Mertens function

Revision as of 16:56, 28 January 2021 by Not a robot (talk | contribs) (Add Fortran)

The Mertens function M(x) is the count of square-free integers up to x that have an even number of prime factors, minus the count of those that have an odd number.

Task
Mertens function
You are encouraged to solve this task according to the task description, using any language you may know.

It is an extension of the Möbius function. Given the Möbius function μ(n), the Mertens function M(x) is the sum of the Möbius numbers from n == 1 through n == x.


Task
  • Write a routine (function, procedure, whatever) to find the Mertens number for any positive integer x.
  • Use that routine to find and display here, on this page, at least the first 99 terms in a grid layout. (Not just one long line or column of numbers.)
  • Use that routine to find and display here, on this page, the number of times the Mertens function sequence is equal to zero in the range M(1) through M(1000).
  • Use that routine to find and display here, on this page, the number of times the Mertens function sequence crosses zero in the range M(1) through M(1000). (Crossing defined as this term equal to zero but preceding term not.)


See also


This is not code golf.   The stackexchange link is provided as an algorithm reference, not as a guide.


Related tasks



8080 Assembly

<lang 8080asm>MAX: equ 1000 ; Amount of numbers to generate org 100h ;;; Generate Mertens numbers lxi b,1 ; Start at place 1; BC = current Mertens number lxi h,MM ; First one is 1 dad b mvi m,1 outer: inx b ; Next Mertens number lxi h,MM dad b mvi m,1 ; Initialize at 1 lxi d,2 ; DE = inner loop counter ('k'), starts at 2 ;;; Now we need to find BC/DE, but there is no hardware divide ;;; We also need to be somewhat clever so it doesn't take forever inner: push d ; Keep both loop counters safe on the stack push b xchg ; Divisor in HL mov d,b ; Dividend in DE mov e,c lxi b,100h ; B = counter, C = zero double: dad h ; Double divisor inr b ; Increment counter call cdehl ; Dividend <= divisor? jnc double ; If so, keep doubling mov a,b ; Keep counter mov b,c ; BC = 0 push b ; Push result variable on stakc (initial 0) mov b,a ; Restore counter xchg ; HL = dividend, DE = doubled divisor subtr: mov a,l ; Try HL -= DE sub e mov l,a mov a,h sbb d mov h,a xthl ; Get result accumulator from stack cmc ; Flip borrow mov a,l ; Rotate into result ral mov l,a mov a,h ral mov h,a mov a,l ; Retrieve flag rar xthl ; Retrieve rest of divisor jc $+4 ; If borrow, dad d ; Add dividend back into divisor xra a  ; DE >> 1 ora d rar mov d,a mov a,e rar mov e,a dcr b  ; Are we there yet? jnz subtr  ; If not, try another subtraction pop h ; HL = quotient ;;; Division is done, do lookup and subraction lxi d,MM ; Look up M[outer/inner] dad d mov e,m ; E = M[BC/DE] pop b ; Restore BC (n) lxi h,MM dad b mov a,m ; A = M[BC] sub e ; A = M[BC] - M[BC/DE] mov m,a ; M[BC] = A pop d ; Restore DE (k) ;;; Update loops inx d ; k++ call cbcde ; DE <= BC? jnc inner lxi h,MAX call chlbc ; BC <= MAX? jnc outer ;;; Print table lxi d,frst99 call puts lxi h,MM+1 ; Start of Merten numbers mvi c,9 ; Column counter table: mov a,m ; Get Merten number ana a ; Set flags mvi b,' ' ; Space jp prtab ; If positive, print space-number-space mvi b,'-' ; Otherwise, print minus sign cma ; And negate the number (make positive) inr a prtab: adi '0' ; Make ASCII digit mov d,a ; Keep number mov a,b ; Print space or minus sign call putc mov a,d ; Restore number call putc ; Print number mvi a,' ' ; Print space call putc dcr c ; Decrement column counter jnz tnext lxi d,nl ; End of columns - print newline call puts mvi c,10 ; Column counter tnext: inx h ; Table done? mov a,l cpi 100 jnz table ; If not, keep going ;;; Find zeroes and crossings lxi b,0 ; B=zeroes, C=crossings lxi d,MAX ; Counter lxi h,MM+1 count: mov a,m ; Get number ana a ; Zero? jnz cnext inr b ; If so, add zero dcx h ; Previous number also zero? mov a,m inx h ana a jz cnext inr c ; If not, add crossiong cnext: inx h dcx d mov a,d ora e jnz count lxi d,zero ; Print zeroes call puts mov a,b call puta lxi d,cross ; Print crossings call puts mov a,c call puta lxi d,tms jmp puts ;;; Print character in A using CP/M, keeping registers putc: push b push d push h mov e,a mvi c,2 call 5 jmp resrgs ;;; Print number in A, keeping registers puta: push b push d push h lxi h,num putad: mvi c,-1 putal: inr c sui 10 jnc putal adi 10+'0' dcx h mov m,a mov a,c ana a jnz putad xchg mvi c,9 call 5 jmp resrgs ;;; Print string in DE using CP/M, keeping registers puts: push b push d push h mvi c,9 call 5 resrgs: pop h pop d pop b ret cdehl: mov a,d cmp h rnz mov a,e cmp l ret cbcde: mov a,b cmp d rnz mov a,c cmp e ret chlbc: mov a,h cmp b rnz mov a,l cmp c ret ;;; Strings db '***' num: db '$' frst99: db 'First 99 Mertens numbers:',13,10,' $' nl: db 13,10,'$' zero: db 'M(N) is zero $' cross: db ' times.',13,10,'M(N) crosses zero $' tms: db ' times.$' ;;; Numbers are stored page-aligned after program MM: equ ($/256)*256+256 </lang>

Output:
First 99 Mertens numbers:
    1  0 -1 -1 -2 -1 -2 -2 -2
-1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2
-3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3
-3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1
-2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2
-2 -1 -1  0  1  2  2  1  1  1
M(N) is zero 92 times.
M(N) crosses zero 59 times.


