SDIRK methods measured off-archive, and the stiff problems that motivate them.
Implicit methods solve an equation at every stage, paying for a Jacobian and Newton iterations to stay stable on stiff problems. A scored record has no field for them, so the run searches two-stage SDIRK methods in a lane of its own: cycles set aside for this class that write an unscored archive. The lane reports the modeled cycles a method needs to reach an accuracy target.
Source: side-track job sdirk.gamma_dyadic_scan, 9 ledger points, 81 distinct gammas plotted. Arithmetic: tableau and stability algebra exact over Fractions; the measured order is float64.
| gamma | R at infinity | A-stable | measured order |
|---|---|---|---|
| 1/32 | 449 | no | 2.1001 |
| 1/16 | 97 | no | 2.0912 |
| 3/32 | 329/9 | no | 2.0835 |
| 1/8 | 17 | no | 2.0779 |
| 5/32 | 217/25 | no | 2.0769 |
| 11/64 | 761/121 | no | 2.0811 |
| 3/16 | 41/9 | no | 2.096 |
| 13/64 | 553/169 | no | 2.1735 |
| 7/32 | 113/49 | no | 1.8356 |
| 29/128 | 1609/841 | no | 1.9477 |
| 15/64 | 353/225 | no | 1.9779 |
| 31/128 | 1217/961 | no | 1.9912 |
| 1/4 | 1 | yes | 1.9982 |
| 33/128 | 833/1089 | yes | 2.0023 |
| 67/256 | 2953/4489 | yes | 2.0036 |
| 17/64 | 161/289 | yes | 2.0046 |
| 69/256 | 2201/4761 | yes | 2.0054 |
| 35/128 | 457/1225 | yes | 2.0059 |
| 71/256 | 1457/5041 | yes | 2.0064 |
| 143/512 | 5089/20449 | yes | 2.0065 |
| 9/32 | 17/81 | yes | 2.0066 |
| 145/512 | 3617/21025 | yes | 2.0067 |
| 73/256 | 721/5329 | yes | 2.0068 |
| 293/1024 | 10073/85849 | yes | 2.0068 |
| 147/512 | 2153/21609 | yes | 2.0069 |
| 295/1024 | 7153/87025 | yes | 2.0069 |
| 37/128 | 89/1369 | yes | 2.0069 |
| 593/2048 | 19873/351649 | yes | 2.0069 |
| 297/1024 | 4241/88209 | yes | 2.0069 |
| 595/2048 | 14057/354025 | yes | 2.0069 |
| 149/512 | 697/22201 | yes | 2.0069 |
| 1193/4096 | 38801/1423249 | yes | 2.0069 |
| 597/2048 | 8249/356409 | yes | 2.0069 |
| 1195/4096 | 27193/1428025 | yes | 2.0069 |
| 299/1024 | 1337/89401 | yes | 2.0069 |
| 1197/4096 | 15593/1432809 | yes | 2.0069 |
| 599/2048 | 2449/358801 | yes | 2.0069 |
| 1199/4096 | 4001/1437601 | yes | 2.0069 |
| 75/256 | -7/5625 | yes | 2.0069 |
| 1201/4096 | -7583/1442401 | yes | 2.0069 |
| 601/2048 | -3343/361201 | yes | 2.0069 |
| 1203/4096 | -19159/1447209 | yes | 2.0069 |
| 301/1024 | -1559/90601 | yes | 2.0069 |
| 1205/4096 | -30727/1452025 | yes | 2.0069 |
| 603/2048 | -9127/363609 | yes | 2.0069 |
| 1207/4096 | -42287/1456849 | yes | 2.0069 |
| 151/512 | -751/22801 | yes | 2.0069 |
| 605/2048 | -14903/366025 | yes | 2.0069 |
| 303/1024 | -4447/91809 | yes | 2.0069 |
| 607/2048 | -20671/368449 | yes | 2.0069 |
| 19/64 | -23/361 | yes | 2.0069 |
| 305/1024 | -7327/93025 | yes | 2.0068 |
| 153/512 | -2191/23409 | yes | 2.0068 |
| 307/1024 | -10199/94249 | yes | 2.0068 |
| 77/256 | -727/5929 | yes | 2.0067 |
| 155/512 | -3623/24025 | yes | 2.0067 |
| 39/128 | -271/1521 | yes | 2.0066 |
| 157/512 | -5047/24649 | yes | 2.0065 |
| 79/256 | -1439/6241 | yes | 2.0064 |
| 5/16 | -7/25 | yes | 2.0062 |
| 81/256 | -2143/6561 | yes | 2.0059 |
| 41/128 | -623/1681 | yes | 2.0056 |
| 83/256 | -2839/6889 | yes | 2.0053 |
| 21/64 | -199/441 | yes | 2.005 |
| 43/128 | -967/1849 | yes | 2.0043 |
| 11/32 | -71/121 | yes | 2.0036 |
| 45/128 | -1303/2025 | yes | 2.0028 |
| 23/64 | -367/529 | yes | 2.0021 |
| 3/8 | -7/9 | yes | 2.0005 |
| 25/64 | -527/625 | yes | 1.999 |
| 13/32 | -151/169 | yes | 1.9976 |
| 27/64 | -679/729 | yes | 1.9963 |
| 7/16 | -47/49 | yes | 1.9952 |
| 15/32 | -223/225 | yes | 1.9936 |
| 1/2 | -1 | yes | 1.9929 |
| 17/32 | -287/289 | yes | 1.9937 |
| 9/16 | -79/81 | yes | 1.9963 |
| 5/8 | -23/25 | yes | 2.0105 |
| 11/16 | -103/121 | yes | 2.0514 |
| 3/4 | -7/9 | yes | 2.205 |
| 13/16 | -119/169 | yes | -2.7267 |
Source: benchmark/results.json, implicit_budget. Arithmetic: float64. Cycle grade: design_estimate.
| problem | rung | steps | analytic cycles | final-state error | status |
|---|---|---|---|---|---|
| enzyme_qssa | 1/4 of the budget | 44 | 16,368 | 2.35477e-05 | ok |
| enzyme_qssa | 1/2 of the budget | 88 | 32,736 | 5.85479e-06 | ok |
| enzyme_qssa | the full budget | 176 | 65,472 | 1.45953e-06 | ok |
| enzyme_qssa | twice the budget | 352 | 130,944 | 3.64339e-07 | ok |
| enzyme_qssa | 4 times the budget | 704 | 261,888 | 9.10147e-08 | ok |
| robertson_scaled | 1/4 of the budget | 25 | 16,300 | no error | diverged |
| robertson_scaled | 1/2 of the budget | 50 | 32,600 | no error | diverged |
| robertson_scaled | the full budget | 100 | 65,200 | no error | diverged |
| robertson_scaled | twice the budget | 200 | 130,400 | no error | diverged |
| robertson_scaled | 4 times the budget | 400 | 260,800 | 4.99529e-06 | ok |
| servo_load_step | 1/4 of the budget | 44 | 16,368 | 2.9464e-05 | ok |
| servo_load_step | 1/2 of the budget | 88 | 32,736 | 7.33205e-06 | ok |
| servo_load_step | the full budget | 176 | 65,472 | 1.82872e-06 | ok |
| servo_load_step | twice the budget | 352 | 130,944 | 4.56642e-07 | ok |
| servo_load_step | 4 times the budget | 704 | 261,888 | 1.14093e-07 | ok |
| problem | steps at budget | cycles per step | error at budget | status |
|---|---|---|---|---|
| enzyme_qssa | 176 | 372 | 1.45953e-06 | ok |
| robertson_scaled | 100 | 652 | n/a | diverged |
| servo_load_step | 176 | 372 | 1.82872e-06 | ok |
Source: side-track job sdirk.jacobian_cost, 3 points; cycle model m0plus_fast. Arithmetic: float64 only; no Q15 effects are included.
