CADICAL(1)

📦

version 2.2.1 · 10 versions

okay bestie, welcome to cadical(1) from User Commands 🫶✨

🌸 NAME 🌸

cadical — Simplified SAT solver 🧚💖

✨ DESCRIPTION ✨

The usage is:

usage: cadical [ <option> ... ] [ <input> [ <proof> ] ]

where <option> is one of the following common options:

-h — print only the common-options list instead 🫶
--help — print this complete list of all options 📜
--version — print version 🏷️
-n — do not print witness (same as --no-witness) 🙈
-v — increase verbosity (also see --verbose below) 🔊
-q — be quiet (same as --quiet) 🤫
-t <sec> — set wall clock time limit ⏰

Or <option> is one of the less common options:

-L<rounds> — run local search initially (default 0 rounds) 🔍
-O<level> — increase limits by 2^<level> or 10^<level> 🚀
-P<rounds> — initial preprocessing (default 0 rounds) 🧼

Note there is no separating space for the options above while the following options require a space after the option name:

-c <limit> — limit the number of conflicts (default unlimited) 💥
-d <limit> — limit the number of decisions (default unlimited) 🧠
-o <output> — write simplified CNF in DIMACS format to file 📝
-e <extend> — write reconstruction/extension stack to file 🧬
--force | -f — parsing broken DIMACS header and writing proofs 💪
--strict — strict parsing (no white space in header) 🧐
-r <sol> — read solution in competition output format to check consistency of learned clauses during testing and debugging 🔎
-w <sol> — write result including a potential witness solution in competition format to the given file 📤
--colors — force colored output 🌈
--no-colors — disable colored output to terminal 🚫
--no-witness — do not print witness (also see -n above) 🙈
--build — print build configuration 🏗️
--copyright — print copyright information ©️

There are pre-defined configurations of advanced internal options:

--default — set default advanced internal options 🎛️
--plain — disable all internal preprocessing options 🧼
--sat — set internal options to target satisfiable instances ✅
--unsat — set internal options to target unsatisfiable instances ❌

Or <option> is one of the following advanced internal options:

