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
Grover planner omits the initial rotation from the success angle · case 01
Success probabilities are reported for angle 2k theta, underestimating the achievable probability.
Grover planner omits the initial rotation from the success angle · case 02
Success probabilities are reported for angle 2k theta, underestimating the achievable probability.
Grover planner omits the initial rotation from the success angle · case 03
Success probabilities are reported for angle 2k theta, underestimating the achievable probability.
Grover planner omits the initial rotation from the success angle · case 04
Success probabilities are reported for angle 2k theta, underestimating the achievable probability.
Grover planner omits the initial rotation from the success angle · case 05
Success probabilities are reported for angle 2k theta, underestimating the achievable probability.
Grover planner rejects a fully marked search space · case 01
N=8, M=8 is reported as "invalid" instead of the trivial zero-iteration plan with success 1.
Grover planner rejects a fully marked search space · case 02
N=8, M=8 is reported as "invalid" instead of the trivial zero-iteration plan with success 1.
Grover planner rejects a fully marked search space · case 03
N=8, M=8 is reported as "invalid" instead of the trivial zero-iteration plan with success 1.
Grover planner rejects a fully marked search space · case 04
N=8, M=8 is reported as "invalid" instead of the trivial zero-iteration plan with success 1.
Grover planner rejects a fully marked search space · case 05
N=8, M=8 is reported as "invalid" instead of the trivial zero-iteration plan with success 1.
Grover planner reports zero marked items as invalid input · case 01
M = 0 yields "invalid" instead of "no-solution".
Grover planner reports zero marked items as invalid input · case 02
M = 0 yields "invalid" instead of "no-solution".
Grover planner reports zero marked items as invalid input · case 03
M = 0 yields "invalid" instead of "no-solution".
Grover planner reports zero marked items as invalid input · case 04
M = 0 yields "invalid" instead of "no-solution".
Grover planner reports zero marked items as invalid input · case 05
M = 0 yields "invalid" instead of "no-solution".
Grover planner rounds the iteration count to nearest · case 01
Some sizes plan one extra iteration, overshooting the peak and lowering success.
Grover planner rounds the iteration count to nearest · case 02
Some sizes plan one extra iteration, overshooting the peak and lowering success.
Grover planner rounds the iteration count to nearest · case 03
Some sizes plan one extra iteration, overshooting the peak and lowering success.
Grover planner rounds the iteration count to nearest · case 04
Some sizes plan one extra iteration, overshooting the peak and lowering success.
Grover planner rounds the iteration count to nearest · case 05
Some sizes plan one extra iteration, overshooting the peak and lowering success.
Error budget charges virtual RZ frame changes · case 01
A circuit of only RZ gates reports success below 1.
Error budget charges virtual RZ frame changes · case 02
A circuit of only RZ gates reports success below 1.
Error budget charges virtual RZ frame changes · case 03
A circuit of only RZ gates reports success below 1.
Error budget charges virtual RZ frame changes · case 04
A circuit of only RZ gates reports success below 1.
Error budget charges virtual RZ frame changes · case 05
A circuit of only RZ gates reports success below 1.
Error budget charges a CNOT once per operand · case 01
Each CNOT multiplies the success by (1-e2)^2.
Error budget charges a CNOT once per operand · case 02
Each CNOT multiplies the success by (1-e2)^2.
Error budget charges a CNOT once per operand · case 03
Each CNOT multiplies the success by (1-e2)^2.
Error budget charges a CNOT once per operand · case 04
Each CNOT multiplies the success by (1-e2)^2.
Error budget charges a CNOT once per operand · case 05
Each CNOT multiplies the success by (1-e2)^2.
Error budget charges a register measurement once · case 01
Measuring three qubits in one op costs (1-er) instead of (1-er)^3.
Error budget charges a register measurement once · case 02
Measuring three qubits in one op costs (1-er) instead of (1-er)^3.
Error budget charges a register measurement once · case 03
Measuring three qubits in one op costs (1-er) instead of (1-er)^3.
Error budget charges a register measurement once · case 04
Measuring three qubits in one op costs (1-er) instead of (1-er)^3.
Error budget charges a register measurement once · case 05
Measuring three qubits in one op costs (1-er) instead of (1-er)^3.
Error budget prices a Toffoli like one CNOT · case 01
A Toffoli costs (1-e2) instead of its six-CNOT, nine-single-qubit decomposition.
