Modulo operation

Variants of the definition

In mathematics, the result of the modulo operation is an equivalence class, and any member of the class may be chosen as representative; however, the usual representative is the least positive residue, the smallest non-negative integer that belongs to that class (i.e., the remainder of the Euclidean division).^[3] However, other conventions are possible. Computers and calculators have various ways of storing and representing numbers; thus their definition of the modulo operation depends on the programming language or the underlying hardware.

In nearly all computing systems, the quotient q and the remainder r of a divided by n satisfy the following conditions:

${\displaystyle {\begin{aligned}q\,&\in \mathbb {Z} \\a\,&=nq+r\\|r|\,&<|n|\end{aligned}}}$

(1)

However, this still leaves a sign ambiguity if the remainder is nonzero: two possible choices for the remainder occur, one negative and the other positive, and two possible choices for the quotient occur. In number theory, the positive remainder is always chosen, but in computing, programming languages choose depending on the language and the signs of a or n.^[a] Standard Pascal and ALGOL 68, for example, give a positive remainder (or 0) even for negative divisors, and some programming languages, such as C90, leave it to the implementation when either of n or a is negative (see the table under § In programming languages for details). a modulo 0 is undefined in most systems, although some do define it as a.

As described by Leijen,

Boute argues that Euclidean division is superior to the other ones in terms of regularity and useful mathematical properties, although floored division, promoted by Knuth, is also a good definition. Despite its widespread use, truncated division is shown to be inferior to the other definitions.

— Daan Leijen, Division and Modulus for Computer Scientists^[6]

However, truncated division satisfies the identity ${\displaystyle (-a)/b=-(a/b)=a/(-b)}$.^[7]

Notation

Some calculators have a mod() function button, and many programming languages have a similar function, expressed as mod(a, n), for example. Some also support expressions that use "%", "mod", or "Mod" as a modulo or remainder operator, such as a % n or a mod n.

For environments lacking a similar function, any of the three definitions above can be used.

Common pitfalls

When the result of a modulo operation has the sign of the dividend (truncating definition), it can lead to surprising mistakes.

For example, to test if an integer is odd, one might be inclined to test if the remainder by 2 is equal to 1:

bool is_odd(int n) {
    return n % 2 == 1;
}

But in a language where modulo has the sign of the dividend, that is incorrect, because when n (the dividend) is negative and odd, n mod 2 returns −1, and the function returns false.

One correct alternative is to test that the remainder is not 0 (because remainder 0 is the same regardless of the signs):

bool is_odd(int n) {
    return n % 2 != 0;
}

Another alternative is to use the fact that for any odd number, the remainder may be either 1 or −1:

bool is_odd(int n) {
    return n % 2 == 1 || n % 2 == -1;
}

Performance issues

Modulo operations might be implemented such that a division with a remainder is calculated each time. For special cases, on some hardware, faster alternatives exist. For example, the modulo of powers of 2 can alternatively be expressed as a bitwise AND operation (assuming x is a positive integer, or using a non-truncating definition):

x % 2n == x & (2n - 1)

Examples:

x % 2 == x & 1

x % 4 == x & 3

x % 8 == x & 7

In devices and software that implement bitwise operations more efficiently than modulo, these alternative forms can result in faster calculations.^[8]

Compiler optimizations may recognize expressions of the form expression % constant where constant is a power of two and automatically implement them as expression & (constant-1), allowing the programmer to write clearer code without compromising performance. This simple optimization is not possible for languages in which the result of the modulo operation has the sign of the dividend (including C), unless the dividend is of an unsigned integer type. This is because, if the dividend is negative, the modulo will be negative, whereas expression & (constant-1) will always be positive. For these languages, the equivalence x % 2n == x < 0 ? x | ~(2n - 1) : x & (2n - 1) has to be used instead, expressed using bitwise OR, NOT and AND operations.

Properties (identities)

Some modulo operations can be factored or expanded similarly to other mathematical operations. This may be useful in cryptography proofs, such as the Diffie–Hellman key exchange.

