Issues

• A table function indexed by a function freezes the trajectory under codegen (regression from #803 step 3; codegen=True and the default above 256 species)

  bugcorrectness concernfamily: table functions
  lanl/bngsim#965 · Filter by author wshlavacek opened on Oct 1, 2026

• A plain run pinned on a state switch whose residual is not 0 on its own surface returns with the switch not taken (Aobs^2 > thr^2: Y = 0 for 0.09)

  bugcorrectness concernfamily: ODE events
  lanl/bngsim#963 · Filter by author wshlavacek opened on Oct 1, 2026

• Tracking: forward-sensitivity runs that are refused where a derivative exists (refusals to lift)

  enhancementfamily: sensitivity termstracking
  lanl/bngsim#961 · Filter by author wshlavacek opened on Oct 1, 2026

• Process: fix review findings in the PR, and refuse forward sensitivities outside a supported envelope

  tracking
  lanl/bngsim#960 · Filter by author wshlavacek opened on Oct 1, 2026

• A plain run of an SBML model stalls at a piecewise condition its species approaches slowly (timeout; the .net form is #928)

  bugfamily: ODE events
  lanl/bngsim#959 · Filter by author wshlavacek opened on Oct 1, 2026

• A model with any event keeps every sensitivity column plain: a window that closes as a power is 0.4% off again (comoving frames are off when n_event_roots > 0)

  bugcorrectness concernfamily: sensitivity terms
  lanl/bngsim#958 · Filter by author wshlavacek opened on Oct 1, 2026
  1

• An event at (time >= tau) with tau 1 to 256 ulp after an output sample never fires (X(6) = 2 for 5, plain run)

  bugcorrectness concernfamily: ODE events
  lanl/bngsim#950 · Filter by author wshlavacek opened on Oct 1, 2026

• A crossing on the closing edge of a window that closes as a power puts the window-length sensitivity 40% off (the isolation bump reads the power a hair from its zero)

  bugcorrectness concernfamily: sensitivity terms
  lanl/bngsim#949 · Filter by author wshlavacek opened on Oct 1, 2026
  1

• A sensitivity to a counter clock's initial value or rate is 0.2-0.4% off under a window that closes as a power, and the run fails under one that opens as a power

  bugcorrectness concernfamily: sensitivity terms
  lanl/bngsim#948 · Filter by author wshlavacek opened on Oct 1, 2026
  1

• Events on time >= tau and time > tau fire as one batch or in sequence depending on the rounding of tau (X ends at 0.50 or 0.15)

  bugcorrectness concernfamily: ODE events
  lanl/bngsim#947 · Filter by author wshlavacek opened on Oct 1, 2026

• A state-dependent switch and a time switch on one instant: the time switch's jump is counted twice (dX/dthr = -1 for -0.5)

  bugcorrectness concernfamily: sensitivity terms
  lanl/bngsim#946 · Filter by author wshlavacek opened on Oct 1, 2026
  1

• A state-dependent switch that crosses within tolerance of an event's time is not refused: the sensitivity comes from one side of a kink

  bugcorrectness concernfamily: sensitivity terms
  lanl/bngsim#945 · Filter by author wshlavacek opened on Oct 1, 2026
  1

• A counter thresholded once directly and once through max() gets a wrong sensitivity to its rate constant (dY/dk = 14.5 for 10.2)

  bugcorrectness concernfamily: sensitivity terms
  lanl/bngsim#938 · Filter by author wshlavacek opened on Oct 1, 2026
  1

• An event whose trigger leaves its start threshold at second order never fires (at (B > 0) with B = t²/2)

  bugcorrectness concern
  lanl/bngsim#934 · Filter by author wshlavacek opened on Oct 1, 2026

• A rate-law condition that reads rateOf stops a forward-sensitivity run with CVODE's no-progress error just before the crossing; the plain run is right

  bugfamily: sensitivity terms
  lanl/bngsim#932 · Filter by author wshlavacek opened on Oct 1, 2026
  2

• The compiled right-hand side reads a Michaelis-Menten kcat from its parameter slot when a function owns that name, so a codegen or forward-sensitivity run integrates the wrong trajectory

  bugcorrectness concernfamily: codegen .net reader
  lanl/bngsim#931 · Filter by author wshlavacek opened on Oct 1, 2026

• differentiable_solve repeats structure-only work every call since routing through Simulator (up to 6x on a 300-species primal)

  enhancement
  lanl/bngsim#919 · Filter by author wshlavacek opened on Sep 30, 2026

• A .net rate law that slides along its state switch runs up to 12x slower since #903 (60 s timeout where 6.7 s sufficed)

  bug
  lanl/bngsim#908 · Filter by author wshlavacek opened on Sep 30, 2026

• A state window a restart would catch is still stepped over when #903's restart test calls the crossing unresolved (16-point sampling, 10·√N band)

  bugcorrectness concern
  lanl/bngsim#907 · Filter by author wshlavacek opened on Sep 30, 2026

• A rooted state window whose closing threshold moves slowly stalls for millions of steps and holds the gate open (SBML today; .net after #903)

  bugcorrectness concern
  lanl/bngsim#905 · Filter by author wshlavacek opened on Sep 30, 2026

• A narrow if() window is still stepped over when written as a number, as one non-monotone comparison, or reached through steady_state() (after #897)

  bugcorrectness concern
  lanl/bngsim#904 · Filter by author wshlavacek opened on Sep 30, 2026

• Nightly wall-time check on bng_nf and bng_ssa times model load, not the simulation: 0.012 s summed against 4004 s run

  enhancement
  lanl/bngsim#901 · Filter by author wshlavacek opened on Sep 30, 2026

• test_run_tests_sh's walk-up probe fails in any checkout without build/, because it forbids the skip test_version_consistency takes there

  bug
  lanl/bngsim#900 · Filter by author wshlavacek opened on Sep 30, 2026
  2

• A step or comparison that reads a mod() of constants gets no crossing stops, so its pulses are stepped over

  bugcorrectness concern
  lanl/bngsim#899 · Filter by author wshlavacek opened on Sep 30, 2026

• Network-free: a time-indexed table function is not refreshed while nothing fires, so a rate that turns on later never does

  bugcorrectness concern
  lanl/bngsim#892 · Filter by author wshlavacek opened on Sep 29, 2026