Define a 1-indexed sequence as follows:

  • A083569(1) = 1
  • A083569(n) where n is an integer greater than 1, is the smallest integer m not occurring earlier such that m+n is a prime number.

Your task is to take in n and return A083569(n).

 n  A083569(n)
 1  1
 2  3
 3  2
 4  7
 5  6
 6  5
 7  4
 8  9
 9  8
10 13
11 12
12 11
13 10
14 15
15 14
16 21
17 20
18 19
19 18
20 17

More testcases can be found here. The original sequence on OEIS can be found here.

This is . Shortest answer in bytes wins. Standard loopholes apply.

  • \$\begingroup\$ @Mr.Xcoder "Define a 1-indexed sequence as follows" \$\endgroup\$
    – Leaky Nun
    Jun 19, 2017 at 9:45

9 Answers 9


Haskell, 87 86 83 80 74 69 bytes

Thanks to xnor for suggesting some changes that saved 3 bytes!

f n=[m|m<-[1..],all((>0).mod(n+m))[2..n+m-1],all((/=m).f)[1..n-1]]!!0

Try it online!

I'm new to Haskell, and Haskell golfing, feedback is appreciated!


We define a function f n. We define f n to be the first element !!0 of the list:


Broken down that is:

[m|          # Numbers m
m<-[1..],    # From the integers greater than 0
all          # Forall x
(>0).mod(n+m)# n+m mod x is not zero
[2..n+m-1]   # from the integers from 2 to n+m-1
all          # Forall
((/=m).f)    # when f is applied the result is not m
[1..n-1]     # from the integers from 1 to n-1
  • 4
    \$\begingroup\$ Welcome to Haskell golfing! The[2,3..] can just be [2..], counting up by 1 is default. There's a built-in notElem. \$\endgroup\$
    – xnor
    Jun 19, 2017 at 4:20
  • \$\begingroup\$ @xnor Thanks! I ended up finding a better way around using notElem but the first tip was helpful and I'll make sure to keep that second one in my back pocket. \$\endgroup\$
    – Wheat Wizard
    Jun 19, 2017 at 4:32
  • \$\begingroup\$ Looks like your new revision gets f 1 wrong, should be 1. \$\endgroup\$
    – xnor
    Jun 20, 2017 at 5:06
  • \$\begingroup\$ @xnor Fixed, unfortunately at the cost of 3 bytes. \$\endgroup\$
    – Wheat Wizard
    Jun 20, 2017 at 5:20

Jelly, 16 15 bytes


This assumes A083569(n) ≤ n² (the sequence appears to be growing linearly).

Try it online!

How it works

Rɓ²R+⁸ÆPTḟḢṭµ/Ṫ  Main link. Argument: n

R                Range; yield [1, ..., n].
 ɓ               Begin a dyadic chain with swapped arguments.
            µ/   Reduce the range by that chain.
                 If we call the chain f, this computes f(2,1), then f(3,f(2,1)),
                 then f(4,f(3,f(2,1)), etc.
                 The left argument is an integer k, the right one an array A.
  ²                Square; yield k².
   R               Range; yield [1, ..., k²].
    +⁸             Add k, yielding [1+k, ..., k²+k].
      ÆP           Test each sum for primality.
        T          Truth; get all indices of 1‘s. This finds all m in [1, ..., k²]
                   such that m+k is prime.
         ḟ         Filterfalse; remove all resulting elements that appear in A.
          Ḣ        Head; extract the first remaining result.
           ṭ       Tack; append the extracted integer to A.
                 This computes the first n elements of the sequence.
              Ṫ  Tail; extract the last, n-th element.
  • 4
    \$\begingroup\$ Indeed, A083569(n) is at most the nth prime larger than n by its definition, which is at most the 2nth prime, which (for n≥3) is less than 4n*log(n) by results of Rosser–Schoenfeld. \$\endgroup\$ Jun 19, 2017 at 5:28
  • \$\begingroup\$ While @GregMartin verified it, it's still a pretty wild assumption to make... \$\endgroup\$ Jun 19, 2017 at 6:18
  • 4
    \$\begingroup\$ @Challenger5 I prefer "educated guess". \$\endgroup\$
    – Dennis
    Jun 19, 2017 at 6:25

Pyth - 18 17 15 bytes

Thanks to @isaacg for saving me two bytes!

Back on this site, after being busy for a while, will hopefully golf this one further.


Try it online here.