8086 Assembly

<lang asm>MAX: equ 1000 ; Amount of Mertens numbers to generate puts: equ 9 ; MS-DOS syscall to print a string putch: equ 2 ; MS-DOS syscall to print a character cpu 8086 org 100h section .text ;;; Generate Mertens numbers mov bx,M ; BX = pointer to start of Mertens numbers mov si,1 ; Current Mertens number mov [si+bx],byte 1 ; First Mertens number is 1 outer: inc si ; Next Mertens number mov [si+bx],byte 1 ; Starts out at 1... mov cx,2 ; CX = from 2 to current number, inner: mov ax,si ; Divide current number, xor dx,dx div cx ; By CX mov di,ax mov al,[di+bx] ; Get value at that location sub [si+bx],al ; Subtract from current number inc cx cmp cx,si jbe inner cmp si,MAX jbe outer ;;; Print the table mov dx,frst99 ; First string call outstr mov si,1 ; Start at index 1 mov dh,9 ; Column count table: mov cl,[si+bx] ; Get item test cl,cl mov dl,' ' jns .print ; Positive? mov dl,'-' ; Otherwise, it is negative, neg cl ; print ' ' and negate .print: call putc ; Print space or minus add cl,'0' ; Add ASCII 0 mov dl,cl call putc ; Print number mov dl,' ' call putc ; Print space dec dh ; One less column left jnz .next mov dx,nl ; Print newline call outstr mov dh,10 .next: inc si ; Done yet? cmp si,100 jb table ; If not, print next item from table ;;; Calculate zeroes and crossings xor cx,cx ; CL = zeroes, CH = crossings mov si,1 mov al,[si+bx] ; AL = current item zc: inc si mov ah,al ; AH = previous item mov al,[si+bx] test al,al ; Zero? jnz .next inc cx ; Then increment zero counter test ah,ah ; Previous one also zero? jz .next inc ch ; Then increment crossing counter .next: cmp si,MAX ; Done yet? jbe zc ;;; Print zeroes and crossings mov dx,zero call outstr mov al,cl call putal mov dx,cross call outstr mov al,ch call putal mov dx,tms jmp outstr putc: mov ah,putch ; Print character int 21h ret ;;; Print AL in decimal format putal: mov di,num .loop: aam ; Extract digit add al,'0' ; Store digit dec di mov [di],al mov al,ah ; Rest of number test al,al ; Done? jnz .loop ; If not, get more digits mov dx,di ; Otherwise, print string outstr: mov ah,puts int 21h ret section .data db '***' ; Number output placeholder num: db '$' frst99: db 'First 99 Mertens numbers:',13,10,' $' nl: db 13,10,'$' zero: db 'M(N) is zero $' cross: db ' times.',13,10,'M(N) crosses zero $' tms: db ' times.$' section .bss mm: resb MAX ; Mertens numbers M: equ mm-1 ; 1-based indexing</lang>

Output:
First 99 Mertens numbers:
    1  0 -1 -1 -2 -1 -2 -2 -2
-1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2
-3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3
-3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1
-2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2
-2 -1 -1  0  1  2  2  1  1  1
M(N) is zero 92 times.
M(N) crosses zero 59 times.

APL

Works with: Dyalog APL

<lang APL>mertens←{

   step  ← {⍵,-⍨/⌽1,⍵[⌊n÷1↓⍳n←1+≢⍵]}
   m1000 ← step⍣999⊢,1
   zero  ← m1000+.=0
   cross ← +/(~∧1⌽⊢)m1000≠0
   ⎕←'First 99 Mertens numbers:'
   ⎕←10 10⍴'∘',m1000
   ⎕←'M(N) is zero ',(⍕zero),' times.'
   ⎕←'M(N) crosses zero ',(⍕cross),' times.'

}</lang>

Output:
First 99 Mertens numbers:
 ∘  1  0 ¯1 ¯1 ¯2 ¯1 ¯2 ¯2 ¯2
¯1 ¯2 ¯2 ¯3 ¯2 ¯1 ¯1 ¯2 ¯2 ¯3
¯3 ¯2 ¯1 ¯2 ¯2 ¯2 ¯1 ¯1 ¯1 ¯2
¯3 ¯4 ¯4 ¯3 ¯2 ¯1 ¯1 ¯2 ¯1  0
 0 ¯1 ¯2 ¯3 ¯3 ¯3 ¯2 ¯3 ¯3 ¯3
¯3 ¯2 ¯2 ¯3 ¯3 ¯2 ¯2 ¯1  0 ¯1
¯1 ¯2 ¯1 ¯1 ¯1  0 ¯1 ¯2 ¯2 ¯1
¯2 ¯3 ¯3 ¯4 ¯3 ¯3 ¯3 ¯2 ¯3 ¯4
¯4 ¯4 ¯3 ¯4 ¯4 ¯3 ¯2 ¯1 ¯1 ¯2
¯2 ¯1 ¯1  0  1  2  2  1  1  1
M(N) is zero 92 times.
M(N) crosses zero 59 times.

BASIC

Note that if you actually try this on a real 8-bit micro, this will take literal hours to run. Interpreted BASIC is not fast.