Error budget prices a Toffoli like one CNOT · case 02
A Toffoli costs (1-e2) instead of its six-CNOT, nine-single-qubit decomposition.
Error budget prices a Toffoli like one CNOT · case 03
A Toffoli costs (1-e2) instead of its six-CNOT, nine-single-qubit decomposition.
Error budget prices a Toffoli like one CNOT · case 04
A Toffoli costs (1-e2) instead of its six-CNOT, nine-single-qubit decomposition.
Error budget prices a Toffoli like one CNOT · case 05
A Toffoli costs (1-e2) instead of its six-CNOT, nine-single-qubit decomposition.
Amplitude damping relaxes toward |1> · case 01
Damping |1> with g = 0.5 gives z = -1 instead of 0; the channel pushes population the wrong way.
Amplitude damping relaxes toward |1> · case 02
Damping |1> with g = 0.5 gives z = -1 instead of 0; the channel pushes population the wrong way.
Amplitude damping relaxes toward |1> · case 03
Damping |1> with g = 0.5 gives z = -1 instead of 0; the channel pushes population the wrong way.
Amplitude damping relaxes toward |1> · case 04
Damping |1> with g = 0.5 gives z = -1 instead of 0; the channel pushes population the wrong way.
Amplitude damping relaxes toward |1> · case 05
Damping |1> with g = 0.5 gives z = -1 instead of 0; the channel pushes population the wrong way.
Amplitude damping shrinks coherences linearly · case 01
Coherences decay as (1-g) instead of sqrt(1-g), so T2 appears equal to T1/2 everywhere.
Amplitude damping shrinks coherences linearly · case 02
Coherences decay as (1-g) instead of sqrt(1-g), so T2 appears equal to T1/2 everywhere.
Amplitude damping shrinks coherences linearly · case 03
Coherences decay as (1-g) instead of sqrt(1-g), so T2 appears equal to T1/2 everywhere.
Amplitude damping shrinks coherences linearly · case 04
Coherences decay as (1-g) instead of sqrt(1-g), so T2 appears equal to T1/2 everywhere.
Amplitude damping shrinks coherences linearly · case 05
Coherences decay as (1-g) instead of sqrt(1-g), so T2 appears equal to T1/2 everywhere.
Dephasing uses 1-p instead of 1-2p · case 01
A phase-flip probability of 0.5 leaves half the coherence instead of erasing it.
Dephasing uses 1-p instead of 1-2p · case 02
A phase-flip probability of 0.5 leaves half the coherence instead of erasing it.
Dephasing uses 1-p instead of 1-2p · case 03
A phase-flip probability of 0.5 leaves half the coherence instead of erasing it.
Dephasing uses 1-p instead of 1-2p · case 04
A phase-flip probability of 0.5 leaves half the coherence instead of erasing it.
Dephasing uses 1-p instead of 1-2p · case 05
A phase-flip probability of 0.5 leaves half the coherence instead of erasing it.
Bit-flip channel contracts the x and y axes · case 01
A bit flip leaves |0> at z = 1 while shrinking x.
Bit-flip channel contracts the x and y axes · case 02
A bit flip leaves |0> at z = 1 while shrinking x.
Bit-flip channel contracts the x and y axes · case 03
A bit flip leaves |0> at z = 1 while shrinking x.
Bit-flip channel contracts the x and y axes · case 04
A bit flip leaves |0> at z = 1 while shrinking x.
Bit-flip channel contracts the x and y axes · case 05
A bit flip leaves |0> at z = 1 while shrinking x.
Bloch Hadamard leaves y unchanged · case 01
H applied to |+i> stays at y = +1 instead of moving to y = -1.
Bloch Hadamard leaves y unchanged · case 02
H applied to |+i> stays at y = +1 instead of moving to y = -1.
Bloch Hadamard leaves y unchanged · case 03
H applied to |+i> stays at y = +1 instead of moving to y = -1.
Bloch Hadamard leaves y unchanged · case 04
H applied to |+i> stays at y = +1 instead of moving to y = -1.
Bloch Hadamard leaves y unchanged · case 05
H applied to |+i> stays at y = +1 instead of moving to y = -1.
Bloch S gate rotates clockwise · case 01
S applied to |+> lands at y = -1 instead of y = +1.
Bloch S gate rotates clockwise · case 02
S applied to |+> lands at y = -1 instead of y = +1.
Bloch S gate rotates clockwise · case 03
S applied to |+> lands at y = -1 instead of y = +1.
