Scalar Perturbations Through a Torsion Bounce: Which Wavelengths Emerge Unchanged and Which Are Processed

At one tenth of the bounce wavenumber the amplification factor is 0.768: long-wavelength modes cross with an order-unity rescaling, neither preserved exactly nor destroyed. At the bounce wavenumber the factor is 0.705. At ten times the bounce wavenumber it is 1.014, again near unity, while the largest value in the scan, 2.321, occurs at an intermediate wavenumber where the mode period and the bounce duration are commensurate and the interaction is resonant in character. The scan therefore shows bounded processing everywhere: no wavenumber diverges, no wavenumber is erased, and the total spread of outcomes across two decades of wavenumber is a factor of a few. That boundedness is itself the headline, because variables less well adapted to the bounce have historically suggested far more violent behaviour at the same crossing. The study is a computation from stated assumptions, performed by the accompanying script; it is simulated, not measured.