For comparison, the 8080 and 8086 assembly versions above run in around 4 minutes and 15 seconds respectively, on real hardware. This took nearly 4 hours on the 8080 (using MBASIC) and a little over 1 hour on the 8086 (using GWBASIC).

<lang BASIC>10 DEFINT C,Z,N,K,M: DIM M(1000) 20 M(1)=1 30 FOR N=2 TO 1000 40 M(N)=1 50 FOR K=2 TO N: M(N) = M(N)-M(INT(N/K)): NEXT 60 NEXT 70 PRINT "First 99 Mertens numbers:" 80 PRINT " "; 90 FOR N=1 TO 99 100 PRINT USING "###";M(N); 110 IF N MOD 10 = 9 THEN PRINT 120 NEXT 130 C=0: Z=0 140 FOR N=1 TO 1000 150 IF M(N)=0 THEN Z=Z+1: IF M(N-1)<>0 THEN C=C+1 160 NEXT 170 PRINT "M(N) is zero";Z;"times." 180 PRINT "M(N) crosses zero";C;"times." 190 END</lang>

Output:
First 99 Mertens numbers:
     1  0 -1 -1 -2 -1 -2 -2 -2
 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2
 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3
 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1
 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2
 -2 -1 -1  0  1  2  2  1  1  1
M(N) is zero 92 times.
M(N) crosses zero 59 times.

C

<lang c>#include <stdio.h>

  1. include <stdlib.h>

int* mertens_numbers(int max) {

   int* m = malloc((max + 1) * sizeof(int));
   if (m == NULL)
       return m;
   m[1] = 1;
   for (int n = 2; n <= max; ++n) {
       m[n] = 1;
       for (int k = 2; k <= n; ++k)
           m[n] -= m[n/k];
   }
   return m;

}

int main() {

   const int max = 1000;
   int* mertens = mertens_numbers(max);
   if (mertens == NULL) {
       fprintf(stderr, "Out of memory\n");
       return 1;
   }
   printf("First 199 Mertens numbers:\n");
   const int count = 200;
   for (int i = 0, column = 0; i < count; ++i) {
       if (column > 0)
           printf(" ");
       if (i == 0)
           printf("  ");
       else
           printf("%2d", mertens[i]);
       ++column;
       if (column == 20) {
           printf("\n");
           column = 0;
       }
   }
   int zero = 0, cross = 0, previous = 0;
   for (int i = 1; i <= max; ++i) {
       int m = mertens[i];
       if (m == 0) {
           ++zero;
           if (previous != 0)
               ++cross;
       }
       previous = m;
   }
   free(mertens);
   printf("M(n) is zero %d times for 1 <= n <= %d.\n", zero, max);
   printf("M(n) crosses zero %d times for 1 <= n <= %d.\n", cross, max);
   return 0;

}</lang>

Output:
First 199 Mertens numbers:
    1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1
 1  0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3
-3 -3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4
-4 -3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0
 0  1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3
-3 -4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8
M(n) is zero 92 times for 1 <= n <= 1000.
M(n) crosses zero 59 times for 1 <= n <= 1000.

C++

<lang cpp>#include <iomanip>

  1. include <iostream>
  2. include <map>

class mertens_calculator { public:

   int mertens_number(int);

private:

   std::map<int, int> cache_;

};

int mertens_calculator::mertens_number(int n) {

   auto i = cache_.find(n);
   if (i != cache_.end())
       return i->second;
   int m = 1;
   for (int k = 2; k <= n; ++k)
       m -= mertens_number(n/k);
   cache_.emplace(n, m);
   return m;

}

void print_mertens_numbers(mertens_calculator& mc, int count) {

   int column = 0;
   for (int i = 0; i < count; ++i) {
       if (column > 0)
           std::cout << ' ';
       if (i == 0)
           std::cout << "  ";
       else
           std::cout << std::setw(2) << mc.mertens_number(i);
       ++column;
       if (column == 20) {
           std::cout << '\n';
           column = 0;
       }
   }

}

int main() {

   mertens_calculator mc;
   std::cout << "First 199 Mertens numbers:\n";
   print_mertens_numbers(mc, 200);
   int zero = 0, cross = 0, previous = 0;
   for (int i = 1; i <= 1000; ++i) {
       int m = mc.mertens_number(i);
       if (m == 0) {
           ++zero;
           if (previous != 0)
               ++cross;
       }
       previous = m;
   }
   std::cout << "M(n) is zero " << zero << " times for 1 <= n <= 1000.\n";
   std::cout << "M(n) crosses zero " << cross << " times for 1 <= n <= 1000.\n";
   return 0;

}</lang>

Output:
First 199 Mertens numbers:
    1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1
 1  0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3
-3 -3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4
-4 -3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0
 0  1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3
-3 -4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8
M(n) is zero 92 times for 1 <= n <= 1000.
M(n) crosses zero 59 times for 1 <= n <= 1000.

