Repository files navigation

Seeded random number generators for TypeScript.

npm install ts-seedrandom

Each generator includes the following methods:

• quick - Default method used. Provides 32 bits of randomness in a float. Can either be called by calling generator instance directly (ex.generator()) or by name (ex.generator.quick()).
• double - Provides 56 bits of randomness.
• int32 - Providers a 32 bit (signed) integer.
• state - Provides internal generator state. Used for saving and restoring states.

import{prngAlea}from'ts-seedrandom';constaleaGenerator=prngAlea('seed');constfirstValue=aleaGenerator();constsecondValue=aleaGenerator();

You also have the option of saving and restoring state of your generator.

import{prngAlea}from'ts-seedrandom';constaleaGenerator=prngAlea('seed');constfirstValue=aleaGenerator();// Return internal generator state, which you can use in other generator instancesconststate=aleaGenerator.state();// This generator starts from the same state as first generator, but runs independentlyconstsecondAleaGenerator=prngAlea('seed',state);

The following PRNG algorithms are available:

1. prngAlea: Alea algorithm
2. prngArc4: ARC4 algorithm
3. prngTychei: Tyche-i algorithm
4. prngXor128: XorShift128 algorithm
5. prngXor4096: XorShift4096 algorithm
6. prngXorshift7: XorShift7 algorithm
7. prngXorwow: Xorwow algorithm
8. prngMulberry32: Mulberry 32 algorithm
9. prngXoshiro128Plus: Xoshiro128+ algorithm
10. prngXoshiro128PlusPlus: Xoshiro128++ algorithm
11. prngSplitMix64: SplitMix64 algorithm
12. prngSplitMix32: SplitMix32 algorithm
13. prngSfc32: SFC32 algorithm
14. prngJsf32: JSF32 algorithm
15. prngXoroshiro128ss: Xoshiro128** algorithm
16. prngXoroshiro128+: Xoroshiro128plus algorithm
17. prngParkMiller: Lehrer (Park-Miller) algorithm

You can import and use any of these algorithms in the same way as demonstrated in the usage examples above.

• What I computed:

  • Mops/s (million iters/sec) = 1000 / (time_ms)
  • per-iteration (ns) ≈ time_ms
  • Slower vs fastest (%) = (1 - (current_speed / fastest_speed)) * 100.

• Test details / machine:Lenovo Legion 5 Pro 16ACH6H (Ryzen 7 5800H — 8 cores / 16 threads, base ≈ 3.2 GHz, turbo ≈ 4.4 GHz, DDR4-3200 memory); Node.js v24.10.0.

• Why numbers vary: JIT warm-up, Node version, single vs multi-thread scheduling, background load, and micro-optimizations in each PRNG implementation all affect timings. Use these as a relative ranking on this machine, not an absolute cross-platform benchmark.

• You can replicate this exact table by runningnpm run compare:performance