Skip to content

indutny/bn.js

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

bn.js

BigNum in pure javascript

Build Status

Install

npm install --save bn.js

Usage

const BN = require('bn.js');

var a = new BN('dead', 16);
var b = new BN('101010', 2);

var res = a.add(b);
console.log(res.toString(10));  // 57047

Note: decimals are not supported in this library.

Sponsors

Scout APM My Open Source work is supported by Scout APM and other sponsors.

Notation

Prefixes

There are several prefixes to instructions that affect the way they work. Here is the list of them in the order of appearance in the function name:

  • i - perform operation in-place, storing the result in the host object (on which the method was invoked). Might be used to avoid number allocation costs
  • u - unsigned, ignore the sign of operands when performing operation, or always return positive value. Second case applies to reduction operations like mod(). In such cases if the result will be negative - modulo will be added to the result to make it positive

Postfixes

  • n - the argument of the function must be a plain JavaScript Number. Decimals are not supported. The number passed must be smaller than 0x4000000 (67_108_864). Otherwise, an error is thrown.
  • rn - both argument and return value of the function are plain JavaScript Numbers. Decimals are not supported.

Examples

  • a.iadd(b) - perform addition on a and b, storing the result in a
  • a.umod(b) - reduce a modulo b, returning positive value
  • a.iushln(13) - shift bits of a left by 13

Instructions

Prefixes/postfixes are put in parens at the end of the line. endian - could be either le (little-endian) or be (big-endian).

Utilities

  • a.clone() - clone number
  • a.toString(base, length) - convert to base-string and pad with zeroes
  • a.toNumber() - convert to Javascript Number (limited to 53 bits)
  • a.toJSON() - convert to JSON compatible hex string (alias of toString(16))
  • a.toArray(endian, length) - convert to byte Array, and optionally zero pad to length, throwing if already exceeding
  • a.toArrayLike(type, endian, length) - convert to an instance of type, which must behave like an Array
  • a.toBuffer(endian, length) - convert to Node.js Buffer (if available). length in bytes. For compatibility with browserify and similar tools, use this instead: a.toArrayLike(Buffer, endian, length)
  • a.bitLength() - get number of bits occupied
  • a.zeroBits() - return number of less-significant consequent zero bits (example: 1010000 has 4 zero bits)
  • a.byteLength() - return number of bytes occupied
  • a.isNeg() - true if the number is negative
  • a.isEven() - no comments
  • a.isOdd() - no comments
  • a.isZero() - no comments
  • a.cmp(b) - compare numbers and return -1 (a < b), 0 (a == b), or 1 (a > b) depending on the comparison result (ucmp, cmpn)
  • a.lt(b) - a less than b (n)
  • a.lte(b) - a less than or equals b (n)
  • a.gt(b) - a greater than b (n)
  • a.gte(b) - a greater than or equals b (n)
  • a.eq(b) - a equals b (n)
  • a.toTwos(width) - convert to two's complement representation, where width is bit width
  • a.fromTwos(width) - convert from two's complement representation, where width is the bit width
  • BN.isBN(object) - returns true if the supplied object is a BN.js instance
  • BN.max(a, b) - return a if a bigger than b
  • BN.min(a, b) - return a if a less than b

Arithmetics

  • a.neg() - negate sign (i)
  • a.abs() - absolute value (i)
  • a.add(b) - addition (i, n, in)
  • a.sub(b) - subtraction (i, n, in)
  • a.mul(b) - multiply (i, n, in)
  • a.sqr() - square (i)
  • a.pow(b) - raise a to the power of b
  • a.div(b) - divide (divn, idivn)
  • a.mod(b) - reduct (u, n) (but no umodn)
  • a.divmod(b) - quotient and modulus obtained by dividing
  • a.divRound(b) - rounded division

Bit operations

  • a.or(b) - or (i, u, iu)
  • a.and(b) - and (i, u, iu, andln) (NOTE: andln is going to be replaced with andn in future)
  • a.xor(b) - xor (i, u, iu)
  • a.setn(b, value) - set specified bit to value
  • a.shln(b) - shift left (i, u, iu)
  • a.shrn(b) - shift right (i, u, iu)
  • a.testn(b) - test if specified bit is set
  • a.maskn(b) - clear bits with indexes higher or equal to b (i)
  • a.bincn(b) - add 1 << b to the number
  • a.notn(w) - not (for the width specified by w) (i)

Reduction

  • a.gcd(b) - GCD
  • a.egcd(b) - Extended GCD results ({ a: ..., b: ..., gcd: ... })
  • a.invm(b) - inverse a modulo b

Fast reduction

When doing lots of reductions using the same modulo, it might be beneficial to use some tricks: like Montgomery multiplication, or using special algorithm for Mersenne Prime.

Reduction context

To enable this trick one should create a reduction context:

var red = BN.red(num);

where num is just a BN instance.

Or:

var red = BN.red(primeName);

Where primeName is either of these Mersenne Primes:

  • 'k256'
  • 'p224'
  • 'p192'
  • 'p25519'

Or:

var red = BN.mont(num);

To reduce numbers with Montgomery trick. .mont() is generally faster than .red(num), but slower than BN.red(primeName).

Converting numbers

Before performing anything in reduction context - numbers should be converted to it. Usually, this means that one should:

  • Convert inputs to reducted ones
  • Operate on them in reduction context
  • Convert outputs back from the reduction context

Here is how one may convert numbers to red:

var redA = a.toRed(red);

Where red is a reduction context created using instructions above

Here is how to convert them back:

var a = redA.fromRed();

Red instructions

Most of the instructions from the very start of this readme have their counterparts in red context:

  • a.redAdd(b), a.redIAdd(b)
  • a.redSub(b), a.redISub(b)
  • a.redShl(num)
  • a.redMul(b), a.redIMul(b)
  • a.redSqr(), a.redISqr()
  • a.redSqrt() - square root modulo reduction context's prime
  • a.redInvm() - modular inverse of the number
  • a.redNeg()
  • a.redPow(b) - modular exponentiation

Number Size

Optimized for elliptic curves that work with 256-bit numbers. There is no limitation on the size of the numbers.

LICENSE

This software is licensed under the MIT License.