| problem | Jacobian | where the cycles go | cycles per step | step total | rhs evaluations per step |
|---|---|---|---|---|---|
| enzyme_qssa | analytic | Newton iterations | 204 | 348 | 6 |
| enzyme_qssa | analytic | LU factorization | 84 | 348 | 6 |
| enzyme_qssa | analytic | other stage arithmetic | 60 | 348 | 6 |
| enzyme_qssa | finite difference | Newton iterations | 204 | 372 | 8 |
| enzyme_qssa | finite difference | LU factorization | 84 | 372 | 8 |
| enzyme_qssa | finite difference | other stage arithmetic | 60 | 372 | 8 |
| enzyme_qssa | finite difference | finite-difference Jacobian | 24 | 372 | 8 |
| robertson_scaled | analytic | Newton iterations | 342 | 598 | 6 |
| robertson_scaled | analytic | LU factorization | 148 | 598 | 6 |
| robertson_scaled | analytic | other stage arithmetic | 108 | 598 | 6 |
| robertson_scaled | finite difference | Newton iterations | 342 | 652 | 9 |
| robertson_scaled | finite difference | LU factorization | 148 | 652 | 9 |
| robertson_scaled | finite difference | other stage arithmetic | 108 | 652 | 9 |
| robertson_scaled | finite difference | finite-difference Jacobian | 54 | 652 | 9 |
| servo_load_step | analytic | Newton iterations | 204 | 348 | 6 |
| servo_load_step | analytic | LU factorization | 84 | 348 | 6 |
| servo_load_step | analytic | other stage arithmetic | 60 | 348 | 6 |
| servo_load_step | finite difference | Newton iterations | 204 | 372 | 8 |
| servo_load_step | finite difference | LU factorization | 84 | 372 | 8 |
| servo_load_step | finite difference | other stage arithmetic | 60 | 372 | 8 |
| servo_load_step | finite difference | finite-difference Jacobian | 24 | 372 | 8 |
Our solvers reached 44 of their 44 targets and the library solvers 96 of their 96; a target is one solver on one problem at one accuracy level. A further 20 rows were skipped and never ran: no analytic Jacobian is derived for this problem. Where our runs carry a cycle count, a reached target costs a median of 3162 analytic cycles; the table gives each row's cost grade.
Source: benchmark/results.json, matched_accuracy (implicit rows); arithmetic per solver in the legend.
| problem | solver | side | rhs evaluations | achieved error | targets met |
|---|---|---|---|---|---|
| battery_2rc | sdirk2_fd_jac | ours | 27 | 0.00656783 | 0.015625 |
| battery_2rc | sdirk2_fd_jac | ours | 36 | 0.00358315 | 0.00390625 |
| battery_2rc | sdirk2_fd_jac | ours | 72 | 0.000856928 | 0.000976562 |
| battery_2rc | sdirk2_fd_jac | ours | 135 | 0.000238932 | 0.000244141 |
| battery_2rc | Radau | library | 51 | 0.000219398 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| battery_2rc | BDF | library | 32 | 0.00656401 | 0.015625 |
| battery_2rc | BDF | library | 42 | 0.000325319 | 0.00390625, 0.000976562 |
| battery_2rc | BDF | library | 54 | 0.000117813 | 0.000244141 |
| battery_2rc | LSODA | library | 17 | 0.00291604 | 0.015625, 0.00390625 |
| battery_2rc | LSODA | library | 21 | 0.000584159 | 0.000976562 |
| battery_2rc | LSODA | library | 29 | 6.01907e-05 | 0.000244141 |
| bicycle_lateral | sdirk2_fd_jac | ours | 32 | 0.000578592 | 0.015625, 0.00390625, 0.000976562 |
| bicycle_lateral | sdirk2_fd_jac | ours | 40 | 0.000220149 | 0.000244141 |
| bicycle_lateral | Radau | library | 72 | 2.18159e-05 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| bicycle_lateral | BDF | library | 46 | 0.001081 | 0.015625, 0.00390625 |
| bicycle_lateral | BDF | library | 60 | 9.62027e-05 | 0.000976562, 0.000244141 |
| bicycle_lateral | LSODA | library | 70 | 6.05922e-05 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| buck_converter | sdirk2_fd_jac | ours | 72 | 0.0117345 | 0.015625 |
| buck_converter | sdirk2_fd_jac | ours | 144 | 0.00378398 | 0.00390625 |
| buck_converter | sdirk2_fd_jac | ours | 248 | 0.000932933 | 0.000976562 |
| buck_converter | sdirk2_fd_jac | ours | 464 | 0.000239097 | 0.000244141 |
| buck_converter | Radau | library | 114 | 0.000478834 | 0.015625, 0.00390625, 0.000976562 |
| buck_converter | Radau | library | 170 | 6.67781e-05 | 0.000244141 |
| buck_converter | BDF | library | 78 | 0.0140095 | 0.015625 |
| buck_converter | BDF | library | 110 | 0.00143434 | 0.00390625 |
| buck_converter | BDF | library | 150 | 0.000219837 | 0.000976562, 0.000244141 |
| buck_converter | LSODA | library | 112 | 0.00117978 | 0.015625, 0.00390625 |
| buck_converter | LSODA | library | 123 | 0.0006592 | 0.000976562 |
| buck_converter | LSODA | library | 143 | 0.000198731 | 0.000244141 |
| enzyme_qssa | sdirk2_analytic_jac | ours | 18 | 0.00584014 | 0.015625 |
| enzyme_qssa | sdirk2_analytic_jac | ours | 24 | 0.0031586 | 0.00390625 |
| enzyme_qssa | sdirk2_analytic_jac | ours | 48 | 0.000746402 | 0.000976562 |
| enzyme_qssa | sdirk2_analytic_jac | ours | 84 | 0.000237889 | 0.000244141 |
| enzyme_qssa | sdirk2_fd_jac | ours | 24 | 0.00584014 | 0.015625 |
| enzyme_qssa | sdirk2_fd_jac | ours | 32 | 0.0031586 | 0.00390625 |
| enzyme_qssa | sdirk2_fd_jac | ours | 64 | 0.000746402 | 0.000976562 |
| enzyme_qssa | sdirk2_fd_jac | ours | 112 | 0.000237889 | 0.000244141 |
| enzyme_qssa | Radau | library | 50 | 5.92607e-05 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| enzyme_qssa | BDF | library | 35 | 0.00177838 | 0.015625, 0.00390625 |
| enzyme_qssa | BDF | library | 52 | 0.000338146 | 0.000976562 |
| enzyme_qssa | BDF | library | 85 | 3.47415e-06 | 0.000244141 |
| enzyme_qssa | LSODA | library | 77 | 0.00309634 | 0.015625, 0.00390625 |
| enzyme_qssa | LSODA | library | 86 | 0.000105883 | 0.000976562, 0.000244141 |
| glucose_minimal | sdirk2_fd_jac | ours | 24 | 0.00962292 | 0.015625 |
| glucose_minimal | sdirk2_fd_jac | ours | 40 | 0.00355833 | 0.00390625 |
| glucose_minimal | sdirk2_fd_jac | ours | 80 | 0.000913168 | 0.000976562 |
| glucose_minimal | sdirk2_fd_jac | ours | 160 | 0.00023147 | 0.000244141 |
| glucose_minimal | Radau | library | 54 | 5.79763e-06 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| glucose_minimal | BDF | library | 42 | 0.00048351 | 0.015625, 0.00390625, 0.000976562 |
| glucose_minimal | BDF | library | 56 | 6.60237e-05 | 0.000244141 |
| glucose_minimal | LSODA | library | 37 | 0.000444164 | 0.015625, 0.00390625, 0.000976562 |
| glucose_minimal | LSODA | library | 45 | 6.83757e-05 | 0.000244141 |
| pll_lock | sdirk2_fd_jac | ours | 32 | 0.0142631 | 0.015625 |
| pll_lock | sdirk2_fd_jac | ours | 48 | 0.00310806 | 0.00390625 |
| pll_lock | sdirk2_fd_jac | ours | 80 | 0.000760809 | 0.000976562 |
| pll_lock | sdirk2_fd_jac | ours | 136 | 0.000229534 | 0.000244141 |
