## Hofstadter’s Sequence

### February 1, 2013

On page 73 of his book *Gödel, Escher, Bach: An Eternal Golden Braid*, Douglas Hofstadter asks you to think about the following sequence:

1, 3, 7, 12, 18, 26, 35, 45, 56, 69, …

Hofstadter leaves you to figure out the rule that produces the sequence, though he does give some hints in the surrounding text.

Your task is to write a program that produces the first *n* elements of Hofstadter’s sequence. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.

[...] today’s Programming Praxis exercise, our goal is to write a program that generates the Hofstadter sequence. [...]

My Haskell solution (see http://bonsaicode.wordpress.com/2013/02/01/programming-praxis-hofstadters-sequence/ for a version with comments):

[...] 1 2 [...]

[...] Question is from here. [...]

Java solution here

Here is a java solution that only stores the Hofstader’s sequence instead of both sequences.

public class Hofstadters {

Vector<Integer> hof = new Vector<Integer>();

final int MAX = 20;

int nextS;

int hofCheckIndex;

public Hofstadters(){

hof.add(1);

nextS = 2;

hofCheckIndex = 1;

for(int i = 1; i < MAX; i++) fillHof();

System.out.println("Hof index: " + MAX + " Hof number: " + hof.lastElement());

}

public void fillHof(){

hof.add(hof.lastElement() + nextS);

if(++nextS == hof.elementAt(hofCheckIndex)){

nextS++;

hofCheckIndex++;

}

}

Python generator

C# solution

C# 2, not as elegant.

List sequenceNumbers = new List();

int i = 1;

while (i < 100)

{

if (sequenceNumbers.IndexOf(i + sequenceNumbers.LastOrDefault()) == -1)

{

sequenceNumbers.Add(i + sequenceNumbers.LastOrDefault());

++i;

}

while (sequenceNumbers.IndexOf(i) != -1)

{

++i;

}

}

Here is Python code with some optimization – we store available numbers (S sequence members) as tuples (min, max) and pops first available. This significantly reduces number of items stored in list.

Here’s my racket solution.

(define (hof length (current 1) (last 0) (s empty))

(cond

((> current length) “done”)

((equal? current 1) (printf “1~%”) (hof length (+ current 1)))

((equal? current 2) (printf “3~%”) (hof length (+ current 1)))

((equal? current 3) (printf “7~%”) (hof length (+ current 1) 7 (list 5 6)))

(else

(let ((this-hof (+ last (car s))))

(printf “~a~%” (+ last (car s)))

(hof

length

(+ current 1)

(+ last (car s))

(append (cdr s) (build-list (- this-hof last 1)

(lambda (x) (+ x last 1)))))))))

Even less pretty but faster C# version. Idea is to enumerate S elements and skip previously found R elements. Got lazy to optimize further. On my i7 finds 1M numbers in about 35ms.

Really interesting problem. I had a crack for a few hours (eheh…) and came up with a solution in Common Lisp on my blog.

The solution works for n = 5, but interestingly, it also works for n = 1 (where it finds one solution; 1), and n = 0 (where it finds no solutions).

I hacked this out in Perl, and after the fact found it very similar to beardedone85’s Java solution

sub hofstadters_sequence{

my $n = shift;

my @R = (1); #Given

my $S = 2; #Given

my $r_index = 1;

for(my $i = 1; $i < $n; $i++){

#Step 1, first calculate the next R

push(@R, $R[$i-1] + $S);

#Step 2, calculate next S

if($S == $R[$r_index] – 1){

$S = $R[$r_index] + 1;

$r_index++;

}

else{$S++;}

}

return \@R;

}

import java.util.Vector;

public class HofstadtersSequence{

public static void main(String[] args){

Vector S = new Vector();

int N = 1000000, R = 2, n = 0;

int lastSkiped = 1;

S.add(1L);

while(N>n++){

System.out.println(S.lastElement()+” “+R);

S.add(S.lastElement()+R++);

if(R == S.get(lastSkiped)){

R++;

lastSkiped++;

}

}

System.out.println(S.lastElement()+” “+R);

}

}

Hofstadter mentions many sequences in his book. This particular one is also known as one of the “Hofstadter Figure-Figure sequences”.

int sum = 0, shadow = 1, count = 2, ng = 1;

for (int i = 1; i <= count – ng; i++)

{

sum = sum + shadow;

Console.Write(" " + sum + ", ");

shadow = shadow + 1;

if (i == count)

{

count = count + 1;

shadow = shadow + 1;

i = 0;

}

else if (sum == 150)

{

goto End;

}

ng = 0;

}

End:

Console.ReadLine();