EDIT::
There was too much iteration happening in the earlier version. This new version contains no loops AND no branching.
Still works with negative n for right rotation and positive n for left rotation for any size n,
Mutation free
function rotate(A,n,l=A.length) {
const offset = (((n % l) + l) %l)
return A.slice(offset).concat(A.slice(0,offset))
}
Golfed version:
const r = (A,n,l=A.length,i=((n%l)+l)%l)=>A.slice(i).concat(A.slice(0,i))
EDIT1::*
Branchless, mutationless implementation.
There was an unnecessary branch. Here is a working branchless solution.
// negative num = right rotate by |num|
// positive num = left rotate by num
function r(A,n,l=A.length) {
return A.map((x,i,a) => A[(((n+i)%l) + l) % l])
}
The equation ((n%l) + l) % l
maps exactly positive and negative numbers of any arbitrarily large values of n
ORIGINAL
Rotate left and right. Rotate left with positive n
, rotate right with negative n
.
Works for obscenely large inputs of n
.
No mutation mode. Too much mutation in these answers.
Also, fewer operations than most answers. No pop, no push, no splice, no shift.
const rotate = (A, num ) => {
return A.map((x,i,a) => {
const n = num + i
return n < 0
? A[(((n % A.length) + A.length) % A.length)]
: n < A.length
? A[n]
: A[n % A.length]
})
}
or
const rotate = (A, num) => A.map((x,i,a, n = num + i) =>
n < 0
? A[(((n % A.length) + A.length) % A.length)]
: n < A.length
? A[n]
: A[n % A.length])
//test
rotate([...Array(5000).keys()],4101) //left rotation
rotate([...Array(5000).keys()],-4101000) //right rotation, num is negative
// will print the first index of the array having been rotated by -i
// demonstrating that the rotation works as intended
[...Array(5000).keys()].forEach((x,i,a) => {
console.log(rotate(a,-i)[0])
})
// prints even numbers twice by rotating the array by i * 2 and getting the first value
//demonstrates the propper mapping of positive number rotation when out of range
[...Array(5000).keys()].forEach((x,i,a) => {
console.log(rotate(a,i*2)[0])
})
Explanation:
map each index of A to the value at index offset. In this case
offset = num
if the offset < 0
then offset + index + positive length of A
will point to the inverse offset.
if offset > 0 and offset < length of A
then simply map the current index to the offset index of A.
Otherwise, modulo the offset and the length to map the offset in the bounds of the array.
Take for instance offset = 4
and offset = -4
.
When offset = -4
, and A = [1,2,3,4,5]
, for each index, offset + index
will make the magnitude (or Math.abs(offset)
) smaller.
Let's explain the calculation for the index of negative n first. A[(((n % A.length) + A.length) % A.length)+0]
and been intimidated. Don't be. It took me 3 minutes in a Repl to work it out.
- We know
n
is negative because the case is n < 0
. If the number is larger than the range of the Array, n % A.length
will map it into the range.
n + A.length
add that number to A.length
to offset n the correct
amount.
- We know
n
is negative because the case is n < 0
. n + A.length
add that number to A.length
to offset n the correct amount.
- Next Map it to the range of the length of A using modulo. The second modulous is necessary to map the result of the calculation into an indexable range

- First index: -4 + 0 = -4. A.length = 5. A.length - 4 = 1. A2 is 2. Map index 0 to 2.
[2,... ]
- Next index, -4 + 1 = -3. 5 + -3 = 2. A2 is 3. Map index 1 to 3.
[2,3... ]
- Etc.
The same process applies to offset = 4
.
When offset = -4
, and A = [1,2,3,4,5]
, for each index, offset + index
will make the magnitude bigger.
4 + 0 = 0
. Map A[0] to the value at A[4]. [5...]
4 + 1 = 5
, 5 is out of bounds when indexing, so map A2 to the
value at the remainder of 5 / 5
, which is 0. A2 = value at
A[0]. [5,1...]
- repeat.