Greetings,

I've checked energy conservation in the G4MuonDecayChannelWithSpin generator using sum of e+, nu_mu and anti_nue energies. In terms of generator I used the sum of the following variables:
total_energy = energy + daughterparticle1->Get4Momentum().e() + daughterparticle2->Get4Momentum().e();
Then I compared my total_energy variable with EMMU (which is the muon mass) by introducing the difference (total_energy - EMMU)/EMMU. So, the difference is very low ~10^{-5} - 10^{-4}. I checked 193 muon decays with this channel and found 4 events with the high difference: -0.011572, -0.015309, -0.019909, -0.007233. It means that energy is not always conserved in the muon decays (2% of considering events!).
I did the same exercise for the G4MuonRadiativeDecayChannelWithSpin. My total energy in terms of generator now is:
total_energy = E + G + daughterparticle2->Get4Momentum().e() + daughterparticle3->Get4Momentum().e();
And again I compared total energy with muon mass by introducing the same difference:
(total_energy - EMMU)/EMMU.
I checked 193 events, the difference is higher than for the non-radiative case ~(1-3)x10^{-3}, but I found two events with very high difference -0.495478, -0.496089, this is large energy deficit! So ~1% of checked events are not consistent with the energy conservation law.
Is this behaviour expected or the energy conservation check have to be added to the generator?

Regards,
Artur