| pll_lock | Radau | library | 72 | 3.85814e-05 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| pll_lock | BDF | library | 48 | 0.000716822 | 0.015625, 0.00390625, 0.000976562 |
| pll_lock | BDF | library | 66 | 0.000172684 | 0.000244141 |
| pll_lock | LSODA | library | 54 | 0.000531216 | 0.015625, 0.00390625, 0.000976562 |
| pll_lock | LSODA | library | 65 | 9.11946e-05 | 0.000244141 |
| robertson_scaled | sdirk2_analytic_jac | ours | 1368 | 0.000320808 | 0.015625, 0.00390625, 0.000976562 |
| robertson_scaled | sdirk2_analytic_jac | ours | 1374 | 0.000236606 | 0.000244141 |
| robertson_scaled | sdirk2_fd_jac | ours | 2052 | 0.000320804 | 0.015625, 0.00390625, 0.000976562 |
| robertson_scaled | sdirk2_fd_jac | ours | 2061 | 0.000236605 | 0.000244141 |
| robertson_scaled | Radau | library | 134 | 1.84472e-06 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| robertson_scaled | BDF | library | 88 | 0.000406376 | 0.015625, 0.00390625, 0.000976562 |
| robertson_scaled | BDF | library | 175 | 0.000140963 | 0.000244141 |
| robertson_scaled | LSODA | library | 100 | 0.00213921 | 0.015625, 0.00390625 |
| robertson_scaled | LSODA | library | 110 | 0.000617852 | 0.000976562 |
| robertson_scaled | LSODA | library | 143 | 0.00019588 | 0.000244141 |
| servo_load_step | sdirk2_analytic_jac | ours | 12 | 0.014 | 0.015625 |
| servo_load_step | sdirk2_analytic_jac | ours | 24 | 0.00382851 | 0.00390625 |
| servo_load_step | sdirk2_analytic_jac | ours | 48 | 0.000925481 | 0.000976562 |
| servo_load_step | sdirk2_analytic_jac | ours | 96 | 0.000226304 | 0.000244141 |
| servo_load_step | sdirk2_fd_jac | ours | 16 | 0.014 | 0.015625 |
| servo_load_step | sdirk2_fd_jac | ours | 32 | 0.00382851 | 0.00390625 |
| servo_load_step | sdirk2_fd_jac | ours | 64 | 0.000925481 | 0.000976562 |
| servo_load_step | sdirk2_fd_jac | ours | 128 | 0.000226304 | 0.000244141 |
| servo_load_step | Radau | library | 72 | 8.77343e-05 | 0.015625, 0.00390625, 0.000976562, 0.000244141 |
| servo_load_step | BDF | library | 43 | 0.00276456 | 0.015625, 0.00390625 |
| servo_load_step | BDF | library | 54 | 0.000277037 | 0.000976562 |
| servo_load_step | BDF | library | 80 | 9.04229e-05 | 0.000244141 |
| servo_load_step | LSODA | library | 77 | 0.0029098 | 0.015625, 0.00390625 |
| servo_load_step | LSODA | library | 83 | 0.00039952 | 0.000976562 |
| servo_load_step | LSODA | library | 93 | 0.000110351 | 0.000244141 |
Ours on these rows means run by this harness through its own solver path, which includes the classical rk4 control, so it does not mean discovered.
Library integrators choose their own step counts, so their work is context and never a same-work comparison with any fixed-step run on this site.
Each row fixes a target for the achieved final-state error and reports what the solver spent to reach it. A row that did not reach its target stays in the table, and its status column says so.
| problem | side | solver | arithmetic | target error | achieved | steps | rhs evaluations | analytic cycles | cost grade | status |
|---|---|---|---|---|---|---|---|---|---|---|
| battery_2rc | library | BDF | compiled_float64 | 0.000244141 | 0.000117813 | 24 | 54 | n/a | none | reached |
| battery_2rc | library | BDF | compiled_float64 | 0.000976562 | 0.000325319 | 20 | 42 | n/a | none | reached |
| battery_2rc | library | BDF | compiled_float64 | 0.00390625 | 0.000325319 | 20 | 42 | n/a | none | reached |
| battery_2rc | library | BDF | compiled_float64 | 0.015625 | 0.00656401 | 15 | 32 | n/a | none | reached |
| battery_2rc | library | LSODA | compiled_float64 | 0.000244141 | 6.01907e-05 | 14 | 29 | n/a | none | reached |
| battery_2rc | library | LSODA | compiled_float64 | 0.000976562 | 0.000584159 | 10 | 21 | n/a | none | reached |
| battery_2rc | library | LSODA | compiled_float64 | 0.00390625 | 0.00291604 | 8 | 17 | n/a | none | reached |
| battery_2rc | library | LSODA | compiled_float64 | 0.015625 | 0.00291604 | 8 | 17 | n/a | none | reached |
| battery_2rc | library | Radau | compiled_float64 | 0.000244141 | 0.000219398 | 7 | 51 | n/a | none | reached |
| battery_2rc | library | Radau | compiled_float64 | 0.000976562 | 0.000219398 | 7 | 51 | n/a | none | reached |
| battery_2rc | library | Radau | compiled_float64 | 0.00390625 | 0.000219398 | 7 | 51 | n/a | none | reached |
| battery_2rc | library | Radau | compiled_float64 | 0.015625 | 0.000219398 | 7 | 51 | n/a | none | reached |
| battery_2rc | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000238932 | 15 | 135 | 9780 | design_estimate | reached |
| battery_2rc | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000856928 | 8 | 72 | 5216 | design_estimate | reached |
| battery_2rc | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.00358315 | 4 | 36 | 2608 | design_estimate | reached |
| battery_2rc | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.00656783 | 3 | 27 | 1956 | design_estimate | reached |
| bicycle_lateral | library | BDF | compiled_float64 | 0.000244141 | 9.62027e-05 | 29 | 60 | n/a | none | reached |
| bicycle_lateral | library | BDF | compiled_float64 | 0.000976562 | 9.62027e-05 | 29 | 60 | n/a | none | reached |
| bicycle_lateral | library | BDF | compiled_float64 | 0.00390625 | 0.001081 | 20 | 46 | n/a | none | reached |
| bicycle_lateral | library | BDF | compiled_float64 | 0.015625 | 0.001081 | 20 | 46 | n/a | none | reached |
| bicycle_lateral | library | LSODA | compiled_float64 | 0.000244141 | 6.05922e-05 | 31 | 70 | n/a | none | reached |
| bicycle_lateral | library | LSODA | compiled_float64 | 0.000976562 | 6.05922e-05 | 31 | 70 | n/a | none | reached |
| bicycle_lateral | library | LSODA | compiled_float64 | 0.00390625 | 6.05922e-05 | 31 | 70 | n/a | none | reached |
| bicycle_lateral | library | LSODA | compiled_float64 | 0.015625 | 6.05922e-05 | 31 | 70 | n/a | none | reached |
| bicycle_lateral | library | Radau | compiled_float64 | 0.000244141 | 2.18159e-05 | 10 | 72 | n/a | none | reached |
| bicycle_lateral | library | Radau | compiled_float64 | 0.000976562 | 2.18159e-05 | 10 | 72 | n/a | none | reached |
| bicycle_lateral | library | Radau | compiled_float64 | 0.00390625 | 2.18159e-05 | 10 | 72 | n/a | none | reached |
| bicycle_lateral | library | Radau | compiled_float64 | 0.015625 | 2.18159e-05 | 10 | 72 | n/a | none | reached |
| bicycle_lateral | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000220149 | 5 | 40 | 1860 | design_estimate | reached |
| bicycle_lateral | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000578592 | 4 | 32 | 1488 | design_estimate | reached |