• Identity:
  • (a mod n) mod n = a mod n.
  • n^x mod n = 0 for all positive integer values of x.
  • If p is a prime number which is not a divisor of b, then ab^p−1 mod p = a mod p, due to Fermat's little theorem.
• Inverse:
  • [(−a mod n) + (a mod n)] mod n = 0.
  • b⁻¹ mod n denotes the modular multiplicative inverse, which is defined if and only if b and n are relatively prime, which is the case when the left hand side is defined: [(b⁻¹ mod n)(b mod n)] mod n = 1.
• Distributive:
  • (a + b) mod n = [(a mod n) + (b mod n)] mod n.
  • ab mod n = [(a mod n)(b mod n)] mod n.
• Division (definition): a/b mod n = [(a mod n)(b⁻¹ mod n)] mod n, when the right hand side is defined (that is when b and n are coprime). Undefined otherwise.
• Inverse multiplication: [(ab mod n)(b⁻¹ mod n)] mod n = a mod n.

In programming languages

In addition, many computer systems provide a divmod functionality, which produces the quotient and the remainder at the same time. Examples include the x86 architecture's IDIV instruction, the C programming language's div() function, and Python's divmod() function.

Generalizations

Modulo with offset

Sometimes it is useful for the result of a modulo n to lie not between 0 and n−1, but between some number d and d+n−1. In that case, d is called an offset. There does not seem to be a standard notation for this operation, so let us tentatively use a mod_dn. We thus have the following definition:^[27]x = a mod_dn just in case d ≤ x ≤ d+n−1 and x mod n = a mod n. Clearly, the usual modulo operation corresponds to zero offset: a mod n = a mod₀n. The operation of modulo with offset is related to the floor function as follows:

${\displaystyle a\operatorname {mod} _{d}n=a-n\left\lfloor {\frac {a-d}{n}}\right\rfloor }$.

(This is easy to see. Let ${\textstyle x=a-n\left\lfloor {\frac {a-d}{n}}\right\rfloor }$. We first show that x mod n = a mod n. It is in general true that (a+bn) mod n = a mod n for all integers b; thus, this is true also in the particular case when ${\textstyle b=-\left\lfloor {\frac {a-d}{n}}\right\rfloor }$; but that means that ${\textstyle x{\bmod {n}}=\left(a-n\left\lfloor {\frac {a-d}{n}}\right\rfloor \right){\bmod {n}}=a{\bmod {n}}}$, which is what we wanted to prove. It remains to be shown that d ≤ x ≤ d+n−1. Let k and r be the integers such that a − d = kn + r with 0 ≤ r ≤ n−1 (see Euclidean division). Then ${\textstyle \left\lfloor {\frac {a-d}{n}}\right\rfloor =k}$, thus ${\textstyle x=a-n\left\lfloor {\frac {a-d}{n}}\right\rfloor =a-nk=d+r}$. Now take 0 ≤ r ≤ n−1 and add d to both sides, obtaining d ≤ d + r ≤ d+n−1. But we've seen that x = d + r, so we are done. □)

The modulo with offset a mod_dn is implemented in Mathematica as^[27]Mod[a, n, d].

Implementing other modulo definitions using truncation

Despite the mathematic elegance of Knuth's floored division and Euclidean division, it is generally much more common to find a truncated division-based modulo in programming languages. Leijen provides the following algorithms for calculating the two divisions given a truncated integer division:^[6]

/* Euclidean and Floored divmod, in the style of C's ldiv() */
typedef struct {
  /* This structure is part of the C stdlib.h, but is reproduced here for clarity */
  long int quot;
  long int rem;
} ldiv_t;

/* Euclidean division */
inline ldiv_t ldivE(long numer, long denom) {
  /* The C99 and C++11 languages define both of these as truncating. */
  long q = numer / denom;
  long r = numer % denom;
  if (r < 0) {
    if (denom > 0) {
      q = q - 1;
      r = r + denom;
    } else {
      q = q + 1;
      r = r - denom;
    }
  }
  return (ldiv_t){.quot = q, .rem = r};
}

/* Floored division */
inline ldiv_t ldivF(long numer, long denom) {
  long q = numer / denom;
  long r = numer % denom;
  if ((r > 0 && denom < 0) || (r < 0 && denom > 0)) {
    q = q - 1;
    r = r + denom;
  }
  return (ldiv_t){.quot = q, .rem = r};
}

Note that for both cases, the remainder can be calculated independently of the quotient, but not vice versa. The operations are combined here to save screen space, as the logical branches are the same.

See also

• Modulo (disambiguation) and modulo (jargon) – many uses of the word modulo, all of which grew out of Carl F. Gauss's introduction of modular arithmetic in 1801.
• Modulo (mathematics), general use of the term in mathematics
• Modular exponentiation
• Turn (unit)

Notes

References

External links

• Modulorama, animation of a cyclic representation of multiplication tables (explanation in French)