of $B\rightarrow D^{*+}(\tau /\mu)\nu$ events between the $D^0\mu$ and $D^{*+}\mu$ samples. It's implemented using this TupleTool in our build of DaVinci, which returns the absolute mass differences of long iso tracks refit with the $D^0$ to form a $D^{*+}$ and the $D^0$ mass (if the difference is $140 < m_{\Delta} < 160$), which then specifies if the event should be vetoed if $\text{min}_i|m_{\Delta ,i}-m_{\Delta ,PDG}|>4$ (where $m_{\Delta ,PDG}=m_{D^{*+}}-m_{D^0}$ from the PDG).
Yipeng did some validation of the $D^{*+}$ veto, but I'm confused about what exactly he was calculating there.
To get the veto eff, $\epsilon_{D^{*+}\text{veto}}$, I regenerated the $D^0$ sample $B^0\rightarrow D^{*+}(\mu /\tau)\nu$ MC without the $D^{*+}$ veto applied, and calculated using the step-2 ntuples (without L0 or PID weights applied, or analysis cuts that happen after step-2--though, including these produces a similar $\epsilon_{D^{*+}\text{veto}}$)
$\epsilon_{D^{*+}\text{veto}}=\frac{N(B^0\rightarrow D^{*+}(\mu /\tau)\nu \text{ reco as } D^0\mu \text{ with }D^{*+} \text{ veto})}{N(B^0\rightarrow D^{*+}(\mu /\tau)\nu \text{ reco as } D^0\mu \text{ without }D^{*+} \text{ veto})}$
and found for $\mu$ and $\tau$, $\epsilon_{D^{*+}\text{veto}}=25.8\%$ and $28.1\%$, respectively. Phoebe checked and told me hers was around $26\%$ too.
For removing the overlap of $B\rightarrow D^{*+}(\tau /\mu)\nu$ events between the $D^0\mu$ and $D^{*+}\mu$ samples, I wrote a short script to compute this using the step-2 ntuples already in rdx-run2-analysis and found the overlap (events shared between $D^{0}/D^{*+}$ samples over number of $B\rightarrow D^{*+}(\tau /\mu)\nu$ events in $D^{*+}$ sample) to be $0.67\%$ and $0.65\%$, respectively, also in good agreement with Phoebe's $\approx 0.5\%$ overlap.
The $D^{*+}$ veto is used to minimize the overlap of $B\rightarrow D^{*+}(\tau /\mu)\nu$ events between the $D^0\mu$ and $D^{*+}\mu$ samples. It's implemented using this TupleTool in our build of DaVinci, which returns the absolute mass differences of long iso tracks refit with the $D^0$ to form a $D^{*+}$ and the $D^0$ mass (if the difference is $140 < m_{\Delta} < 160$), which then specifies if the event should be vetoed if $\text{min}_i|m_{\Delta ,i}-m_{\Delta ,PDG}|>4$ (where $m_{\Delta ,PDG}=m_{D^{*+}}-m_{D^0}$ from the PDG).
Yipeng did some validation of the $D^{*+}$ veto, but I'm confused about what exactly he was calculating there.
To get the veto eff, $\epsilon_{D^{*+}\text{veto}}$, I regenerated the $D^0$ sample $B^0\rightarrow D^{*+}(\mu /\tau)\nu$ MC without the $D^{*+}$ veto applied, and calculated using the step-2 ntuples (without L0 or PID weights applied, or analysis cuts that happen after step-2--though, including these produces a similar $\epsilon_{D^{*+}\text{veto}}$)
$\epsilon_{D^{*+}\text{veto}}=\frac{N(B^0\rightarrow D^{*+}(\mu /\tau)\nu \text{ reco as } D^0\mu \text{ with }D^{*+} \text{ veto})}{N(B^0\rightarrow D^{*+}(\mu /\tau)\nu \text{ reco as } D^0\mu \text{ without }D^{*+} \text{ veto})}$
and found for $\mu$ and $\tau$, $\epsilon_{D^{*+}\text{veto}}=25.8\%$ and $28.1\%$, respectively. Phoebe checked and told me hers was around $26\%$ too.
For removing the overlap of $B\rightarrow D^{*+}(\tau /\mu)\nu$ events between the $D^0\mu$ and $D^{*+}\mu$ samples, I wrote a short script to compute this using the step-2 ntuples already in
rdx-run2-analysisand found the overlap (events shared between $D^{0}/D^{*+}$ samples over number of $B\rightarrow D^{*+}(\tau /\mu)\nu$ events in $D^{*+}$ sample) to be $0.67\%$ and $0.65\%$, respectively, also in good agreement with Phoebe's $\approx 0.5\%$ overlap.