| bicycle_lateral | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.000578592 | 4 | 32 | 1488 | design_estimate | reached |
| bicycle_lateral | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.000578592 | 4 | 32 | 1488 | design_estimate | reached |
| buck_converter | library | BDF | compiled_float64 | 0.000244141 | 0.000219837 | 74 | 150 | n/a | none | reached |
| buck_converter | library | BDF | compiled_float64 | 0.000976562 | 0.000219837 | 74 | 150 | n/a | none | reached |
| buck_converter | library | BDF | compiled_float64 | 0.00390625 | 0.00143434 | 54 | 110 | n/a | none | reached |
| buck_converter | library | BDF | compiled_float64 | 0.015625 | 0.0140095 | 38 | 78 | n/a | none | reached |
| buck_converter | library | LSODA | compiled_float64 | 0.000244141 | 0.000198731 | 70 | 143 | n/a | none | reached |
| buck_converter | library | LSODA | compiled_float64 | 0.000976562 | 0.0006592 | 60 | 123 | n/a | none | reached |
| buck_converter | library | LSODA | compiled_float64 | 0.00390625 | 0.00117978 | 54 | 112 | n/a | none | reached |
| buck_converter | library | LSODA | compiled_float64 | 0.015625 | 0.00117978 | 54 | 112 | n/a | none | reached |
| buck_converter | library | Radau | compiled_float64 | 0.000244141 | 6.67781e-05 | 24 | 170 | n/a | none | reached |
| buck_converter | library | Radau | compiled_float64 | 0.000976562 | 0.000478834 | 16 | 114 | n/a | none | reached |
| buck_converter | library | Radau | compiled_float64 | 0.00390625 | 0.000478834 | 16 | 114 | n/a | none | reached |
| buck_converter | library | Radau | compiled_float64 | 0.015625 | 0.000478834 | 16 | 114 | n/a | none | reached |
| buck_converter | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000239097 | 58 | 464 | 21576 | design_estimate | reached |
| buck_converter | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000932933 | 31 | 248 | 11532 | design_estimate | reached |
| buck_converter | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.00378398 | 18 | 144 | 6696 | design_estimate | reached |
| buck_converter | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.0117345 | 9 | 72 | 3348 | design_estimate | reached |
| enzyme_qssa | library | BDF | compiled_float64 | 0.000244141 | 3.47415e-06 | 33 | 85 | n/a | none | reached |
| enzyme_qssa | library | BDF | compiled_float64 | 0.000976562 | 0.000338146 | 22 | 52 | n/a | none | reached |
| enzyme_qssa | library | BDF | compiled_float64 | 0.00390625 | 0.00177838 | 14 | 35 | n/a | none | reached |
| enzyme_qssa | library | BDF | compiled_float64 | 0.015625 | 0.00177838 | 14 | 35 | n/a | none | reached |
| enzyme_qssa | library | LSODA | compiled_float64 | 0.000244141 | 0.000105883 | 43 | 86 | n/a | none | reached |
| enzyme_qssa | library | LSODA | compiled_float64 | 0.000976562 | 0.000105883 | 43 | 86 | n/a | none | reached |
| enzyme_qssa | library | LSODA | compiled_float64 | 0.00390625 | 0.00309634 | 37 | 77 | n/a | none | reached |
| enzyme_qssa | library | LSODA | compiled_float64 | 0.015625 | 0.00309634 | 37 | 77 | n/a | none | reached |
| enzyme_qssa | library | Radau | compiled_float64 | 0.000244141 | 5.92607e-05 | 6 | 50 | n/a | none | reached |
| enzyme_qssa | library | Radau | compiled_float64 | 0.000976562 | 5.92607e-05 | 6 | 50 | n/a | none | reached |
| enzyme_qssa | library | Radau | compiled_float64 | 0.00390625 | 5.92607e-05 | 6 | 50 | n/a | none | reached |
| enzyme_qssa | library | Radau | compiled_float64 | 0.015625 | 5.92607e-05 | 6 | 50 | n/a | none | reached |
| enzyme_qssa | ours | sdirk2_analytic_jac | float64 | 0.000244141 | 0.000237889 | 14 | 84 | 4872 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_analytic_jac | float64 | 0.000976562 | 0.000746402 | 8 | 48 | 2784 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_analytic_jac | float64 | 0.00390625 | 0.0031586 | 4 | 24 | 1392 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_analytic_jac | float64 | 0.015625 | 0.00584014 | 3 | 18 | 1044 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000237889 | 14 | 112 | 5208 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000746402 | 8 | 64 | 2976 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.0031586 | 4 | 32 | 1488 | design_estimate | reached |
| enzyme_qssa | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.00584014 | 3 | 24 | 1116 | design_estimate | reached |
| glucose_minimal | library | BDF | compiled_float64 | 0.000244141 | 6.60237e-05 | 27 | 56 | n/a | none | reached |
| glucose_minimal | library | BDF | compiled_float64 | 0.000976562 | 0.00048351 | 20 | 42 | n/a | none | reached |
| glucose_minimal | library | BDF | compiled_float64 | 0.00390625 | 0.00048351 | 20 | 42 | n/a | none | reached |
| glucose_minimal | library | BDF | compiled_float64 | 0.015625 | 0.00048351 | 20 | 42 | n/a | none | reached |
| glucose_minimal | library | LSODA | compiled_float64 | 0.000244141 | 6.83757e-05 | 22 | 45 | n/a | none | reached |
| glucose_minimal | library | LSODA | compiled_float64 | 0.000976562 | 0.000444164 | 17 | 37 | n/a | none | reached |
| glucose_minimal | library | LSODA | compiled_float64 | 0.00390625 | 0.000444164 | 17 | 37 | n/a | none | reached |
| glucose_minimal | library | LSODA | compiled_float64 | 0.015625 | 0.000444164 | 17 | 37 | n/a | none | reached |
| glucose_minimal | library | Radau | compiled_float64 | 0.000244141 | 5.79763e-06 | 7 | 54 | n/a | none | reached |
| glucose_minimal | library | Radau | compiled_float64 | 0.000976562 | 5.79763e-06 | 7 | 54 | n/a | none | reached |
| glucose_minimal | library | Radau | compiled_float64 | 0.00390625 | 5.79763e-06 | 7 | 54 | n/a | none | reached |
| glucose_minimal | library | Radau | compiled_float64 | 0.015625 | 5.79763e-06 | 7 | 54 | n/a | none | reached |
| glucose_minimal | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.00023147 | 20 | 160 | 7440 | design_estimate | reached |
| glucose_minimal | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000913168 | 10 | 80 | 3720 | design_estimate | reached |
| glucose_minimal | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.00355833 | 5 | 40 | 1860 | design_estimate | reached |
| glucose_minimal | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.00962292 | 3 | 24 | 1116 | design_estimate | reached |
| pll_lock | library | BDF | compiled_float64 | 0.000244141 | 0.000172684 | 32 | 66 | n/a | none | reached |
| pll_lock | library | BDF | compiled_float64 | 0.000976562 | 0.000716822 | 23 | 48 | n/a | none | reached |
