Wrong cerenkov photon directions in eic-opticks

[ggalgoczi]

Yousen Zhang reported that when trying to simulate Cerenkov photons he observed that the Cerenkov photons have wrong Cerenkov cone. He simulated a sphere of water with r index of 1.1 and shot a 3 GeV muon into it. Also there are some photons that are going the wrong way? The latter one is speculation from my side. See screenshot created by dumping the photon histories into numpy files and plotting them:

[ggalgoczi]

Yousen reran with an older container from Feb. He got better cone angle numbers:

[ggalgoczi]

Simon split the WITH_OLD_FRAME refactor in simoncblyth/opticks into two commits _ c9cfeaab6 (2026-02-06, the refactor itself, hiding sframe behind WITH_OLD_FRAME) and d292d8bf0 (2026-03-18, the companion fix that restored the genstep matline lookup on the new code path, with Simon's own note that the effect was "verbose qcerenkov kernel logging and flat cerenkov wavelength"). eic-opticks pulled the first via 71ebf76 (Sync with upstream #204, 2026-03-03) but never picked up the second, so every GPU Cerenkov genstes silently fell back to matline=0 from 2026-03-03 until f6ee3f7 today.

Proposing fix in #309

[plexoos]

Could you add a regression test to verify the fix in #309?

[ggalgoczi]

Could you add a regression test to verify the fix in #309?

Not at the moment. Can get back to it at a later point.