num-bigint

num-bigint provides BigInt and BigUint, arbitrary-precision signed and unsigned integer types for Rust, implementing the standard arithmetic, bitwise, and comparison operator traits so they behave like Rust’s built-in integer types but without a fixed bit width. It’s part of the rust-num family of crates, integrating with num-traits/num-integer for shared numeric trait implementations, and optionally supports serde serialization, rand (multiple versions) for random big-integer generation, and no_std + alloc environments.

What You Get

• BigUint and BigInt types with full arithmetic, bitwise, and comparison operator trait implementations
• Conversions to/from Rust’s built-in integer types, strings (multiple radixes), and byte slices
• Modular exponentiation, root extraction, and other numeric algorithms exposed as trait methods
• Optional rand_0_9/rand_0_10 feature flags for generating cryptographically-relevant random big integers
• Optional serde serialization and no_std+alloc support for constrained environments

Common Use Cases

• Cryptographic primitives and protocol implementations requiring modular exponentiation over large integers (RSA, Diffie-Hellman)
• Arbitrary-precision arithmetic for financial, scientific, or combinatorics calculations that exceed 64/128-bit integer ranges
• Blockchain and distributed-ledger implementations that manipulate large numeric identifiers or balances
• Porting algorithms from languages with native bignum support (Python, JavaScript BigInt) into Rust without losing precision

Under The Hood

Architecture: the crate stores BigUint as a Vec of BigDigits (big_digit.rs) in a little-endian, base-2^32/2^64 representation, with biguint.rs implementing the unsigned arithmetic and bigint.rs layering a sign onto BigUint to implement signed BigInt, while bigrand.rs and macros.rs provide random-generation support and shared trait-impl macros respectively. Tech Stack: pure Rust (edition 2021, minimum supported Rust version 1.60), depending on the sibling num-integer and num-traits crates from the same rust-num organization for shared numeric trait definitions, with optional dependencies gated behind Cargo features (serde, rand_0_9, rand_0_10, quickcheck, arbitrary) so consumers only pull in what they use. Code Quality: the tests/ directory includes dedicated suites for signed/unsigned arithmetic, scalar operations, bitwise operations, modular exponentiation, root extraction, and a fuzzed.rs suite specifically covering fuzzer-discovered edge cases, reflecting the correctness bar expected of a widely-depended-on numeric primitive (nearly half a billion cumulative downloads per crates.io). API Design: BigUint/BigInt implement the same standard operator traits (Add, Sub, Mul, Shl, etc.) as Rust’s native integers, so idiomatic code (a + b, a.pow(n)) works without learning a bignum-specific method-call API, which keeps the learning curve close to zero for anyone already writing Rust.