| pll_lock | library | BDF | compiled_float64 | 0.00390625 | 0.000716822 | 23 | 48 | n/a | none | reached |
| pll_lock | library | BDF | compiled_float64 | 0.015625 | 0.000716822 | 23 | 48 | n/a | none | reached |
| pll_lock | library | LSODA | compiled_float64 | 0.000244141 | 9.11946e-05 | 31 | 65 | n/a | none | reached |
| pll_lock | library | LSODA | compiled_float64 | 0.000976562 | 0.000531216 | 25 | 54 | n/a | none | reached |
| pll_lock | library | LSODA | compiled_float64 | 0.00390625 | 0.000531216 | 25 | 54 | n/a | none | reached |
| pll_lock | library | LSODA | compiled_float64 | 0.015625 | 0.000531216 | 25 | 54 | n/a | none | reached |
| pll_lock | library | Radau | compiled_float64 | 0.000244141 | 3.85814e-05 | 10 | 72 | n/a | none | reached |
| pll_lock | library | Radau | compiled_float64 | 0.000976562 | 3.85814e-05 | 10 | 72 | n/a | none | reached |
| pll_lock | library | Radau | compiled_float64 | 0.00390625 | 3.85814e-05 | 10 | 72 | n/a | none | reached |
| pll_lock | library | Radau | compiled_float64 | 0.015625 | 3.85814e-05 | 10 | 72 | n/a | none | reached |
| pll_lock | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000229534 | 17 | 136 | 6324 | design_estimate | reached |
| pll_lock | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000760809 | 10 | 80 | 3720 | design_estimate | reached |
| pll_lock | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.00310806 | 6 | 48 | 2232 | design_estimate | reached |
| pll_lock | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.0142631 | 4 | 32 | 1488 | design_estimate | reached |
| robertson_scaled | library | BDF | compiled_float64 | 0.000244141 | 0.000140963 | 64 | 175 | n/a | none | reached |
| robertson_scaled | library | BDF | compiled_float64 | 0.000976562 | 0.000406376 | 29 | 88 | n/a | none | reached |
| robertson_scaled | library | BDF | compiled_float64 | 0.00390625 | 0.000406376 | 29 | 88 | n/a | none | reached |
| robertson_scaled | library | BDF | compiled_float64 | 0.015625 | 0.000406376 | 29 | 88 | n/a | none | reached |
| robertson_scaled | library | LSODA | compiled_float64 | 0.000244141 | 0.00019588 | 72 | 143 | n/a | none | reached |
| robertson_scaled | library | LSODA | compiled_float64 | 0.000976562 | 0.000617852 | 53 | 110 | n/a | none | reached |
| robertson_scaled | library | LSODA | compiled_float64 | 0.00390625 | 0.00213921 | 43 | 100 | n/a | none | reached |
| robertson_scaled | library | LSODA | compiled_float64 | 0.015625 | 0.00213921 | 43 | 100 | n/a | none | reached |
| robertson_scaled | library | Radau | compiled_float64 | 0.000244141 | 1.84472e-06 | 12 | 134 | n/a | none | reached |
| robertson_scaled | library | Radau | compiled_float64 | 0.000976562 | 1.84472e-06 | 12 | 134 | n/a | none | reached |
| robertson_scaled | library | Radau | compiled_float64 | 0.00390625 | 1.84472e-06 | 12 | 134 | n/a | none | reached |
| robertson_scaled | library | Radau | compiled_float64 | 0.015625 | 1.84472e-06 | 12 | 134 | n/a | none | reached |
| robertson_scaled | ours | sdirk2_analytic_jac | float64 | 0.000244141 | 0.000236606 | 229 | 1374 | 136942 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_analytic_jac | float64 | 0.000976562 | 0.000320808 | 228 | 1368 | 136344 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_analytic_jac | float64 | 0.00390625 | 0.000320808 | 228 | 1368 | 136344 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_analytic_jac | float64 | 0.015625 | 0.000320808 | 228 | 1368 | 136344 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000236605 | 229 | 2061 | 149308 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000320804 | 228 | 2052 | 148656 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.000320804 | 228 | 2052 | 148656 | design_estimate | reached |
| robertson_scaled | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.000320804 | 228 | 2052 | 148656 | design_estimate | reached |
| servo_load_step | library | BDF | compiled_float64 | 0.000244141 | 9.04229e-05 | 37 | 80 | n/a | none | reached |
| servo_load_step | library | BDF | compiled_float64 | 0.000976562 | 0.000277037 | 26 | 54 | n/a | none | reached |
| servo_load_step | library | BDF | compiled_float64 | 0.00390625 | 0.00276456 | 18 | 43 | n/a | none | reached |
| servo_load_step | library | BDF | compiled_float64 | 0.015625 | 0.00276456 | 18 | 43 | n/a | none | reached |
| servo_load_step | library | LSODA | compiled_float64 | 0.000244141 | 0.000110351 | 48 | 93 | n/a | none | reached |
| servo_load_step | library | LSODA | compiled_float64 | 0.000976562 | 0.00039952 | 41 | 83 | n/a | none | reached |
| servo_load_step | library | LSODA | compiled_float64 | 0.00390625 | 0.0029098 | 37 | 77 | n/a | none | reached |
| servo_load_step | library | LSODA | compiled_float64 | 0.015625 | 0.0029098 | 37 | 77 | n/a | none | reached |
| servo_load_step | library | Radau | compiled_float64 | 0.000244141 | 8.77343e-05 | 10 | 72 | n/a | none | reached |
| servo_load_step | library | Radau | compiled_float64 | 0.000976562 | 8.77343e-05 | 10 | 72 | n/a | none | reached |
| servo_load_step | library | Radau | compiled_float64 | 0.00390625 | 8.77343e-05 | 10 | 72 | n/a | none | reached |
| servo_load_step | library | Radau | compiled_float64 | 0.015625 | 8.77343e-05 | 10 | 72 | n/a | none | reached |
| servo_load_step | ours | sdirk2_analytic_jac | float64 | 0.000244141 | 0.000226304 | 16 | 96 | 5568 | design_estimate | reached |
| servo_load_step | ours | sdirk2_analytic_jac | float64 | 0.000976562 | 0.000925481 | 8 | 48 | 2784 | design_estimate | reached |
| servo_load_step | ours | sdirk2_analytic_jac | float64 | 0.00390625 | 0.00382851 | 4 | 24 | 1392 | design_estimate | reached |
| servo_load_step | ours | sdirk2_analytic_jac | float64 | 0.015625 | 0.014 | 2 | 12 | 696 | design_estimate | reached |
| servo_load_step | ours | sdirk2_fd_jac | float64 | 0.000244141 | 0.000226304 | 16 | 128 | 5952 | design_estimate | reached |
| servo_load_step | ours | sdirk2_fd_jac | float64 | 0.000976562 | 0.000925481 | 8 | 64 | 2976 | design_estimate | reached |
| servo_load_step | ours | sdirk2_fd_jac | float64 | 0.00390625 | 0.00382851 | 4 | 32 | 1488 | design_estimate | reached |
| servo_load_step | ours | sdirk2_fd_jac | float64 | 0.015625 | 0.014 | 2 | 16 | 744 | design_estimate | reached |
A further 20 rows are left out of the table because they never ran: sdirk2_analytic_jac on battery_2rc, bicycle_lateral, buck_converter, glucose_minimal, pll_lock, each recorded with the same reason: no analytic Jacobian is derived for this problem.