Bloch S gate rotates clockwise · case 04
S applied to |+> lands at y = -1 instead of y = +1.
Bloch S gate rotates clockwise · case 05
S applied to |+> lands at y = -1 instead of y = +1.
Rotation merger normalizes pi to minus pi · case 01
rz(pi/2) followed by rz(pi/2) is emitted as rz(-1) instead of rz(1).
Rotation merger normalizes pi to minus pi · case 02
rz(pi/2) followed by rz(pi/2) is emitted as rz(-1) instead of rz(1).
Rotation merger normalizes pi to minus pi · case 03
rz(pi/2) followed by rz(pi/2) is emitted as rz(-1) instead of rz(1).
Rotation merger normalizes pi to minus pi · case 04
rz(pi/2) followed by rz(pi/2) is emitted as rz(-1) instead of rz(1).
Rotation merger normalizes pi to minus pi · case 05
rz(pi/2) followed by rz(pi/2) is emitted as rz(-1) instead of rz(1).
Rotation merger lets rz slide past a CNOT target · case 01
rz on qubit 1 before and after cx(0,1) are merged although they do not commute.
Rotation merger lets rz slide past a CNOT target · case 02
rz on qubit 1 before and after cx(0,1) are merged although they do not commute.
Rotation merger lets rz slide past a CNOT target · case 03
rz on qubit 1 before and after cx(0,1) are merged although they do not commute.
Rotation merger lets rz slide past a CNOT target · case 04
rz on qubit 1 before and after cx(0,1) are merged although they do not commute.
Rotation merger lets rz slide past a CNOT target · case 05
rz on qubit 1 before and after cx(0,1) are merged although they do not commute.
Rotation merger keeps a merged identity rotation · case 01
rz(3pi/4) followed by rz(5pi/4) leaves an rz(0) in the output.
Rotation merger keeps a merged identity rotation · case 02
rz(3pi/4) followed by rz(5pi/4) leaves an rz(0) in the output.
Rotation merger keeps a merged identity rotation · case 03
rz(3pi/4) followed by rz(5pi/4) leaves an rz(0) in the output.
Rotation merger keeps a merged identity rotation · case 04
rz(3pi/4) followed by rz(5pi/4) leaves an rz(0) in the output.
Rotation merger keeps a merged identity rotation · case 05
rz(3pi/4) followed by rz(5pi/4) leaves an rz(0) in the output.
Rotation merger leaves isolated angles unnormalized · case 01
A lone rz(5pi/2) is emitted as "5/2" instead of "1/2".
Rotation merger leaves isolated angles unnormalized · case 02
A lone rz(5pi/2) is emitted as "5/2" instead of "1/2".
Rotation merger leaves isolated angles unnormalized · case 03
A lone rz(5pi/2) is emitted as "5/2" instead of "1/2".
Rotation merger leaves isolated angles unnormalized · case 04
A lone rz(5pi/2) is emitted as "5/2" instead of "1/2".
Rotation merger leaves isolated angles unnormalized · case 05
A lone rz(5pi/2) is emitted as "5/2" instead of "1/2".
Rotation merger cancels X gates on different qubits · case 01
x on qubit 0 followed by x on qubit 1 cancels both.
Rotation merger cancels X gates on different qubits · case 02
x on qubit 0 followed by x on qubit 1 cancels both.
Rotation merger cancels X gates on different qubits · case 03
x on qubit 0 followed by x on qubit 1 cancels both.
Rotation merger cancels X gates on different qubits · case 04
x on qubit 0 followed by x on qubit 1 cancels both.
Rotation merger cancels X gates on different qubits · case 05
x on qubit 0 followed by x on qubit 1 cancels both.
Register expansion advances the qubit counter for classical registers · case 01
Declaring a creg between two qregs shifts every qubit of the second qreg by the creg size.
Register expansion advances the qubit counter for classical registers · case 02
Declaring a creg between two qregs shifts every qubit of the second qreg by the creg size.
Register expansion advances the qubit counter for classical registers · case 03
Declaring a creg between two qregs shifts every qubit of the second qreg by the creg size.
Register expansion advances the qubit counter for classical registers · case 04
Declaring a creg between two qregs shifts every qubit of the second qreg by the creg size.
Register expansion advances the qubit counter for classical registers · case 05
Declaring a creg between two qregs shifts every qubit of the second qreg by the creg size.
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 ↗