There is an error in the normalization after photon-absorption when finite-temperature modes are used in composites (MCWF evolution). This might entail that the jump probability is incorrect as indicated also by incorrect time averages.
BinarySystem and Composite produce the same.
Testcase:
TwoModes_d --nTh0 .05 --nTh1 .025 --logLevel 2 --dc 1 --cutoff0 20 --cutoff1 30 --kappa0 20 --kappa1 40 --timeAverage
@ time=0.016 and 0.0211667 the error is manifest, since the photonnumber should never become non-integer in this case (on the next timestep, the renormalization corrects it).
The incorrect value of the photon number equals the inverse of the dimension of the other system, which explains why the error was not seen with a single mode.
The output and the jumps follow
resolved on [574aaa]
Related
Commit: [574aaa]