Cowgol

<lang cowgol>include "cowgol.coh";

const MAX := 1000;

  1. Table output

sub printtab(n: int8) is

   if n<0 then 
       print_char('-');
       n := -n;
   else 
       print_char(' ');
   end if;
   print_char(n as uint8 + '0');
   print_char(' ');

end sub;

  1. Generate Merten numbers

var M: int8[MAX+1]; M[0] := 0; M[1] := 1;

var n: @indexof M := 2; while n < @sizeof M loop

   M[n] := 1;
   var k: @indexof M := 2;
   while k <= n loop
       M[n] := M[n] - M[n/k];
       k := k + 1;
   end loop;
   n := n + 1;

end loop;

  1. Find zeroes and crossings

var zero: uint8 := 0; var cross: uint8 := 0; n := 1; while n < @sizeof M loop

   if M[n] == 0 then
       zero := zero + 1;
       if M[n-1] != 0 then
           cross := cross + 1;
       end if;
   end if;
   n := n + 1;

end loop;

  1. Print table

print("The first 99 Mertens numbers are:\n"); print(" "); n := 1; var col: uint8 := 9; while n < 100 loop

   printtab(M[n]);
   col := col - 1;
   if col == 0 then    
       print_nl();
       col := 10;
   end if;
   n := n + 1;

end loop;

print("M(n) is zero "); print_i8(zero); print(" times\n"); print("M(n) crosses zero "); print_i8(cross); print(" times\n");</lang>

Output:
The first 99 Mertens numbers are:
    1  0 -1 -1 -2 -1 -2 -2 -2
-1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2
-3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3
-3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1
-2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2
-2 -1 -1  0  1  2  2  1  1  1
M(n) is zero 92 times
M(n) crosses zero 59 times

Factor

Works with: Factor version 0.99 2020-01-23

<lang factor>USING: formatting grouping io kernel math math.extras math.ranges math.statistics prettyprint sequences ;

! Take the cumulative sum of the mobius sequence to avoid ! summing lower terms over and over.

mertens-upto ( n -- seq ) [1,b] [ mobius ] map cum-sum ;

"The first 199 terms of the Mertens sequence:" print 199 mertens-upto " " prefix 20 group [ [ "%3s" printf ] each nl ] each nl

"In the first 1,000 terms of the Mertens sequence there are:" print 1000 mertens-upto [ [ zero? ] count bl pprint bl "zeros." print ] [

   2 <clumps> [ first2 [ 0 = not ] [ zero? ] bi* and ] count bl
   pprint bl "zero crossings." print

] bi</lang>

Output:
The first 199 terms of the Mertens sequence:
     1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1
  1  0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3
 -3 -3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4
 -4 -3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0
  0  1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3
 -3 -4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8

In the first 1,000 terms of the Mertens sequence there are:
 92 zeros.
 59 zero crossings.

Fortran

<lang Fortran> program Mertens

     implicit none
     integer M(1000), n, k, zero, cross
      

C Generate Mertens numbers

     M(1) = 1
     do 10 n=2, 1000
         M(n) = 1
         do 10 k=2, n
             M(n) = M(n) - M(n/k)
10   continue 

C Print table

     write (*,"('The first 99 Mertens numbers are:')")
     write (*,"('   ')",advance='no')
     k = 9
     do 20 n=1, 99
         write (*,'(I3)',advance='no') M(n)
         k = k-1
         if (k .EQ. 0) then
             k=10
             write (*,*)
         end if
20   continue

C Calculate zeroes and crossings

     zero = 0
     cross = 0
     do 30 n=2, 1000
         if (M(n) .EQ. 0) then
             zero = zero + 1
             if (M(n-1) .NE. 0) cross = cross+1
         end if    
30   continue 

40   format("M(N) is zero ",I2," times.")
     write (*,40) zero
50   format("M(N) crosses zero ",I2," times.")
     write (*,50) cross
     end program</lang>
Output:
The first 99 Mertens numbers are:
     1  0 -1 -1 -2 -1 -2 -2 -2
 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2
 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3
 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1
 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2
 -2 -1 -1  0  1  2  2  1  1  1
M(N) is zero 92 times.
M(N) crosses zero 59 times.

FreeBASIC

<lang freebasic>function padto( i as ubyte, j as integer ) as string

   return wspace(i-len(str(j)))+str(j)

end function

dim as integer M( 1 to 1000 ), n, col, k, psum dim as integer num_zeroes = 0, num_cross = 0 dim as string outstr

M(1) = 1 for n = 2 to 1000

   psum = 0
   for k = 2 to n
       psum += M(int(n/k))
   next k
   M(n) = 1 - psum
   if M(n) = 0 then
       num_zeroes += 1
       if M(n-1)<>0 then
           num_cross += 1
       end if
   end if

next n

print using "There are ### zeroes in the range 1 to 1000."; num_zeroes print using "There are ### crossings in the range 1 to 1000."; num_cross print "The first 100 Mertens numbers are: "

for n=1 to 100

   outstr += padto(3, M(n))+"  "
   if n mod 10 = 0 then
       print outstr
       outstr = ""
   end if

next n</lang>

Output:
There are  92 zeroes in the range 1 to 1000.
There are  59 crossings in the range 1 to 1000.
The first 100 Mertens numbers are: 
  1    0   -1   -1   -2   -1   -2   -2   -2   -1  
 -2   -2   -3   -2   -1   -1   -2   -2   -3   -3  
 -2   -1   -2   -2   -2   -1   -1   -1   -2   -3  
 -4   -4   -3   -2   -1   -1   -2   -1    0    0  
 -1   -2   -3   -3   -3   -2   -3   -3   -3   -3  
 -2   -2   -3   -3   -2   -2   -1    0   -1   -1  
 -2   -1   -1   -1    0   -1   -2   -2   -1   -2  
 -3   -3   -4   -3   -3   -3   -2   -3   -4   -4  
 -4   -3   -4   -4   -3   -2   -1   -1   -2   -2  
 -1   -1    0    1    2    2    1    1    1    1