Three grades of cycle number sit in this document and each row says which it is: the pinned model cross-checked against the reference listing for Q15 explicit rows, a design estimate resting on a software-reciprocal guess for the implicit rows, and an upper bound with unpriced branches for the Q15 adaptive rows. Library rows carry no cycle number at all, because SciPy has no cost model and inventing one for it would be inventing the comparison.
The archive fixes a cycle budget and measures the error a method reaches. The lane fixes an accuracy target on the axis-T ladder and measures the modeled cycles a candidate needs to reach it, so neither number converts into the other. Every run is float64. The pinned order-condition checker never saw these coefficients, so none is order-verified.
Source: rk-work/implicit_archive/elites.json. float64 runs, cycles priced on design_estimate_div32; not order-verified.
Analytic entries need 2713 cycles at a21 1/8 to 2; finite difference entries need 2713 cycles at a21 1/8 to 5/4 and 2736 cycles at a21 11/8 to 2.
| rank | key | gamma | a21 | Jacobian | median cycles at the elite target |
|---|---|---|---|---|---|
| 1 | 41dd2666fd089999 | 7/8 | 1/8 | analytic | 2713 |
| 2 | e4c5840e0062c423 | 7/8 | 1/2 | analytic | 2713 |
| 3 | ddd24e7b3ecddb17 | 7/8 | 9/8 | analytic | 2713 |
| 4 | f535a65221c5c803 | 7/8 | 9/8 | finite difference | 2713 |
| 5 | c395c89ce9fa86f5 | 7/8 | 1/2 | finite difference | 2713 |
| 6 | 2f17abeb74a71f15 | 7/8 | 3/8 | analytic | 2713 |
| 7 | ecd697af33221456 | 7/8 | 3/2 | analytic | 2713 |
| 8 | c5fc63a59ad8243c | 7/8 | 3/4 | finite difference | 2713 |
| 9 | 22c8983196ef9dfc | 7/8 | 5/8 | analytic | 2713 |
| 10 | 807bdbcf1a3c544c | 7/8 | 5/8 | finite difference | 2713 |
| 11 | 53cd469f17220b3e | 7/8 | 5/4 | analytic | 2713 |
| 12 | a2569d009416261a | 7/8 | 1 | finite difference | 2713 |
| 13 | 88561348a06dc291 | 7/8 | 11/8 | analytic | 2713 |
| 14 | bbc73954b096cfad | 7/8 | 5/4 | finite difference | 2713 |
| 15 | e6c3f208afccf349 | 7/8 | 7/8 | analytic | 2713 |
| 16 | b17493037b3bb360 | 7/8 | 1/4 | analytic | 2713 |
| 17 | 985e256a9ff00861 | 7/8 | 13/8 | analytic | 2713 |
| 18 | 279b5561b2af2dd9 | 7/8 | 3/8 | finite difference | 2713 |
| 19 | 08ece01a5d113337 | 7/8 | 2 | analytic | 2713 |
| 20 | 04c47dad7bb1002b | 7/8 | 15/8 | analytic | 2713 |
| 21 | cc1e5a16fd157bad | 7/8 | 1/4 | finite difference | 2713 |
| 22 | 1f6af89598f650f3 | 7/8 | 1 | analytic | 2713 |
| 23 | 8577a41431f59811 | 7/8 | 7/4 | analytic | 2713 |
| 24 | 0d49be473923f968 | 7/8 | 1/8 | finite difference | 2713 |
| 25 | a86159f088f3fad7 | 7/8 | 3/4 | analytic | 2713 |
| 26 | 351b3f20dbe8b529 | 7/8 | 7/8 | finite difference | 2713 |
| 27 | 299b715ac28b5cdb | 7/8 | 11/8 | finite difference | 2736 |
| 28 | 5f803686bb5bcd7e | 7/8 | 7/4 | finite difference | 2736 |
| 29 | 3ca6c764e30c3f76 | 7/8 | 3/2 | finite difference | 2736 |
| 30 | ce3ec43a4dc6498c | 7/8 | 13/8 | finite difference | 2736 |
| 31 | 08793af994d3cd74 | 7/8 | 15/8 | finite difference | 2736 |
| 32 | d9d40926782b7ec7 | 7/8 | 2 | finite difference | 2736 |
Ranking rule, quoted from the document: lowest median cycles at the elite target across the problems where the status is reached; a candidate that reached the elite target nowhere ranks below every candidate that reached it somewhere; ties broken by the count of targets reached, highest leading, then by record_hash. No clock is read and no value is sampled, so the order is total and two runs over the same records agree.
