Understand the failure.
Verify the repair.
Small, reproducible software failures. The broken implementation, the fix that didn’t work, and the one that passed—preserved together.
Explore the cases ↓How results are verified ↗WHAT THE ARCHIVE CONTAINS
100840 executable cases. 20168 are open.
Every case records the implementation that fails, the fix that did not work, and the repair that passed its checks—with recorded outputs and source hashes. This release adds 100840 cases across 20168 failure mechanisms and 254 domains.
The open tier gives you the failure and the unsuccessful fix for one case in every mechanism. The remaining 80672 cases, 5 variants per mechanism, are member-only: the verified repair, its recorded checks, and the full fixture suite are held in the member archive. Read the methodology ↗
A RECORD OF WHAT WENT WRONG
Browse the archive / 100840
Pendulum kinetic energy uses one power of the rod length · case 01
Energy diagnostics disagree with the integrated state for any rod length other than one.
Pendulum kinetic energy uses one power of the rod length · case 02
Energy diagnostics disagree with the integrated state for any rod length other than one.
Pendulum kinetic energy uses one power of the rod length · case 03
Energy diagnostics disagree with the integrated state for any rod length other than one.
Pendulum kinetic energy uses one power of the rod length · case 04
Energy diagnostics disagree with the integrated state for any rod length other than one.
Pendulum kinetic energy uses one power of the rod length · case 05
Energy diagnostics disagree with the integrated state for any rod length other than one.
Stability discriminant uses raw stiffness instead of k over m · case 01
Heavy bodies are classified with the eigenvalues of a unit-mass system.
Stability discriminant uses raw stiffness instead of k over m · case 02
Heavy bodies are classified with the eigenvalues of a unit-mass system.
Stability discriminant uses raw stiffness instead of k over m · case 03
Heavy bodies are classified with the eigenvalues of a unit-mass system.
Stability discriminant uses raw stiffness instead of k over m · case 04
Heavy bodies are classified with the eigenvalues of a unit-mass system.
Stability discriminant uses raw stiffness instead of k over m · case 05
Heavy bodies are classified with the eigenvalues of a unit-mass system.
Stability test checks only the first characteristic root · case 01
Overdamped systems whose fast root leaves the unit disc are reported stable.
Stability test checks only the first characteristic root · case 02
Overdamped systems whose fast root leaves the unit disc are reported stable.
Stability test checks only the first characteristic root · case 03
Overdamped systems whose fast root leaves the unit disc are reported stable.
Stability test checks only the first characteristic root · case 04
Overdamped systems whose fast root leaves the unit disc are reported stable.
Stability test checks only the first characteristic root · case 05
Overdamped systems whose fast root leaves the unit disc are reported stable.
Underdamped eigenvalues use twice the imaginary part · case 01
Oscillatory systems report too much amplification and are flagged unstable early.
Underdamped eigenvalues use twice the imaginary part · case 02
Oscillatory systems report too much amplification and are flagged unstable early.
Underdamped eigenvalues use twice the imaginary part · case 03
Oscillatory systems report too much amplification and are flagged unstable early.
Underdamped eigenvalues use twice the imaginary part · case 04
Oscillatory systems report too much amplification and are flagged unstable early.
Underdamped eigenvalues use twice the imaginary part · case 05
Oscillatory systems report too much amplification and are flagged unstable early.
Complex eigenvalues are tested with the real-axis step criterion · case 01
Undamped oscillators are called stable although explicit Euler always gains energy on them.
Complex eigenvalues are tested with the real-axis step criterion · case 02
Undamped oscillators are called stable although explicit Euler always gains energy on them.
Complex eigenvalues are tested with the real-axis step criterion · case 03
Undamped oscillators are called stable although explicit Euler always gains energy on them.
Complex eigenvalues are tested with the real-axis step criterion · case 04
Undamped oscillators are called stable although explicit Euler always gains energy on them.
Complex eigenvalues are tested with the real-axis step criterion · case 05
Undamped oscillators are called stable although explicit Euler always gains energy on them.
Unit amplification is labeled stable before the marginal test · case 01
A free damped particle (zero stiffness) is reported stable although its drift mode never decays.
Unit amplification is labeled stable before the marginal test · case 02
A free damped particle (zero stiffness) is reported stable although its drift mode never decays.
Unit amplification is labeled stable before the marginal test · case 03
A free damped particle (zero stiffness) is reported stable although its drift mode never decays.