--arena=bool — allocate clauses in arena [true] 🏟️
--arenacompact=bool — keep clauses compact [true] 🧳
--arenasort=bool — sort clauses in arena [true] 🔤
--arenatype=1..3 — 1 = clause, 2 = var, 3 = queue [3] 🧩
--backbone=0..2 — binary clause backbone [1] 🦴
--backboneeffort=0..1e5 — binary effort in per mile [20] 💪
--backbonemaxrounds=0..1e5 — backbone rounds limit [1e3] 🔁
--backbonerounds=0..1e5 — backbone rounds limit [100] 🔁
--backbonethresh=0..1e9 — delay if ticks smaller thresh*clauses [5] ⏳
--binary=bool — use binary proof format [true] 💾
--block=bool — blocked clause elimination [false] 🚫
--blockmaxclslim=1..2e9 — maximum clause size [1e5] 📏
--blockminclslim=2..2e9 — minimum clause size [2] 📐
--blockocclim=1..2e9 — occurrence limit [1e2] 🔢
--bump=bool — bump variables [true] 📈
--bumpreason=bool — bump reason literals too [true] 🪄
--bumpreasondepth=1..3 — bump reason depth [1] 🕳️
--bumpreasonlimit=1..2e9 — bump reason limit [10] 🚧
--bumpreasonrate=1..2e9 — bump reason decision rate [100] ⚡
--check=bool — enable internal checking [false] 🔍
--checkassumptions=bool — check assumptions satisfied [true] ✅
--checkconstraint=bool — check constraint satisfied [true] ✅
--checkfailed=bool — check failed literals form core [true] 🧪
--checkfrozen=bool — check all frozen semantics [false] 🧊
--checkproof=0..3 — 1 = drat, 2 = lrat, 3 = both [3] 📝
--checkwitness=bool — check witness internally [true] 👀
--chrono=0..2 — chronological backtracking [1] ⏱️
--chronoalways=bool — force always chronological [false] 🕰️
--chronolevelim=0..2e9 — chronological level limit [1e2] 🚦
--chronoreusetrail=bool — reuse trail chronologically [true] ♻️
--compact=bool — compact internal variables [true] 🧳
--compactint=1..2e9 — compacting interval [2e3] ⏲️
--compactlim=0..1e3 — inactive limit per mille [1e2] 📉
--compactmin=1..2e9 — minimum inactive limit [1e2] 📉
--condition=bool — globally blocked clause elim [false] 🚫
--conditioneffort=1..1e5 — relative efficiency per mille [100] ⚙️
--conditionint=1..2e9 — initial conflict interval [1e4] ⏱️
--conditionmaxeff=0..2e9 — maximum condition efficiency [1e7] 📈
--conditionmaxrat=1..2e9 — maximum clause variable ratio [100] ⚖️
--conditionmineff=0..2e9 — minimum condition efficiency [0] 📉
--congruence=bool — congruence closure [true] 🔗
--congruenceand=bool — extract AND gates [true] 🚪
--congruenceandarity=2..5e7 — AND gate arity limit [1e6] 📏
--congruencebinaries=bool — extract binary and strengthen ternary clauses [true] 🧱
--congruenceite=bool — extract ITE gates [true] 🚪
--congruencexor=bool — extract XOR gates [true] 🚪
--congruencexorarity=2..31 — XOR gate arity limit [4] 📏
--congruencexorcounts=1..5e6 — XOR gate round [1] 🔁
--cover=bool — covered clause elimination [false] 🧹
--covereffort=1..1e5 — relative efficiency per mille [4] ⚙️
--covermaxclslim=1..2e9 — maximum clause size [1e5] 📏
--covermaxeff=0..2e9 — maximum cover efficiency [1e8] 📈
--coverminclslim=2..2e9 — minimum clause size [2] 📐
--covermineff=0..2e9 — minimum cover efficiency [0] 📉
--decompose=bool — decompose BIG in SCCs and ELS [true] 🧩
--decomposerounds=1..16 — number of decompose rounds [2] 🔁
--deduplicate=bool — remove duplicated binaries [true] 🧹
--deduplicateallinit=bool — remove duplicated clauses once before solving [false] 🧹
--eagersubsume=bool — subsume recently learned [true] 🍽️
--eagersubsumelim=1..1e3 — limit on subsumed candidates [20] 🚧
--elim=bool — bounded variable elimination [true] ✂️
--elimands=bool — find AND gates [true] 🔎
--elimbackward=bool — eager backward subsumption [true] ↩️
--elimboundmax=-1..2e6 — maximum elimination bound [16] 📈
--elimboundmin=-1..2e6 — minimum elimination bound [0] 📉
--elimclslim=2..2e9 — resolvent size limit [1e2] 📏
--elimdef=bool — mine definitions with kitten [false] 🐱