Rank is the position the document gives under the rule above. Each row shows the candidate's shape, the median modeled cycles it needed at the elite target, how many problems it reached that target on, and the cost basis its cycles were priced on.
| rank | candidate | gamma | a21 | newton iterations | jacobian | A-stable | median cycles at target | problems at target | targets reached | cost basis |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 41dd2666fd089999 | 7/8 | 1/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 2 | e4c5840e0062c423 | 7/8 | 1/2 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 3 | ddd24e7b3ecddb17 | 7/8 | 9/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 4 | f535a65221c5c803 | 7/8 | 9/8 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 5 | c395c89ce9fa86f5 | 7/8 | 1/2 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 6 | 2f17abeb74a71f15 | 7/8 | 3/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 7 | ecd697af33221456 | 7/8 | 3/2 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 8 | c5fc63a59ad8243c | 7/8 | 3/4 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 9 | 22c8983196ef9dfc | 7/8 | 5/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 10 | 807bdbcf1a3c544c | 7/8 | 5/8 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 11 | 53cd469f17220b3e | 7/8 | 5/4 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 12 | a2569d009416261a | 7/8 | 1 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 13 | 88561348a06dc291 | 7/8 | 11/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 14 | bbc73954b096cfad | 7/8 | 5/4 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 15 | e6c3f208afccf349 | 7/8 | 7/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 16 | b17493037b3bb360 | 7/8 | 1/4 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 17 | 985e256a9ff00861 | 7/8 | 13/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 18 | 279b5561b2af2dd9 | 7/8 | 3/8 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 19 | 08ece01a5d113337 | 7/8 | 2 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 20 | 04c47dad7bb1002b | 7/8 | 15/8 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 21 | cc1e5a16fd157bad | 7/8 | 1/4 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 22 | 1f6af89598f650f3 | 7/8 | 1 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 23 | 8577a41431f59811 | 7/8 | 7/4 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 24 | 0d49be473923f968 | 7/8 | 1/8 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 25 | a86159f088f3fad7 | 7/8 | 3/4 | 2 | analytic | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 26 | 351b3f20dbe8b529 | 7/8 | 7/8 | 2 | finite_difference | true | 2713 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 27 | 299b715ac28b5cdb | 7/8 | 11/8 | 2 | finite_difference | true | 2736 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 28 | 5f803686bb5bcd7e | 7/8 | 7/4 | 2 | finite_difference | true | 2736 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 29 | 3ca6c764e30c3f76 | 7/8 | 3/2 | 2 | finite_difference | true | 2736 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 30 | ce3ec43a4dc6498c | 7/8 | 13/8 | 2 | finite_difference | true | 2736 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 31 | 08793af994d3cd74 | 7/8 | 15/8 | 2 | finite_difference | true | 2736 | 8 of 8 | 32 of 32 | design_estimate_div32 |
| 32 | d9d40926782b7ec7 | 7/8 | 2 | 2 | finite_difference | true | 2736 | 8 of 8 | 32 of 32 | design_estimate_div32 |
design_estimate_div32
One table per job and one row per point, from its newest ledger line. A point counts as measured only under the code hash that measured it, so an older line from earlier code is left out; that rule and the retry policy are in the measurement ledger. Plan, job catalogue and invariants: docs/SIDETRACK-AUTOMATION.md in the harness repository. Designs these feed: docs/EPOCH2-DESIGN.md and docs/EPOCH3-DESIGN.md.
25 points
Closes: EPOCH3-DESIGN: the L-stability window in J4 is 17 candidates wide and nothing pins that width, and the second free parameter of the 2-stage SDIRK is never varied
Arithmetic: tableau and stability algebra exact over Fractions; the measured order is float64
A = [[gamma, 0], [a21, gamma]], c is the row sum, b is solved from the two order-2 conditions, so b2 = (1/2 - gamma) / a21. The numerator of R(z) works out to (1, 1 - 2 gamma, 1/2 - 2 gamma + gamma^2) whatever a21 is, so |R(inf)|, A-stability and L-stability are functions of gamma alone. The a21 axis moves the order-3 residuals, stiff accuracy, the exactness of b, its CSD cost and the measured order.
| point | cycle | a21_values | a_stable_gammas | b_exact_pairs | gammas | l_stable_gammas | min_abs_r_inf | min_order3_residual | pairs | r_inf_a21_invariant | stiffly_accurate_pairs |
|---|---|---|---|---|---|---|---|---|---|---|---|
| s04_t04 | 2520 | 32 | 12 | 144 | 15 | 0 | 0.28 | 0.109375 | 480 | true | 0 |
| s04_t06 | 2520 | 65 | 12 | 125 | 15 | 0 | 0.28 | 0.105469 | 975 | true | 0 |
| s04_t08 | 2520 | 65 | 12 | 71 | 15 | 0 | 0.28 | 0.104492 | 975 | true | 0 |
| s04_t10 | 2520 | 65 | 12 | 65 | 15 | 0 | 0.28 | 0.116292 | 975 | true | 0 |
| s04_t12 | 2520 | 65 | 12 | 65 | 15 | 0 | 0.28 | 0.120687 | 975 | true | 0 |
| s06_t04 | 2520 | 32 | 48 | 596 | 63 | 0 | 0.0637119 | 0.108724 | 2016 | true | 0 |
| s06_t06 | 2520 | 65 | 48 | 425 | 63 | 0 | 0.0637119 | 0.104736 | 4095 | true | 0 |
| s06_t08 | 2520 | 65 | 48 | 113 | 63 | 0 | 0.0637119 | 0.104431 | 4095 | true | 0 |
| s06_t10 | 2520 | 65 | 48 | 71 | 63 | 0 | 0.0637119 | 0.112976 | 4095 | true | 0 |
| s06_t12 | 2520 | 65 | 48 | 65 | 63 | 0 | 0.0637119 | 0.114461 | 4095 | true | 0 |
| s08_t04 | 2520 | 32 | 44 | 615 | 65 | 0 | 0.00124444 | 0.107747 | 2080 | true | 0 |
| s08_t06 | 2520 | 65 | 44 | 444 | 65 | 0 | 0.00124444 | 0.104706 | 4225 | true | 0 |
| s08_t08 | 2520 | 65 | 44 | 75 | 65 | 0 | 0.00124444 | 0.104202 | 4225 | true | 0 |
| s08_t10 | 2520 | 65 | 44 | 11 | 65 | 0 | 0.00124444 | 0.112172 | 4225 | true | 0 |
| s08_t12 | 2520 | 65 | 44 | 1 | 65 | 0 | 0.00124444 | 0.11372 | 4225 | true | 0 |
| s10_t04 | 2520 | 32 | 65 | 614 | 65 | 0 | 0.00124444 | 0.107503 | 2080 | true | 0 |
| s10_t06 | 2520 | 65 | 65 | 443 | 65 | 0 | 0.00124444 | 0.104677 | 4225 | true | 0 |
| s10_t08 | 2520 | 65 | 65 | 97 | 65 | 0 | 0.00124444 | 0.104431 | 4225 | true | 0 |
| s10_t10 | 2520 | 65 | 65 | 20 | 65 | 0 | 0.00124444 | 0.114232 | 4225 | true | 0 |
| s10_t12 | 2520 | 65 | 65 | 5 | 65 | 0 | 0.00124444 | 0.118352 | 4225 | true | 0 |
| s12_t04 | 2520 | 32 | 65 | 611 | 65 | 0 | 0.00124444 | 0.108236 | 2080 | true | 0 |
| s12_t06 | 2520 | 65 | 65 | 441 | 65 | 0 | 0.00124444 | 0.106004 | 4225 | true | 0 |
| s12_t08 | 2520 | 65 | 65 | 94 | 65 | 0 | 0.00124444 | 0.105652 | 4225 | true | 0 |
| s12_t10 | 2520 | 65 | 65 | 16 | 65 | 0 | 0.00124444 | 0.118352 | 4225 | true | 0 |
| s12_t12 | 2520 | 65 | 65 | 2 | 65 | 0 | 0.00124444 | 0.123021 | 4225 | true | 0 |
9 points, from 18 ledger lines
Closes: EPOCH3-DESIGN: L-stability is to be checked numerically for the snapped gamma, and the NOT_L_STABLE threshold is a placeholder
Arithmetic: tableau and stability algebra exact over Fractions; the measured order is float64
gamma is snapped to a dyadic, a21 is fixed at 1 - gamma so that c2 = 1, and b is solved from the two order conditions. Stiff accuracy and order 2 can only hold together at the irrational gamma, so order is the one kept.