Go

<lang go>package main

import "fmt"

func mertens(to int) ([]int, int, int) {

   if to < 1 {
       to = 1
   }
   merts := make([]int, to+1)
   primes := []int{2}
   var sum, zeros, crosses int
   for i := 1; i <= to; i++ {
       j := i
       cp := 0      // counts prime factors
       spf := false // true if there is a square prime factor
       for _, p := range primes {
           if p > j {
               break
           }
           if j%p == 0 {
               j /= p
               cp++
           }
           if j%p == 0 {
               spf = true
               break
           }
       }
       if cp == 0 && i > 2 {
           cp = 1
           primes = append(primes, i)
       }
       if !spf {
           if cp%2 == 0 {
               sum++
           } else {
               sum--
           }
       }
       merts[i] = sum
       if sum == 0 {
           zeros++
           if i > 1 && merts[i-1] != 0 {
               crosses++
           }
       }
   }
   return merts, zeros, crosses

}

func main() {

   merts, zeros, crosses := mertens(1000)
   fmt.Println("Mertens sequence - First 199 terms:")
   for i := 0; i < 200; i++ {
       if i == 0 {
           fmt.Print("    ")
           continue
       }
       if i%20 == 0 {
           fmt.Println()
       }
       fmt.Printf("  % d", merts[i])
   }
   fmt.Println("\n\nEquals zero", zeros, "times between 1 and 1000")
   fmt.Println("\nCrosses zero", crosses, "times between 1 and 1000")

}</lang>

Output:
Mertens sequence - First 199 terms:
       1   0  -1  -1  -2  -1  -2  -2  -2  -1  -2  -2  -3  -2  -1  -1  -2  -2  -3
  -3  -2  -1  -2  -2  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1  -1  -2  -1   0
   0  -1  -2  -3  -3  -3  -2  -3  -3  -3  -3  -2  -2  -3  -3  -2  -2  -1   0  -1
  -1  -2  -1  -1  -1   0  -1  -2  -2  -1  -2  -3  -3  -4  -3  -3  -3  -2  -3  -4
  -4  -4  -3  -4  -4  -3  -2  -1  -1  -2  -2  -1  -1   0   1   2   2   1   1   1
   1   0  -1  -2  -2  -3  -2  -3  -3  -4  -5  -4  -4  -5  -6  -5  -5  -5  -4  -3
  -3  -3  -2  -1  -1  -1  -1  -2  -2  -1  -2  -3  -3  -2  -1  -1  -1  -2  -3  -4
  -4  -3  -2  -1  -1   0   1   1   1   0   0  -1  -1  -1  -2  -1  -1  -2  -1   0
   0   1   1   0   0  -1   0  -1  -1  -1  -2  -2  -2  -3  -4  -4  -4  -3  -2  -3
  -3  -4  -5  -4  -4  -3  -4  -3  -3  -3  -4  -5  -5  -6  -5  -6  -6  -7  -7  -8

Equals zero 92 times between 1 and 1000

Crosses zero 59 times between 1 and 1000

Haskell

<lang haskell>import Data.List.Split (chunksOf) import qualified Data.MemoCombinators as Memo import Math.NumberTheory.Primes (unPrime, factorise) import Text.Printf (printf)

moebius :: Integer -> Int moebius = product . fmap m . factorise

 where
   m (p, e)
     | unPrime p == 0 = 0
     | e == 1 = -1
     | otherwise = 0

mertens :: Integer -> Int mertens = Memo.integral (\n -> sum $ fmap moebius [1..n])

countZeros :: [Integer] -> Int countZeros = length . filter ((==0) . mertens)

crossesZero :: [Integer] -> Int crossesZero = length . go . fmap mertens

 where
   go (x:y:xs) 
     | y == 0 && x /= 0 = y : go (y:xs)
     | otherwise        = go (y:xs)
   go _ = []

main :: IO () main = do

 printf "The first 99 terms for M(1..99):\n\n   "
 mapM_ (printf "%3d" . mertens) [1..9] >> printf "\n"
 mapM_ (\row -> mapM_ (printf "%3d" . mertens) row >> printf "\n") $ chunksOf 10 [10..99]
 printf "\nM(n) is zero %d times for 1 <= n <= 1000.\n" $ countZeros [1..1000]
 printf "M(n) crosses zero %d times for 1 <= n <= 1000.\n" $ crossesZero [1..1000]</lang>
Output:
The first 99 terms for M(1..99):

     1  0 -1 -1 -2 -1 -2 -2 -2
 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2
 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3
 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1
 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2
 -2 -1 -1  0  1  2  2  1  1  1

M(n) is zero 92 times for 1 <= n <= 1000.
M(n) crosses zero 59 times for 1 <= n <= 1000.

J

<lang J>mu =: 0:`(1 - 2 * 2|#@{.)@.(1: = */@{:)@(2&p:)"0 M =: +/@([: mu 1:+i.)

m1000 =: (M"0) 1+i.1000 zero =: +/ m1000 = 0 cross =: +/ (-.*.1:|]) m1000 ~: 0

echo 'The first 99 Merten numbers are' echo 10 10$ __, 99{.m1000 echo 'M(N) is zero ',(":zero),' times.' echo 'M(N) crosses zero ',(":cross),' times.' exit</lang>

Output:
The first 99 Merten numbers are
__  1  0 _1 _1 _2 _1 _2 _2 _2
_1 _2 _2 _3 _2 _1 _1 _2 _2 _3
_3 _2 _1 _2 _2 _2 _1 _1 _1 _2
_3 _4 _4 _3 _2 _1 _1 _2 _1  0
 0 _1 _2 _3 _3 _3 _2 _3 _3 _3
_3 _2 _2 _3 _3 _2 _2 _1  0 _1
_1 _2 _1 _1 _1  0 _1 _2 _2 _1
_2 _3 _3 _4 _3 _3 _3 _2 _3 _4
_4 _4 _3 _4 _4 _3 _2 _1 _1 _2
_2 _1 _1  0  1  2  2  1  1  1
M(N) is zero 92 times.
M(N) crosses zero 0 times.