--elimdefcores=1..100 — number of unsat cores [1] 🧩
--elimdefticks=0..2e9 — kitten ticks limit [2e5] ⏱️
--elimeffort=1..1e5 — relative efficiency per mille [1e3] ⚙️
--elimequivs=bool — find equivalence gates [true] 🟰
--elimint=1..2e9 — elimination interval [2e3] ⏲️
--elimites=bool — find if-then-else gates [true] 🚪
--elimlimited=bool — limit resolutions [true] 🚧
--elimmaxeff=0..2e9 — maximum elimination efficiency [2e9] 📈
--elimmineff=0..2e9 — minimum elimination efficiency [1e7] 📉
--elimocclim=0..2e9 — occurrence limit [1e2] 🔢
--elimprod=0..1e4 — elim score product weight [1] ✖️
--elimrounds=1..512 — usual number of rounds [2] 🔁
--elimsubst=bool — elimination by substitution [true] 🔄
--elimsum=0..1e4 — elimination score sum weight [1] ➕
--elimxorlim=2..27 — maximum XOR size [5] 📏
--elimxors=bool — find XOR gates [true] 🔎
--emadecisions=1..2e9 — window decision rate [1e5] 🪟
--emagluefast=1..2e9 — window fast glue [33] 🪟
--emaglueslow=1..2e9 — window slow glue [1e5] 🪟
--emajump=1..2e9 — window back-jump level [1e5] 🪟
--emalevel=1..2e9 — window back-track level [1e5] 🪟
--emasize=1..2e9 — window learned clause size [1e5] 🪟
--ematrailfast=1..2e9 — window fast trail [1e2] 🪟
--ematrailslow=1..2e9 — window slow trail [1e5] 🪟
--exteagerreasons=bool — eagerly ask for all reasons (0: only when needed) [true] 🗣️
--exteagerrecalc=bool — after eagerly asking for reasons recalculate all levels (0: trust the external tool) [true] 🔄
--externallrat=bool — external lrat [false] 🌐
--factor=bool — bounded variable addition [false] ➕
--factorcandrounds=0..2e9 — candidates reduction rounds [2] 🔁
--factorcheck=0..2 — API checks that variables have been declared (1 = only with factor on, 2 = always) [1] ✅
--factordelay=0..12 — delay bounded variable addition between eliminations [4] ⏳
--factoreffort=0..1e6 — relative effort per mille [50] ⚙️
--factoriniticks=1..1e6 — initial effort in millions [300] 🚀
--factorsize=2..2e9 — clause size limit [5] 📏
--factorthresh=0..100 — delay if ticks smaller thresh*clauses [7] ⏳
--factorunbump=bool — extension variable with lowest importance (1: as in kissat) [true] 📉
--fastelim=bool — fast BVE during preprocessing [true] ⚡
--fastelimbound=1..1e3 — fast BVE bound during preprocessing [8] 📈
--fastelimclslim=2..2e9 — fast BVE resolvent size limit [1e2] 📏
--fastelimocclim=1..2e9 — fast BVE occurence limit during preprocessing [100] 🔢
--fastelimrounds=1..512 — number of fastelim rounds [4] 🔁
--flush=bool — flush redundant clauses [false] 🚽
--flushfactor=1..1e3 — interval increase [3] 📈
--flushint=1..2e9 — initial limit [1e5] 🚧
--forcephase=bool — always use initial phase [false] 📌
--frat=0..2 — 1 = frat(lrat), 2 = frat(drat) [0] 📝
--idrup=bool — incremental proof format [false] 💾
--ilb=0..2 — ILB (incremental lazy backtrack) (0: no, 1: assumptions only, 2: everything) [0] 🔄
--incdecay=0..4 — decay clauses when doing incremental clauses [1] 📉
--incdecayint=1..2e9 — decay interval when doing incremental clauses [1e6] ⏲️
--inprobeint=1..2e9 — inprobing interval [100] ⏱️
--inprobing=bool — enable probe inprocessing [true] 🔍
--inprocessing=bool — enable general inprocessing [true] ⚙️
--instantiate=bool — variable instantiation [false] 🧬
--instantiateclslim=2..2e9 — minimum clause size [3] 📐
--instantiateocclim=1..2e9 — maximum occurrence limit [1] 🔢
--instantiateonce=bool — instantiate each clause once [true] 1️⃣
--lidrup=bool — linear incremental proof format 💅 ([false])
--lrat=bool — use LRAT proof format 📜 ([false])
--lucky=bool — lucky phases 🍀 ([true])
--luckyassumptions=bool — lucky phases with assumptions 🍀✨ ([true])
--luckyearly=bool — lucky phases before preprocessing 🍀⬅️ ([true])
--luckylate=bool — lucky phases after preprocessing 🍀➡️ ([true])
--minimize=bool — minimize learned clauses ✂️ ([true])