| point | cycle | candidates | a_stable | best_gamma | best_measured_order | best_r_at_infinity | l_stable |
|---|---|---|---|---|---|---|---|
| s04 | 2520 | 13 | 10 | 5/16 | 2.0062 | -0.28 | 0 |
| s05 | 2520 | 17 | 10 | 9/32 | 2.0066 | 0.209877 | 0 |
| s06 | 2520 | 17 | 12 | 19/64 | 2.0069 | -0.0637119 | 0 |
| s07 | 2520 | 17 | 14 | 19/64 | 2.0069 | -0.0637119 | 0 |
| s08 | 2520 | 17 | 17 | 75/256 | 2.0069 | -0.00124444 | 0 |
| s09 | 2520 | 17 | 17 | 75/256 | 2.0069 | -0.00124444 | 0 |
| s10 | 2520 | 17 | 17 | 75/256 | 2.0069 | -0.00124444 | 0 |
| s11 | 2520 | 17 | 17 | 75/256 | 2.0069 | -0.00124444 | 0 |
| s12 | 2520 | 17 | 17 | 75/256 | 2.0069 | -0.00124444 | 0 |
3 points
Closes: EPOCH3-DESIGN: the optional analytic Jacobian field is unpriced because every side-track SDIRK number is costed with the finite difference
Arithmetic: float64 only; no Q15 effects are included
the analytic Jacobian evaluation is excluded from the cycle count by the same convention costmodel.cycle_count uses for rhs evaluations, since both are application-supplied code. The reported saving is therefore the finite-difference assembly arithmetic plus the n extra rhs evaluations per step that assembly needs.
| point | cycle | best_error_analytic | best_error_finite_difference | cycles_analytic | cycles_finite_difference | cycles_saved | f_evals_saved | max_abs_diff | max_rel_diff |
|---|---|---|---|---|---|---|---|---|---|
| enzyme_qssa | 2540 | 2.76241e-06 | 2.76241e-06 | 348 | 372 | 24 | 2 | 9.69249e-09 | 1.69982e-10 |
| robertson_scaled | 2540 | n/a | n/a | 598 | 652 | 54 | 3 | 3.72529e-07 | 3.72529e-07 |
| servo_load_step | 2520 | 3.46048e-06 | 3.46048e-06 | 348 | 372 | 24 | 2 | 9.93411e-09 | 1.02767e-09 |
4 points, from 8 ledger lines
Closes: EPOCH3-DESIGN: three Newton iterations is the prototype's setting, not a ruling
Arithmetic: float64 only; no Q15 effects are included
EPOCH3-DESIGN.md: three iterations is the prototype's setting, not a ruling; the count enters the pinned evaluator config at the epoch boundary, so it should be chosen on data
| point | cycle | cycles_per_step_2_states | newton_iters | order_estimate | stiff_best_errors |
|---|---|---|---|---|---|
| it1 | 2520 | 236 | 1 | 2.0069 | enzyme_qssa 3.21588e-06; robertson_scaled 4.06153; servo_load_step 3.46048e-06 |
| it2 | 2520 | 304 | 2 | 2.0069 | enzyme_qssa 2.76307e-06; robertson_scaled n/a; servo_load_step 3.46048e-06 |
| it3 | 2520 | 372 | 3 | 2.0069 | enzyme_qssa 2.76241e-06; robertson_scaled n/a; servo_load_step 3.46048e-06 |
| it4 | 2520 | 440 | 4 | 2.0069 | enzyme_qssa 2.76241e-06; robertson_scaled n/a; servo_load_step 3.46048e-06 |
3 points, from 6 ledger lines
Closes: EPOCH3-DESIGN: the stiff case rests on rc_thermal and a synthetic two-rate system, while the suite now has three purpose-built stiff problems
Arithmetic: float64 only; no Q15 effects are included
cycle estimates exclude rhs and Jacobian evaluations, matching costmodel.cycle_count; SDIRK2 is priced with a finite-difference Jacobian because the validation problems supply no analytic one
| point | cycle | explicit_finishers | finishers | rk4_min_stable_n | sdirk2_best_error | sdirk2_min_stable_n |
|---|---|---|---|---|---|---|
| enzyme_qssa | 2520 | none | sdirk2 | n/a | 6.89217e-07 | 8 |
| robertson_scaled | 2520 | none | sdirk2 | n/a | 1.0198e-05 | 256 |
| servo_load_step | 2520 | none | sdirk2 | n/a | 8.63721e-07 | 8 |
3 points
Closes: EPOCH3-DESIGN: the stiff ladder in J3 stops at 256 steps while the harness scores at a matched cycle budget, where the cheap explicit methods take thousands of steps
Arithmetic: float64 only; no Q15 effects are included
cycle estimates exclude rhs and Jacobian evaluations, matching costmodel.cycle_count; SDIRK2 is priced with a finite-difference Jacobian because that is how every other side-track SDIRK number is costed. The budget is not read from any results file: an artifact that depended on host state would stop being reproducible.
| point | cycle | budget_cycles | error_at_budget | finishers_at_budget | sdirk2_error_at_budget | steps_at_budget |
|---|---|---|---|---|---|---|
| enzyme_qssa | 2520 | 65536 | euler 1.89045e-05; heun2 3.42268e-08; midpoint 2.09683e-08; rk4 3.24934e-11; sdirk2 1.45953e-06 | euler, heun2, midpoint, rk4, sdirk2 | 1.45953e-06 | euler 6553; heun2 2520; midpoint 2978; rk4 992; sdirk2 176 |
| robertson_scaled | 2520 | 65536 | euler 8.80946e-05; heun2 3.81289e-07; midpoint 3.16523e-07; rk4 n/a; sdirk2 n/a | euler, heun2, midpoint | n/a | euler 4369; heun2 1680; midpoint 1985; rk4 661; sdirk2 100 |
| servo_load_step | 2520 | 65536 | euler 2.3216e-05; heun2 3.67262e-08; midpoint 2.62928e-08; rk4 2.62516e-13; sdirk2 1.82872e-06 | euler, heun2, midpoint, rk4, sdirk2 | 1.82872e-06 | euler 6553; heun2 2520; midpoint 2978; rk4 992; sdirk2 176 |
| document | state | what it holds |
|---|---|---|
| rk-work/validation/axes.json | not written yet | the cycles-to-tolerance rows for this class, next to the explicit class on the same accuracy targets |
| rk-work/implicit_archive/YYYY-MM-DD.jsonl | written | the lane's per-day search log: one line per candidate the lane enumerated, ranked against nothing |
| rk-work/schedule/shares.json | written | the per-cycle lane log summarised per method class |
The traceability rule lists key_findings.json, validation/results.json, benchmark/results.json, the side-track ledger with its artifacts, and the two lane elites documents, adaptive_archive/elites.json and implicit_archive/elites.json. Each document named above sits outside that list and says so in its own schema, so this page names it and reports whether it exists without publishing a number from it. Only the owner can add one to the list.