Unit amplification is labeled stable before the marginal test · case 04
A free damped particle (zero stiffness) is reported stable although its drift mode never decays.
Unit amplification is labeled stable before the marginal test · case 05
A free damped particle (zero stiffness) is reported stable although its drift mode never decays.
Damping rate multiplies by mass instead of dividing · case 01
Heavy bodies appear strongly damped and light ones barely damped.
Damping rate multiplies by mass instead of dividing · case 02
Heavy bodies appear strongly damped and light ones barely damped.
Damping rate multiplies by mass instead of dividing · case 03
Heavy bodies appear strongly damped and light ones barely damped.
Damping rate multiplies by mass instead of dividing · case 04
Heavy bodies appear strongly damped and light ones barely damped.
Damping rate multiplies by mass instead of dividing · case 05
Heavy bodies appear strongly damped and light ones barely damped.
Springs attached to a static anchor are ignored when limiting the step · case 01
Stiff tethers to the world explode because they never shrink the substep.
Springs attached to a static anchor are ignored when limiting the step · case 02
Stiff tethers to the world explode because they never shrink the substep.
Springs attached to a static anchor are ignored when limiting the step · case 03
Stiff tethers to the world explode because they never shrink the substep.
Springs attached to a static anchor are ignored when limiting the step · case 04
Stiff tethers to the world explode because they never shrink the substep.
Springs attached to a static anchor are ignored when limiting the step · case 05
Stiff tethers to the world explode because they never shrink the substep.
Effective spring mass uses the larger inverse mass only · case 01
Two light bodies connected by a stiff spring get a substep that is too long.
Effective spring mass uses the larger inverse mass only · case 02
Two light bodies connected by a stiff spring get a substep that is too long.
Effective spring mass uses the larger inverse mass only · case 03
Two light bodies connected by a stiff spring get a substep that is too long.
Effective spring mass uses the larger inverse mass only · case 04
Two light bodies connected by a stiff spring get a substep that is too long.
Effective spring mass uses the larger inverse mass only · case 05
Two light bodies connected by a stiff spring get a substep that is too long.
Critical step divides by k*w without the square root · case 01
Stiffness changes the substep quadratically wrong, overshooting or wasting substeps.
Critical step divides by k*w without the square root · case 02
Stiffness changes the substep quadratically wrong, overshooting or wasting substeps.
Critical step divides by k*w without the square root · case 03
Stiffness changes the substep quadratically wrong, overshooting or wasting substeps.
Critical step divides by k*w without the square root · case 04
Stiffness changes the substep quadratically wrong, overshooting or wasting substeps.
Critical step divides by k*w without the square root · case 05
Stiffness changes the substep quadratically wrong, overshooting or wasting substeps.
Critical step uses one over omega instead of two over omega · case 01
The solver runs twice as many substeps as the contract bound requires.
Critical step uses one over omega instead of two over omega · case 02
The solver runs twice as many substeps as the contract bound requires.
Critical step uses one over omega instead of two over omega · case 03
The solver runs twice as many substeps as the contract bound requires.
Critical step uses one over omega instead of two over omega · case 04
The solver runs twice as many substeps as the contract bound requires.
Critical step uses one over omega instead of two over omega · case 05
The solver runs twice as many substeps as the contract bound requires.
Substep count floors the frame to critical step ratio · case 01
Frames needing 2.3 critical steps run only two, each above the stability limit.
Substep count floors the frame to critical step ratio · case 02
Frames needing 2.3 critical steps run only two, each above the stability limit.
Substep count floors the frame to critical step ratio · case 03
Frames needing 2.3 critical steps run only two, each above the stability limit.
Substep count floors the frame to critical step ratio · case 04
Frames needing 2.3 critical steps run only two, each above the stability limit.
Substep count floors the frame to critical step ratio · case 05
Frames needing 2.3 critical steps run only two, each above the stability limit.
Substep clamp flag is evaluated after clamping · case 01
The caller is never told the frame ran above the stability limit.
Substep clamp flag is evaluated after clamping · case 02
The caller is never told the frame ran above the stability limit.
Substep clamp flag is evaluated after clamping · case 03
The caller is never told the frame ran above the stability limit.
Substep clamp flag is evaluated after clamping · case 04
The caller is never told the frame ran above the stability limit.
Substep clamp flag is evaluated after clamping · case 05
The caller is never told the frame ran above the stability limit.
Critical step keeps the loosest spring rather than the stiffest · case 01
A single stiff spring in a soft network is integrated with the soft spring step.