--minimizedepth=0..1e3 — minimization depth 📉 ([1e3])
--minimizeticks=bool — increment ticks in minimization ⏱️ ([true])
--otfs=bool — on-the-fly self subsumption 🪄 ([true])
--phase=bool — initial phase 🌗 ([true])
--preprocessinit=0..2e9 — initial preprocessing base limit 🏗️ ([2e6])
--preprocesslight=bool — lightweight preprocessing 🪶 ([true])
--probe=bool — failed literal probing 🔍 ([true])
--probeeffort=1..1e5 — relative efficiency per mille ⚡ ([8])
--probehbr=bool — learn hyper binary clauses 🧠 ([true])
--probethresh=0..100 — delay if ticks smaller thresh*clauses 🐢 ([0])
--profile=0..4 — profiling level 📊 ([2])
--quiet=bool — disable all messages 🤫 ([false])
--radixsortlim=0..2e9 — radix sort limit 🔢 ([32])
--randec=bool — random decisions 🎲 ([false])
--randecfocused=bool — random decisions in focused mode 🎲🔦 ([true])
--randecinit=2..2e9 — inital random decision interval 🎲⏳ ([1e3]) — yes bestie, "inital", that's the original's spelling, we don't touch it 💀
--randecint=0..2e9 — random conflict length 🎲📏 ([500])
--randeclength=1..1e9 — length random decisions phases 🎲🌊 ([10])
--randecstable=bool — random decisions in stable mode 🎲🧘 ([false])
--realtime=bool — real instead of process time ⌚ ([false])
--recomputetier=bool — recompute tiers 🧮 ([true])
--reduce=bool — reduce useless clauses 🗑️ ([true])
--reduceinit=1..1e6 — initial interval 🕐 ([300])
--reduceint=2..1e6 — reduce interval 🔁 ([25])
--reduceopt=0..2 — 0=prct, 1=sqrt, 2=max 🎛️ ([1])
--reducetarget=10..1e2 — reduce fraction in percent 💯 ([75])
--reducetier1glue=1..2e9 — glue of kept learned clauses 🍯 ([2])
--reducetier2glue=1..2e9 — glue of tier two clauses 🍯🍯 ([6])
--reluctant=bool — stable reluctant doubling restarts 😤 ([true])
--reluctantint=0..2e9 — reluctant doubling period ⏲️ ([1024])
--reluctantmax=0..2e9 — maximum reluctant doubling period ⏲️⬆️ ([1048576])
--rephase=0..2 — enable resetting phase (0=no, 1=always, 2=stable-only) 🔄 ([1])
--rephaseint=1..2e9 — rephase interval 🔄⏳ ([1e3])
--report=bool — enable reporting 📢 ([false])
--reportall=bool — report even if not successful 📢🙃 ([false])
--reportsolve=bool — use solving not process time 📢⌛ ([false])
--restart=bool — enable restarts 🔃 ([true])
--restartint=1..2e9 — restart interval 🔃⏱️ ([2])
--restartmarginfocused=0..25 — focused slow fast margin in percent 🎯 ([10])
--restartmarginstable=0..25 — stable slow fast margin in percent 🧘🎯 ([25])
--restartreusetrail=bool — enable trail reuse 🛤️ ([true])
--restoreall=0..2 — restore all clauses (2=really) 😳 ([0]) — yes the "really" is doing a LOT of heavy lifting there
--restoreflush=bool — remove satisfied clauses 🚿 ([false])
--reverse=bool — reverse variable ordering 🔀 ([false])
--score=bool — use EVSIDS scores 🏆 ([true])
--scorefactor=500..1e3 — score factor per mille 🏆⚖️ ([950])
--seed=0..2e9 — random seed 🌱 ([0])
--shrink=0..3 — shrink conflict clause (1=binary-only, 2=minimize-on-pulling, 3=full) 🫧 ([3])
--shrinkreap=bool — use a reap for shrinking 🌾 ([true])
--shuffle=bool — shuffle variables 🔀🎴 ([false])
--shufflequeue=bool — shuffle variable queue 🔀📋 ([true])
--shufflerandom=bool — not reverse but random 🙃 ([false])
--shufflescores=bool — shuffle variable scores 🔀🏆 ([true])
--stabilize=bool — enable stabilizing phases 🧘 ([true])
--stabilizeinit=1..2e9 — stabilizing interval 🧘⏳ ([1e3])
--stabilizeonly=bool — only stabilizing phases 🧘🙅 ([false])
--stats=bool — print all statistics at the end of the run 📊 ([false])
--stubbornIOfocused=bool — force phases to I/O in focused mode every once in a while (requires rephase==2) 🙈 ([false])
--subsume=bool — enable clause subsumption 🫂 ([true])
--subsumebinlim=0..2e9 — watch list length limit 👀 ([1e4])
--subsumeclslim=0..2e9 — clause length limit 📏 ([1e2])
--subsumeeffort=1..1e5 — relative efficiency per mille ⚡ ([1e3])
