Systems That Exhibit Combinatorial Explosion
Here are the main domains, grouped by field, with typical scale where it's meaningful.
Cryptography & Security
- Symmetric key search**: AES-128 has 2128 ≈ 3.4×1038 keys; AES-256 has ≈ 1.2×1077
- Asymmetric keys**: RSA-2048 factoring space, elliptic curve discrete log (~2256)
- Password and passphrase spaces**: grow as (alphabet size)length
- Hash preimage and collision search**: SHA-256 preimage ≈ 2256
- Side-channel and fault-injection key-recovery spaces**
- Lattice problems** (post-quantum): shortest vector search in high dimensions
Games & Puzzles
- Chess**: ~1044 positions, game tree ~10120 (Shannon number)
- Go (19×19)**: ~2×10170 legal positions
- Shogi, Xiangqi, Arimaa, Stratego**: 1071 to 10535 game-tree complexity
- Rubik's Cube**: 4.3×1019 states (larger cubes: 7×7×7 ≈ 10160)
- Sudoku/Latin squares**: ~6.7×1021 valid 9×9 grids
- Poker (Texas Hold'em)**: ~10161 decision points in no-limit
- Sliding-tile puzzles**: 24-puzzle ≈ 7.8×1024; 35-puzzle ≈ 1041
Optimization & Operations Research
- Traveling Salesman Problem**: (n-1)!/2 tours; 60 cities ≈ 1080
- Vehicle routing, job-shop and flow-shop scheduling**
- Knapsack, bin packing, set cover, graph coloring**
- Quadratic assignment problem** (facility layout)
- Timetabling** (exams, courses, airline crew pairing)
- Integer programming** (branching trees)
- Network design and facility location**
Logic, Computation & Verification
- Boolean satisfiability (SAT)**: 2n assignments for n variables
- Constraint satisfaction problems**
- Model checking**: state-space explosion in concurrent systems (interleavings of threads)
- Program analysis and symbolic execution**: path explosion from branching
- Software and hardware testing**: input-combination spaces, feature-interaction bugs
- Theorem proving and automated planning**: proof/plan search trees
- Circuit design verification** (Boolean function spaces: 2(2n) functions of n inputs)
Biology & Chemistry
- Protein folding** (Levinthal's paradox): ~10143 conformations for a 100-residue chain
- Protein sequence space**: 20n sequences (a 100-residue protein ≈ 10130)
- DNA/RNA sequences**: 4n (a human genome ≈ 4(3×109))
- Chemical space**: ~1060 drug-like molecules
- Molecular docking and conformer search**
- Gene regulatory networks and epistasis**: combinations of mutations
- Phylogenetic tree inference**: (2n-5)!! trees (50 species ≈ 1076)
- Multiple sequence alignment**
Physics & Materials Science
- Quantum many-body systems**: Hilbert space dimension 2n (exponential in particles)
- Ising and spin-glass models**: 2N configurations
- Crystal structure prediction** and alloy composition search
- Lattice gauge theory configurations**
- Statistical mechanics partition functions**
Artificial Intelligence & Machine Learning
- Neural architecture search** and hyperparameter combinations
- Feature subset selection**: 2n subsets
- Reinforcement learning state spaces** (e.g., Atari pixels, robotics)
- Bayesian network structure learning**: super-exponential in nodes
- Natural language**: possible sentences from a vocabulary V of length L is VL
- Image space**: 256(w×h×3) possible images
- Combinatorial action spaces** in strategy games (StarCraft ≈ 1026 actions per step)
Mathematics
- Graph enumeration**: number of labeled graphs on n vertices is 2(n(n-1)/2)
- Permutations and partitions**: n! and Bell numbers
- Ramsey numbers** and extremal combinatorics
- Latin squares, block designs, error-correcting code search**
- Boolean function and cellular automata rule spaces**
- Collatz-like and Busy Beaver searches**
Engineering & Industry
- Chip layout/placement and routing**: ~102500 placements for modern chips
- Configuration management** (product variants, option combinations)
- Supply chain and logistics planning**
- Power-grid unit commitment**
- Telecom network routing and spectrum allocation**
- Satellite constellation scheduling**
- Architectural and mechanical design spaces**
Finance & Economics
- Portfolio optimization** with discrete constraints
- Combinatorial auctions**: 2n bundles for n items
- Option and derivative path-dependency**
- Market microstructure and agent-based simulations**
Social & Everyday Systems
- Scheduling and seating arrangements**
- Matching problems** (with constraints)
- Genealogy and network analysis**
- Possible shuffles of a deck**: 52! ≈ 8×1067
Why These Explode
Three recurring patterns drive the growth:
- Exponential**: kn (keys, SAT, sequences)
- Factorial**: n! (permutations, TSP, orderings)
- Super-exponential**: nn or 2(2n) (Boolean functions, graph structures)
Common Resolution Strategies
Since brute force fails at these scales, the field relies on:
- Pruning**: branch-and-bound, alpha-beta, constraint propagation
- Heuristics/metaheuristics**: simulated annealing, genetic algorithms, tabu search
- Learned guidance**: Monte Carlo tree search, neural-guided search (AlphaGo, AlphaFold)
- Decomposition**: dynamic programming, divide-and-conquer, symmetry reduction
- Relaxation and approximation**: LP relaxation, approximation algorithms
- Probabilistic methods**: sampling, SAT solvers with clause learning (CDCL)
- Structure exploitation**: algebraic attacks, meet-in-the-middle, number-field sieve
- Quantum algorithms**: Grover (square-root speedup), Shor (factoring)
IMMORTALITY