Java

<lang java> public class MertensFunction {

   public static void main(String[] args) {
       System.out.printf("First 199 terms of the merten function are as follows:%n    ");
       for ( int n = 1 ; n < 200 ; n++ ) {
           System.out.printf("%2d  ", mertenFunction(n));
           if ( (n+1) % 20 == 0 ) {
               System.out.printf("%n");
           }
       }
       
       for ( int exponent = 3 ; exponent<= 8 ; exponent++ ) {
           int zeroCount = 0;
           int zeroCrossingCount = 0;
           int positiveCount = 0;
           int negativeCount = 0;
           int mSum = 0;
           int mMin = Integer.MAX_VALUE;
           int mMinIndex = 0;
           int mMax = Integer.MIN_VALUE;
           int mMaxIndex = 0;
           int nMax = (int) Math.pow(10, exponent);
           for ( int n = 1 ; n <= nMax ; n++ ) {
               int m = mertenFunction(n);
               mSum += m;
               if ( m < mMin ) {
                   mMin = m;
                   mMinIndex = n;
               }
               if ( m > mMax ) {
                   mMax = m;
                   mMaxIndex = n;
               }
               if ( m > 0 ) {
                   positiveCount++;
               }
               if ( m < 0 ) {
                   negativeCount++;
               }
               if ( m == 0 ) {
                   zeroCount++;
               }
               if ( m == 0 && mertenFunction(n - 1) != 0 ) {
                   zeroCrossingCount++;
               }
           }
           System.out.printf("%nFor M(x) with x from 1 to %,d%n", nMax);        
           System.out.printf("The maximum of M(x) is M(%,d) = %,d.%n", mMaxIndex, mMax);
           System.out.printf("The minimum of M(x) is M(%,d) = %,d.%n", mMinIndex, mMin);
           System.out.printf("The sum of M(x) is %,d.%n", mSum);
           System.out.printf("The count of positive M(x) is %,d, count of negative M(x) is %,d.%n", positiveCount, negativeCount);
           System.out.printf("M(x) has %,d zeroes in the interval.%n", zeroCount);
           System.out.printf("M(x) has %,d crossings in the interval.%n", zeroCrossingCount);
       }
   }
   
   private static int MU_MAX = 100_000_000;
   private static int[] MU = null;
   private static int[] MERTEN = null;
       
   //  Compute mobius and merten function via sieve
   private static int mertenFunction(int n) {
       if ( MERTEN != null ) {
           return MERTEN[n];
       }
       
       //  Populate array
       MU = new int[MU_MAX+1];
       MERTEN = new int[MU_MAX+1];
       MERTEN[1] = 1;
       int sqrt = (int) Math.sqrt(MU_MAX);
       for ( int i = 0 ; i < MU_MAX ; i++ ) {
           MU[i] = 1;
       }
       
       for ( int i = 2 ; i <= sqrt ; i++ ) {
           if ( MU[i] == 1 ) {
               //  for each factor found, swap + and -
               for ( int j = i ; j <= MU_MAX ; j += i ) {
                   MU[j] *= -i;
               }
               //  square factor = 0
               for ( int j = i*i ; j <= MU_MAX ; j += i*i ) {
                   MU[j] = 0;
               }
           }
       }
       
       int sum = 1;
       for ( int i = 2 ; i <= MU_MAX ; i++ ) {
           if ( MU[i] == i ) {
               MU[i] = 1;
           }
           else if ( MU[i] == -i ) {
               MU[i] = -1;
           }
           else if ( MU[i] < 0 ) {
               MU[i] = 1;               
           }
           else if ( MU[i] > 0 ) {
               MU[i] = -1;
           }
           sum += MU[i];
           MERTEN[i] = sum;
       }
       return MERTEN[n];
   }

} </lang>

Output:
First 199 terms of the merten function are as follows:
     1   0  -1  -1  -2  -1  -2  -2  -2  -1  -2  -2  -3  -2  -1  -1  -2  -2  -3  
-3  -2  -1  -2  -2  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1  -1  -2  -1   0  
 0  -1  -2  -3  -3  -3  -2  -3  -3  -3  -3  -2  -2  -3  -3  -2  -2  -1   0  -1  
-1  -2  -1  -1  -1   0  -1  -2  -2  -1  -2  -3  -3  -4  -3  -3  -3  -2  -3  -4  
-4  -4  -3  -4  -4  -3  -2  -1  -1  -2  -2  -1  -1   0   1   2   2   1   1   1  
 1   0  -1  -2  -2  -3  -2  -3  -3  -4  -5  -4  -4  -5  -6  -5  -5  -5  -4  -3  
-3  -3  -2  -1  -1  -1  -1  -2  -2  -1  -2  -3  -3  -2  -1  -1  -1  -2  -3  -4  
-4  -3  -2  -1  -1   0   1   1   1   0   0  -1  -1  -1  -2  -1  -1  -2  -1   0  
 0   1   1   0   0  -1   0  -1  -1  -1  -2  -2  -2  -3  -4  -4  -4  -3  -2  -3  
-3  -4  -5  -4  -4  -3  -4  -3  -3  -3  -4  -5  -5  -6  -5  -6  -6  -7  -7  -8  

For M(x) with x from 1 to 1,000
The maximum of M(x) is M(586) = 7.
The minimum of M(x) is M(665) = -12.
The sum of M(x) is -1,572.
The count of positive M(x) is 254, count of negative M(x) is 654.
M(x) has 92 zeroes in the interval.
M(x) has 59 crossings in the interval.