--subsumelimited=bool — limit subsumption checks 🚧 ([true])
--subsumemaxeff=0..2e9 — maximum subsuming efficiency 📈 ([1e8])
--subsumemineff=0..2e9 — minimum subsuming efficiency 📉 ([0])
--subsumeocclim=0..2e9 — watch list length limit 👀 ([1e2])
--subsumestr=bool — subsume strengthen 💪 ([true])
--sweep=bool — enable SAT sweeping 🧹 ([true])
--sweepclauses=0..2e9 — environment clauses 🌐 ([1024])
--sweepcomplete=bool — run SAT sweeping to completion 🧹✅ ([false])
--sweepcountbinary=bool — count binaries to environment 🌐🔢 ([true])
--sweepdepth=0..2e9 — environment depth 🌐🕳️ ([2])
--sweepeffort=0..1e4 — relative effort in ticks per mille ⚡ ([1e2])
--sweepfliprounds=0..2e9 — flipping rounds 🔄 ([1])
--sweepmaxclauses=2..2e9 — maximum environment clauses 🌐⬆️ ([3e5])
--sweepmaxdepth=1..2e9 — maximum environment depth 🌐⬆️🕳️ ([3])
--sweepmaxvars=2..2e9 — maximum environment variables 🌐🔤 ([8192])
--sweeprand=bool — randomize sweeping environment 🎲🌐 ([false])
--sweepthresh=0..100 — delay if ticks smaller thresh*clauses 🐢 ([5])
--sweepvars=0..2e9 — environment variables 🌐🔤 ([256])
--target=0..2 — target phases (1=stable only) 🎯 ([1])
--terminateint=0..1e4 — termination check interval 🛑⏱️ ([10])
--ternary=bool — hyper ternary resolution 3️⃣ ([true])
--ternaryeffort=1..1e5 — relative efficiency per mille ⚡ ([8])
--ternarymaxadd=0..1e4 — max clauses added in percent ➕💯 ([1e3])
--ternaryocclim=1..2e9 — ternary occurrence limit 3️⃣👀 ([1e2])
--ternaryrounds=1..16 — maximum ternary rounds 3️⃣🔁 ([2])
--ternarythresh=0..100 — delay if ticks smaller thresh*clauses 🐢 ([6])
--tier1limit=0..100 — limit of tier1 usage in percentage 💯 ([50])
--tier1minglue=0..100 — lowest tier1 limit 🍯⬇️ ([0])
--tier2limit=0..100 — limit of tier2 usage in percentage 💯 ([90])
--tier2minglue=0..100 — lowest tier2 limit 🍯⬇️ ([0])
--transred=bool — transitive reduction of BIG 🫠 ([true])
--transredeffort=1..1e5 — relative efficiency per mille ⚡ ([1e2])
--transredmaxeff=0..2e9 — maximum efficiency 📈 ([1e8])
--transredmineff=0..2e9 — minimum efficiency 📉 ([0])
--verbose=0..3 — more verbose messages 🗣️ ([0])
--veripb=0..4 — odd=check-deletions, >2 drat 🧾 ([0])
--vivify=bool — vivification ✨ ([true])
--vivifycalctier=bool — recalculate tier limits 🧮 ([false])
--vivifydemote=bool — demote irredundant or delete directly ⬇️🗑️ ([false])
--vivifyeffort=0..1e5 — overall efficiency per mille ⚡ ([50])
--vivifyflush=bool — flush subsumed before vivification rounds 🚿 ([true])
--vivifyinst=bool — instantiate last literal when vivify ✨🔤 ([true])
--vivifyirred=bool — vivification of irredundant clauses ✨ ([true])
--vivifyirredeff=1..100 — irredundant efficiency per mille ⚡ ([3])
--vivifyonce=0..2 — vivify once: 1=red, 2=red+irr 1️⃣ ([0])
--vivifyretry=0..5 — re-vivify clause if vivify was successful 🔁✨ ([0])
--vivifyschedmax=10..2e9 — maximum schedule size 📅⬆️ ([5e3])
--vivifythresh=0..100 — delay if ticks smaller thresh*clauses 🐢 ([20])
--vivifytier1=bool — vivification tier1 ✨1️⃣ ([true])
--vivifytier1eff=0..100 — relative tier1 effort ⚡1️⃣ ([4])
--vivifytier2=bool — vivification tier2 ✨2️⃣ ([true])
--vivifytier2eff=1..100 — relative tier2 effort ⚡2️⃣ ([2])
--vivifytier3=bool — vivification tier3 ✨3️⃣ ([true])
--vivifytier3eff=1..100 — relative tier3 effort ⚡3️⃣ ([1])
--walk=bool — enable random walks 🚶 ([true])
--walkeffort=1..1e5 — relative efficiency per mille ⚡ ([80])
--walkfullocc=bool — use Kissat's full occurrences instead of the single watched 👀 ([false])
--walkmaxeff=0..2e9 — maximum efficiency (in 1e3 ticks) 📈 ([1e7])
--walkmineff=0..1e7 — minimum efficiency 📉 ([0])
--walknonstable=bool — walk in non-stabilizing phase 🚶🚫 ([true])
--walkredundant=bool — walk redundant clauses too 🚶🔁 ([false])
--warmup=bool — warmup before walk using propagation 🔥🚶 ([true])

