Issues

• GMM dim-group: test_empty_dim_group_key_is_dropped_and_harmless fails on GPU (A100), passes CPU

  oshaughnessy-junior/research-projects-RIT#424 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• Re-measure --sampler-warmstart-cover-frac 0.1 under-coverage claim after #412

  oshaughnessy-junior/research-projects-RIT#423 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• AV contraction loses parts of the live region; lnZ low after #412 on a JAX ILE fixture

  oshaughnessy-junior/research-projects-RIT#422 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• Seeded ILE adaptive_cartesian_gpu is not reproducible on GPU

  oshaughnessy-junior/research-projects-RIT#421 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• GMM: refit collapses a correct seed toward the centre; cold start never resolves separated per-group modes

  oshaughnessy-junior/research-projects-RIT#418 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• shape gate: pytest entry runs on GPU on pcdev hosts (cupy loaded before conftest hides devices)

  oshaughnessy-junior/research-projects-RIT#417 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• statutils: log aggregate is NaN after an all -inf chunk, and stays NaN

  oshaughnessy-junior/research-projects-RIT#416 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• CIP AV/GMM exits 1 under cupy: float128 lnL sent to the device (mcsamplerAdaptiveVolume:2025)

  oshaughnessy-junior/research-projects-RIT#414 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• Import-bound xpy=xpy_default defaults remain in AV, portfolio, NFlow, generic samplers

  oshaughnessy-junior/research-projects-RIT#413 · Filter by author oshaughnessy-junior opened on Oct 7, 2026

• util_ManualOverlapGrid.py ignores --srate on the lalsuite path (always 16384 Hz)

  bug
  oshaughnessy-junior/research-projects-RIT#376 · Filter by author oshaughnessy-junior opened on Sep 30, 2026

• jax_ile: --angle-marg-scheme multipeak (#313) is not traceable under the sampler jit; Table 3 arms cannot run it

  oshaughnessy-junior/research-projects-RIT#316 · Filter by author oshaughnessy-junior opened on Sep 9, 2026
  1

• four-axis policy: local branch accepts +16 nat at rho 163 under the #301 capacities (max_time_nodes 256, max_starts 128, budget 1e-2)

  oshaughnessy-junior/research-projects-RIT#308 · Filter by author oshaughnessy-junior opened on Sep 9, 2026

• jax gate: test_jax_slowrot segfaults on CIT when test_flow_reuse_default.py is collected first (pre-existing on rift_O4d)

  oshaughnessy-junior/research-projects-RIT#300 · Filter by author oshaughnessy-junior opened on Sep 9, 2026

• jax gate: CI runs an unpinned jax that no local environment has

  oshaughnessy-junior/research-projects-RIT#292 · Filter by author oshaughnessy-junior opened on Sep 8, 2026
  5

• anglemarg buffer cap does not cover the flowMC sampling path: 833 GiB request, laplace unusable under flowmc-phipsimarg

  oshaughnessy-junior/research-projects-RIT#253 · Filter by author oshaughnessy-junior opened on Sep 5, 2026

• _runtime_amp_failsafe's SUSPECT label is uninformative under vmap (i.e. on every production path)

  oshaughnessy-junior/research-projects-RIT#240 · Filter by author oshaughnessy-junior opened on Sep 3, 2026

• Vectorized non-GPU time marginalization offsets by ONE detector's max while exponentiating the accumulated network sum (platform-conditional; float64 platforms get #232's 745-nat budget)

  oshaughnessy-junior/research-projects-RIT#236 · Filter by author oshaughnessy-junior opened on Sep 3, 2026

• batchmode defaults --interpolate-time to 'nearest', which its own help text says is never competitive — and the jax driver defaults to 'sinc'

  oshaughnessy-junior/research-projects-RIT#233 · Filter by author oshaughnessy-junior opened on Sep 3, 2026
  1

• Vectorized time-marginalized likelihood offsets by the BATCH max, not the per-sample max: lnL underflows to -inf above rho ~ 40 and mcsamplerAV collapses (device-independent; affects a delivered O4b result)

  oshaughnessy-junior/research-projects-RIT#232 · Filter by author oshaughnessy-junior opened on Sep 3, 2026
  2

• AV + --distance-marginalization without --force-xpy: TypeError on every point, and the driver exits 0 with no output files

  oshaughnessy-junior/research-projects-RIT#231 · Filter by author oshaughnessy-junior opened on Sep 3, 2026

• mcsamplerGPU reports a mismatched CUDA stack as "no cupy" and silently runs numpy under --gpu --force-xpy (162 of 171 runs in one campaign)

  oshaughnessy-junior/research-projects-RIT#229 · Filter by author oshaughnessy-junior opened on Sep 2, 2026

• test/jax: seven slowrot_fs_lib callers silently inherit a 4x finer rholm sampling

  oshaughnessy-junior/research-projects-RIT#197 · Filter by author oshaughnessy-junior opened on Aug 26, 2026

• slowrot: rotation_post_phase is only exercised where it is numerically the identity

  oshaughnessy-junior/research-projects-RIT#173 · Filter by author oshaughnessy-junior opened on Aug 19, 2026
  1

• slowrot Path D: antenna_response_fd / F_fd_expanded carry the same unpaired-Nyquist non-Hermiticity as #164

  oshaughnessy-junior/research-projects-RIT#168 · Filter by author oshaughnessy-junior opened on Aug 19, 2026

• Vision: define an atomic campaign-assimilation boundary

  oshaughnessy-junior/research-projects-RIT#141 · Filter by author oshaughnessy-junior opened on Aug 18, 2026