For M(x) with x from 1 to 10,000
The maximum of M(x) is M(8,511) = 35.
The minimum of M(x) is M(9,861) = -43.
The sum of M(x) is -20,409.
The count of positive M(x) is 3,965, count of negative M(x) is 5,629.
M(x) has 406 zeroes in the interval.
M(x) has 256 crossings in the interval.

For M(x) with x from 1 to 100,000
The maximum of M(x) is M(48,433) = 96.
The minimum of M(x) is M(96,014) = -132.
The sum of M(x) is -516,879.
The count of positive M(x) is 47,830, count of negative M(x) is 50,621.
M(x) has 1,549 zeroes in the interval.
M(x) has 949 crossings in the interval.

For M(x) with x from 1 to 1,000,000
The maximum of M(x) is M(992,998) = 311.
The minimum of M(x) is M(926,265) = -368.
The sum of M(x) is -14,244,200.
The count of positive M(x) is 472,963, count of negative M(x) is 521,676.
M(x) has 5,361 zeroes in the interval.
M(x) has 3,269 crossings in the interval.

For M(x) with x from 1 to 10,000,000
The maximum of M(x) is M(9,993,034) = 1,143.
The minimum of M(x) is M(7,109,110) = -1,078.
The sum of M(x) is -194,680,528.
The count of positive M(x) is 4,938,188, count of negative M(x) is 5,049,266.
M(x) has 12,546 zeroes in the interval.
M(x) has 7,646 crossings in the interval.

For M(x) with x from 1 to 100,000,000
The maximum of M(x) is M(92,418,127) = 3,290.
The minimum of M(x) is M(76,015,339) = -3,448.
The sum of M(x) is -608,757,258.
The count of positive M(x) is 54,659,906, count of negative M(x) is 45,298,186.
M(x) has 41,908 zeroes in the interval.
M(x) has 25,525 crossings in the interval.

Julia

The OEIS A002321 reference suggests the Mertens function has a negative bias, which it does below 1 million, but this bias seems to switch to a positive bias by 1 billion. There may simply be large swings in the bias overall, which get larger and longer as the sequence continues. <lang julia>using Primes, Formatting

function moebius(n::Integer)

   @assert n > 0
   m(p, e) = p == 0 ? 0 : e == 1 ? -1 : 0
   return reduce(*, m(p, e) for (p, e) in factor(n) if p  ≥ 0; init=1)

end μ(n) = moebius(n)

mertens(x) = sum(n -> μ(n), 1:x) M(x) = mertens(x)

print("First 99 terms of the Mertens function for positive integers:\n ") for n in 1:99

   print(lpad(M(n), 3), n % 10 == 9 ? "\n" : "")

end

function maximinM(N)

   z, cros, lastM, maxi, maxM, mini, minM, sumM, pos, neg = 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
   for i in 1:N
       m = μ(i) + lastM
       if m == 0 && lastM != 0
           cros += 1
       end
       sumM += m
       lastM = m
       if m > maxM
           maxi = i
           maxM = m
       elseif m < minM
           mini = i
           minM = m
       end
       if m > 0
           pos += 1
       elseif m < 0
           neg += 1
       else
           z += 1
       end
   end
   println("\nFor M(x) with x from 1 to $(format(N, commas=true)):")
   println("The maximum of M(x) is M($(format(maxi, commas=true)) = $maxM.")
   println("The minimum of M(x) is M($(format(mini, commas=true))) = $minM.")
   println("The sum of M(x) is $(format(sumM, commas=true)).")
   println("The count of positive M(x) is $(format(pos, commas=true)), count of negative M(x) is $(format(neg, commas=true)).")
   println("M(x) has $(format(z, commas=true)) zeroes in the interval.")
   println("M(x) has $(format(cros, commas=true)) crossings in the interval.")
   diff = pos - neg
   if diff > 0
       println("Positive M(x) exceed negative ones by $(format(diff, commas=true)).")
   else
       println("Negative M(x) exceed positive ones by $(format(-diff, commas=true)).")
   end

end

foreach(maximinM, (1000, 1_000_000, 1_000_000_000))

</lang>

Output:
First 99 terms of the Mertens function for positive integers:
     1  0 -1 -1 -2 -1 -2 -2 -2
 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2
 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3
 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1
 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2
 -2 -1 -1  0  1  2  2  1  1  1

For M(x) with x from 1 to 1,000:
The maximum of M(x) is M(586 = 7.
The minimum of M(x) is M(665) = -12.
The sum of M(x) is -1,572.
The count of positive M(x) is 254, count of negative M(x) is 654.
M(x) has 92 zeroes in the interval.
M(x) has 59 crossings in the interval.
Negative M(x) exceed positive ones by 400.

For M(x) with x from 1 to 1,000,000:
The maximum of M(x) is M(992,998 = 311.
The minimum of M(x) is M(926,265) = -368.
The sum of M(x) is -14,244,200.
The count of positive M(x) is 472,963, count of negative M(x) is 521,676.
M(x) has 5,361 zeroes in the interval.
M(x) has 3,269 crossings in the interval.
Negative M(x) exceed positive ones by 48,713.

