A Cross-Target Methane Replication Test of the V38 Predictive Target-Response Layer for a Frozen ECSM Electron-Like Packet

This paper reports V39 of the ECSM electron-like packet measured-scattering programme. V38 introduced a low-parameter target-response layer that reduced held-out sevoflurane molecular scattering residuals without changing the frozen ECSM packet parameters. V39 performs a cross-target replication test using measured methane elastic electron-scattering differential-cross-section data.

The V39 suite has four linked parts. V39A locks the measured methane dataset from published Table 2 values over the measured angular domain 25 to 125 degrees, excluding extrapolated angular regions. V39B applies the previously trained V38 sevoflurane target-response parameters directly to methane as a cold-transfer test. V39C keeps the same V38 response form but refits the target-response parameters using methane training energies only: 50 eV, 150 eV, and 250 eV. The held-out methane energies are 100 eV, 200 eV, and 300 eV. V39D compares the sevoflurane and methane response parameters.

The methane dataset lock passes with 126 measured rows. The cold transfer from sevoflurane to methane also passes, giving a held-out mean relative residual of 0.208, below the predeclared 0.25 pass threshold. The same-form methane train/test replication gives a held-out mean relative residual of 0.174, compared with a bare ECSM held-out mean residual of 0.870. The final verdict is PASS_V39_CROSS_TARGET_SAME_FORM_REPLICATION.

This is not claimed as a full molecular QED calculation, recovered free-electron scattering, or proof of universal target-response parameters. Rather, it shows that the V38 response form transfers to a second measured molecular target while the frozen ECSM electron-like packet remains unchanged.