Critical step keeps the loosest spring rather than the stiffest · case 02
A single stiff spring in a soft network is integrated with the soft spring step.
Critical step keeps the loosest spring rather than the stiffest · case 03
A single stiff spring in a soft network is integrated with the soft spring step.
Critical step keeps the loosest spring rather than the stiffest · case 04
A single stiff spring in a soft network is integrated with the soft spring step.
Critical step keeps the loosest spring rather than the stiffest · case 05
A single stiff spring in a soft network is integrated with the soft spring step.
Accumulator requires strictly more than one step of time · case 01
A frame exactly one fixed step long runs no physics step and the renderer lags a frame.
Accumulator requires strictly more than one step of time · case 02
A frame exactly one fixed step long runs no physics step and the renderer lags a frame.
Accumulator requires strictly more than one step of time · case 03
A frame exactly one fixed step long runs no physics step and the renderer lags a frame.
Accumulator requires strictly more than one step of time · case 04
A frame exactly one fixed step long runs no physics step and the renderer lags a frame.
Accumulator requires strictly more than one step of time · case 05
A frame exactly one fixed step long runs no physics step and the renderer lags a frame.
Accumulator adds unclamped frame deltas · case 01
A debugger pause produces a burst of dropped time and a huge catch-up.
Accumulator adds unclamped frame deltas · case 02
A debugger pause produces a burst of dropped time and a huge catch-up.
Accumulator adds unclamped frame deltas · case 03
A debugger pause produces a burst of dropped time and a huge catch-up.
Accumulator adds unclamped frame deltas · case 04
A debugger pause produces a burst of dropped time and a huge catch-up.
Accumulator adds unclamped frame deltas · case 05
A debugger pause produces a burst of dropped time and a huge catch-up.
Leftover whole steps are carried into the next frame after the step budget · case 01
After a long frame the accumulator keeps backlog and every later frame runs at the budget limit.
Leftover whole steps are carried into the next frame after the step budget · case 02
After a long frame the accumulator keeps backlog and every later frame runs at the budget limit.
Leftover whole steps are carried into the next frame after the step budget · case 03
After a long frame the accumulator keeps backlog and every later frame runs at the budget limit.
Leftover whole steps are carried into the next frame after the step budget · case 04
After a long frame the accumulator keeps backlog and every later frame runs at the budget limit.
Leftover whole steps are carried into the next frame after the step budget · case 05
After a long frame the accumulator keeps backlog and every later frame runs at the budget limit.
Step budget counter is not reset each frame · case 01
After a few frames the budget is exhausted permanently and physics stops.
Step budget counter is not reset each frame · case 02
After a few frames the budget is exhausted permanently and physics stops.
Step budget counter is not reset each frame · case 03
After a few frames the budget is exhausted permanently and physics stops.
Step budget counter is not reset each frame · case 04
After a few frames the budget is exhausted permanently and physics stops.
Step budget counter is not reset each frame · case 05
After a few frames the budget is exhausted permanently and physics stops.
Render interpolation factor is returned as raw leftover seconds · case 01
Interpolation weights are tiny and rendering visibly snaps between states.
Render interpolation factor is returned as raw leftover seconds · case 02
Interpolation weights are tiny and rendering visibly snaps between states.
Render interpolation factor is returned as raw leftover seconds · case 03
Interpolation weights are tiny and rendering visibly snaps between states.
Render interpolation factor is returned as raw leftover seconds · case 04
Interpolation weights are tiny and rendering visibly snaps between states.
Render interpolation factor is returned as raw leftover seconds · case 05
Interpolation weights are tiny and rendering visibly snaps between states.
Dropped-time statistic records the kept remainder · case 01
Telemetry reports tiny losses while the simulation actually discarded whole steps.
Dropped-time statistic records the kept remainder · case 02
Telemetry reports tiny losses while the simulation actually discarded whole steps.
Dropped-time statistic records the kept remainder · case 03
Telemetry reports tiny losses while the simulation actually discarded whole steps.
Dropped-time statistic records the kept remainder · case 04
Telemetry reports tiny losses while the simulation actually discarded whole steps.
Dropped-time statistic records the kept remainder · case 05
Telemetry reports tiny losses while the simulation actually discarded whole steps.
INSPECTABLE BY DESIGN
Every result has a runnable source.
Runnable implementations with recorded outputs, source hashes, and explicit contracts. Related variants share a failure mechanism and belong together in evaluation splits.
Read the methodology ↗