How this instrument works
Floor division answers a blunter question than ordinary division: not 'what real number is a ⁄ b', but 'how many whole times does b fit into a, discarding whatever is left over'. Formally it is ⌊a ⁄ b⌋ — evaluate the ordinary quotient a ⁄ b, then round it down to the whole number sitting directly beneath it, never up to the one above. Ordinary division and floor division start from that same real-number answer, but the floor then strips away everything after the decimal point, and that missing piece is exactly the number the division algorithm calls the remainder.
Every floor division is half of an older, tidier statement: the division algorithm, a = b·q + r with 0 ≤ r < b, a result attributed to Euclid's Elements. Floor division supplies q, the quotient; the leftover r is what a modulo operation would return. Because r is pinned between 0 and b, the pair (q, r) is unique for any a and any positive b — there is exactly one way to split a into that many full groups of b plus a remainder too small to form another group.
The one place floor division surprises people is a negative dividend. Rounding toward zero — the kind a pocket calculator's plain integer-division button gives — would send −17 ⁄ 5 to −3, dropping the 0.4. Floor division instead keeps descending past −3, landing on −4, since −4 is the largest whole number that still sits at or below −3.4. Python's // operator follows this floor rule on purpose; C's classic integer division follows truncation instead, and that gap between the two rules is a real source of bugs whenever code moves from one convention to the other.
- Enter the number being divided into the a field; any real value works, whole or fractional, above or below zero.
- Enter the divisor into the b field; it must be greater than zero, since the result is undefined at b = 0.
- Read ⌊a ⁄ b⌋ for the whole-number quotient — the count of full times b fits into a, with any leftover discarded.
- Try a negative a, such as −17, and confirm ⌊a ⁄ b⌋ settles on −4 — a full group beyond −3 — rather than the zero-bound answer a truncating calculator would give.
Worked example — 17 into groups of 5
Set a to 17 and b to 5. Ordinary division gives a ⁄ b = 3.4, but ⌊a ⁄ b⌋ discards the 0.4 and returns 3 — the whole number of times 5 fits inside 17. Picture 17 apples sorted into bags of 5: three bags fill completely and 2 apples are left over unbagged, since 17 = 5 × 3 + 2 with the remainder r = 2 satisfying 0 ≤ r < 5. Floor division reports only the 3 full bags, not the leftover fruit.
Flip the sign of a to −17 with b left at 5 and the answer is not −3, the truncated guess, but −4: −4 × 5 = −20, and −17 − (−20) = 3 is the remainder, still inside the required range 0 ≤ r < 5. The quotient moved past −3 to the next integer down because floor division keeps descending until it clears −3.4 entirely, never snapping back up toward zero the way truncation does, so a negative dividend needs one more whole group than truncation would suggest.
Questions
What does floor division actually compute?
It computes the whole number of times the divisor fits into the dividend, discarding any leftover: ⌊a ⁄ b⌋, the floor of the ordinary quotient a ⁄ b. For 17 and 5 that is 3, because 5 fits into 17 three complete times before 2 is left over — the 0.4 fractional part of 3.4 never appears in the answer.
How is floor division different from truncating toward zero?
For positive numbers the two agree, but they split apart the moment the dividend goes negative. Truncation rounds −17 ⁄ 5 = −3.4 toward zero, to −3. Floor division instead keeps going one step further, to −4, since −4 is the largest whole number that still sits at or below −3.4. Python's // and this sheet both follow the floor rule; C's plain integer division follows truncation, which is why the same expression can give different answers in different languages.
How does the remainder relate to the floor division result?
They come from the same identity: a = b·q + r, with q = ⌊a ⁄ b⌋ and 0 ≤ r < b. Once the quotient is known, the remainder falls out as r = a − b·q. For 17 and 5, q = 3 and r = 17 − 15 = 2; for −17 and 5, q = −4 and r = −17 − (−20) = 3 — always non-negative, never equal to or larger than b.
Why must the divisor b be greater than zero here?
Division by zero is undefined — there is no whole number of times zero fits into anything — so b = 0 is rejected outright. Restricting b to positive values also keeps the remainder identity 0 ≤ r < b unambiguous; negative divisors are mathematically possible but shift which direction the remainder is pinned to, and most practical uses of floor division never need that case.
Where does floor division show up outside a math class?
Anywhere a total must be split into fixed-size groups with a leftover set aside: paginating 137 search results into pages of 20 gives ⌊137 ⁄ 20⌋ = 6 full pages plus a partial seventh; converting 137 minutes into whole hours gives ⌊137 ⁄ 60⌋ = 2; indexing schemes in most programming languages lean on the same operation to find which row a flattened position belongs to.
Is floor division just the floor function applied after dividing?
Yes — ⌊a ⁄ b⌋ means take the ordinary real-number quotient a ⁄ b and then apply the floor function, the greatest-integer rule that always rounds down. The only difference from an ordinary floor calculation is that the input here is itself a division of two numbers you supply, rather than a single value already computed elsewhere.