  • 4
    \$\begingroup\$ Welcome back to PPCG! \$\endgroup\$
    – Leaky Nun
    Jun 19, 2017 at 3:53
  • \$\begingroup\$ @LeakyNun Thanks :) \$\endgroup\$
    – Maltysen
    Jun 19, 2017 at 3:58
  • 1
    \$\begingroup\$ -TY is one byte shorter than !/YT, and truthy in the same cases. \$\endgroup\$
    – isaacg
    Jun 19, 2017 at 8:04
  • \$\begingroup\$ You can save another byte by changing +hdT to +Th. \$\endgroup\$
    – isaacg
    Jun 19, 2017 at 8:08
  • \$\begingroup\$ @isaacg, oh does it cast the first element to a list? That's really cool. \$\endgroup\$
    – Maltysen
    Jun 19, 2017 at 11:27

C# (.NET Core), 169 bytes

n=>{if(n<2)return 1;var p=new int[n-1];int i=0,j,s;for(;i<n-1;)p[i]=f(++i);for(i=1;;i++){for(j=2,s=i+n;j<s&&s%j++>0;);if(j==s&!System.Array.Exists(p,e=>e==i))return i;}}

Try it online!

By far the most inefficient way to calculate the results, so please refrain from calculating f(n) for n>=30 with this code. The first step is to recursively calculate the values from f(1) to f(n-1) and then proceed to calculate f(n) by searching for the first i such that n+i is prime and i is not in the previous values list.


x86-64 Assembly, 57 55 bytes

I'm new to golfing, so comments/feedback are appreciated.

Note: this is optimized for machine code length, not for source length.

0: 89 f8 ff cf 74 32 97 89 fe 89 f1 ff c6 89 f0 99
1: f7 f1 85 d2 e0 f7 85 c9 75 ed 89 f9 ff c9 56 29
2: fe 56 57 51 89 fc e8 d3 ff ff ff 59 5f 5e 39 c6
3: e0 ef 96 5e 74 d1 c3

Defines a function, using the standard convention (i.e. return value in eax, first argument in edi, all registers caller-saved except ebx) that takes an unsigned 32-bit integer and returns the smallest m etc.


    .globl a083569
    // edi = original, probably don't touch
    // esi = candidate prime, if it's not a repeat we return edi-this
    mov %edi, %eax
    dec %edi
    jz end
    xchg %eax, %edi
    mov %edi, %esi
    mov %esi, %ecx
    inc %esi
    mov %esi, %eax
    div %ecx
    test %edx, %edx
    loopnz primeloop
/* end */
    // if esi isn't prime, then ecx is now one or greater.
    test %ecx, %ecx
    jnz primecheck
    // esi is now our target prime: check if it's not already one
    mov %edi, %ecx
    dec %ecx
    push %rsi   /* we need a flag-safe way to restore this later */
    sub %edi, %esi
    push %rsi
    push %rdi
    push %rcx
    mov %ecx, %edi
    call a083569
    pop %rcx
    pop %rdi
    pop %rsi
    cmp %eax, %esi
    loopne chkdup
/* end loop - chkdup */
    xchg %esi, %eax
    pop %rsi
    je primecheck
/* end outer loop - primecheck */

Try it online!


Clojure, 158 155 bytes

#(loop[r[0 1]i 1](if(= i %)(last r)(recur(conj r(nth(for[j(range):when(=((set r)j)(seq(for[k(range 2(+ 1 i j)):when(=(mod(+ 1 i j)k)0)]j)))]j)0))(inc i))))

This might still have some fat, I'm not quite happy with (+ 1 i j) but this was the easiest way to handle base case n = 1 and the rest. ((set r)j) returns nil if j is not in the set, and (seq ()) on an empty list returns nil as well. Calculates n = 1000 in 48 seconds.

Update: removed nil from = check as the code works correctly also without it.


Ruby, 62+8 = 70 bytes

Uses the -rprime flag.

f=->n,o=[]{o<<f[n-1,o]if n>1;Prime.find{|i|i>n&&o|[i-n]!=o}-n}

Try it online!


Python, 194 170 110 bytes

84 bytes saved by Leaky Nun

2 bytes saved by mathmandan

def s(n):
 a=[s(j)for j in range(1,n)];i=1
 while(i in a)|any((i+n)%j<1for j in range(2,i+n)):i+=1
 return i

Defines a function s(n) that takes a number as input and returns A083569(n).

Try it Online

  • 1
    \$\begingroup\$ You might consider including this TIO link. \$\endgroup\$
    – Leaky Nun
    Jun 19, 2017 at 3:55
  • 1
    \$\begingroup\$ You can use p=lambda n:any(n%i<1for i in range(2,n)) for the primality check. \$\endgroup\$
    – Leaky Nun
    Jun 19, 2017 at 4:06
  • 1
    \$\begingroup\$ 110 bytes \$\endgroup\$
    – Leaky Nun
    Jun 19, 2017 at 7:09
  • 1
    \$\begingroup\$ You can use bitwise or to save a couple bytes: while(i in a)|any(... \$\endgroup\$
    – mathmandan
    Jun 19, 2017 at 16:38

Husk, 16 bytes


Try it online!



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.