okay so 💖 the internal options have their default value printed in brackets right after their description, like you just saw. And there's cute shorthand:

--<name> is literally the same as --<name>=1 ✨
--no-<name> is literally the same as --<name>=0 ✨

and you can also use true instead of 1, false instead of 0, and numbers with positive exponent like 1e3 instead of 1000. math girlie behavior, we love it 🧚

Alternatively you can drop option values in the header of the DIMACS file, like this:

c --elim=false

or you can set them through environment variables, like:

CADICAL_ELIM=false

Priority check bestie, do NOT mix this up: embedded options in the DIMACS file have the highest priority, then command line options, and then values specified through environment variables. That's the order. Periodt. 👑

The input gets read from <input> and it's assumed to be in DIMACS format 📄. Incremental p inccnf files are supported too, with cubes at the end 🧊.

If <proof> is given then a DRAT proof gets written to that file 🧾✨.

If <input> is missing then the solver reads from <stdin> 🌸, also if - is used as input path name <input>. Similarly,

For incremental files each cube is solved in turn 🧊🧊🧊. The solver stops at the first satisfied cube if there is one and uses that one for the witness to print 🏆. Conflict and decision limits are applied to each individual cube solving call, while -P, -L and -t remain global 🌍. Only if all cubes were unsatisfiable does the solver print the standard unsatisfiable solution line:

s UNSATISFIABLE

By default the proof is stored in the binary DRAT format 💾 unless the option --no-binary is specified, or the proof is written to <stdout> and <stdout> is connected to a terminal 🖥️. (because dumping raw binary at your terminal would be a whole crime scene 😭)

Okay so compression! 💅 If you hand over an input file explicitly and it's got a .gz, .bz2, .xz or .7z suffix, we're gonna assume it's compressed, bestie. Same energy for the output file — suffix says compressed, we treat it as compressed. ✨

And like… obviously you need the actual tools for that, because we're not doing magic here. Depending on the format you gotta have gzip, bzip, xz or 7z installed on your system, fr fr. No utility, no zip-unzip, it's giving missing dependency. 😭

But here's the safety net 🫶 — the solver actually peeks at the file type signatures, and if the signature does NOT match, it just falls back to reading the file non-compressed. So a sneaky little mislabeled file isn't gonna ruin your whole day. Iconic behaviour honestly. 👑

And that's a wrap on CADICAL(1) babes!! 🌸 This one's cadical 2.2.1, dated April 2026 — go forth and solve, my gorgeous little SAT-solver-using main characters. 🦋💖 periodt. ⭐

📦 where this page comes from

fedora 44: cadical 1:2.2.1-4.fc44 · MIT
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-10-13: a language model's rewrite of the original, under its terms