For M(x) with x from 1 to 1,000,000,000:
The maximum of M(x) is M(903,087,703 = 10246.
The minimum of M(x) is M(456,877,618) = -8565.
The sum of M(x) is 510,495,361,261.
The count of positive M(x) is 510,200,302, count of negative M(x) is 489,658,577.
M(x) has 141,121 zeroes in the interval.
M(x) has 85,652 crossings in the interval.
Positive M(x) exceed negative ones by 20,541,725.

MAD

<lang MAD> NORMAL MODE IS INTEGER

           DIMENSION M(1000)
                 
           M(1) = 1
           THROUGH GENMRT, FOR N=2, 1, N.G.1000
           M(N) = 1
           THROUGH GENMRT, FOR K=2, 1, K.G.N

GENMRT M(N) = M(N) - M(N/K)

           PRINT COMMENT $ FIRST 99 MERTEN NUMBERS ARE$
           VECTOR VALUES F9 = $S3,9(I2,S1)*$
           VECTOR VALUES F10 = $10(I2,S1)*$
           
           PRINT FORMAT F9, M(1), M(2), M(3), M(4), M(5), M(6),
         0                  M(7), M(8), M(9)
         
           THROUGH SHOW, FOR N=10, 10, N.GE.100

SHOW PRINT FORMAT F10, M(N), M(N+1), M(N+2), M(N+3), M(N+4),

         0   M(N+5), M(N+6), M(N+7), M(N+8), M(N+9), M(N+10)
         
           ZERO = 0
           CROSS = 0
           THROUGH ZC, FOR N=1, 1, N.G.1000
           WHENEVER M(N).E.0, ZERO = ZERO + 1

ZC WHENEVER M(N).E.0 .AND. M(N-1).NE.0, CROSS = CROSS + 1

           VECTOR VALUES FZ = $13HM(N) IS ZERO ,I2,S1,5HTIMES*$
           PRINT FORMAT FZ, ZERO
           
           VECTOR VALUES FC = $18HM(N) CROSSES ZERO ,I2,S1,5HTIMES*$
           PRINT FORMAT FC, CROSS
           
           END OF PROGRAM </lang>
Output:
FIRST 99 MERTEN NUMBERS ARE
    1  0 -1 -1 -2 -1 -2 -2 -2
-1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2
-3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3
-3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1
-2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2
-2 -1 -1  0  1  2  2  1  1  1
M(N) IS ZERO 92 TIMES
M(N) CROSSES ZERO 59 TIMES

Pascal

Works with: Free Pascal

Nearly the same as Square-free_integers#Pascal Instead here marking all multiples, starting at factor 2, of a prime by incrementing the factor count.
runtime ~log(n)*n <lang pascal>program Merten; {$IFDEF FPC}

 {$MODE DELPHI}
 {$Optimization ON,ALL}

{$ELSE}

  {$APPTYPE CONSOLE}

{$ENDIF} uses

 sysutils;

const

 BigLimit = 10*1000*1000*1000;//1e10

type

 tSieveElement = Int8;
 tpSieve = pInt8;
 tMoebVal = array[-1..1] of Int64;

var

 MertensValues : array[-40000..50500] of NativeInt;
 primes : array of byte;
 sieve : array of tSieveElement;

procedure CompactPrimes; //searching for needed primes //last primes are marked with -1 var

 pSieve : tpSieve;
 i,lmt,dp:NativeInt;

Begin

 setlength(Primes,74500);//suffices for primes to calc square upto 1e12
 //extract difference of primes
 i := 2;
 lmt := 0;
 dp := 2;
 pSieve :=@sieve[0];
 repeat
   IF pSieve[i]= 0 then
   Begin
     //mark for Moebius
     pSieve[i]:= -1;
     primes[lmt] := dp;
     dp := 0;
     inc(lmt);
   end;
   inc(dp);
   inc(i);
 until i*i >BigLimit;
 setlength(Primes,lmt+1);
 repeat
   IF pSieve[i]= 0 then
     //mark for Moebius
     pSieve[i]:= -1;
   inc(i);
 until i >BigLimit;

end;

procedure SieveSquares; //mark all powers >=2 of prime => all powers = 2 is sufficient var

 pSieve : tpSieve;
 i,sq,k,prime : NativeInt;

Begin

 pSieve := @sieve[0];
 prime := 0;
 For i := 0 to High(primes) do
 Begin
   prime := prime+primes[i];
   sq := prime*prime;
   k := sq;
   if sq > BigLimit then
     break;
   repeat
     pSieve[k] := 0;
     inc(k,sq);
   until k> BigLimit;
 end;

end;

procedure initPrimes; var

 pSieve : tpSieve;
 fakt,
 sieveprime : NativeUint;

begin

 pSieve := @sieve[0];
 sieveprime := 2;
 repeat
   if pSieve[sieveprime]=0 then
   begin
     fakt := sieveprime+sieveprime;
     while fakt <=BigLimit do
     Begin
       //count divisors
       inc(pSieve[fakt]);
       inc(fakt,sieveprime);
     end;
   end;
   inc(sieveprime);
 until sieveprime>BigLimit DIV 2;
 //Möbius of 1
 pSieve[1] := 1;
 //convert to Moebius
 For fakt := 2 to BigLimit do
 Begin
   sieveprime := pSieve[fakt];
   IF sieveprime<>0 then
     pSieve[fakt] := 1-(2*(sieveprime AND 1)) ;
 end;
 CompactPrimes;
 SieveSquares;

end;

procedure OutMerten10(Lmt,ZeroCross:NativeInt;Const MoebVal:tMoebVal); var

 i,j: NativeInt;

Begin

 Writeln(lmt:11,MoebVal[-1]:11,MoebVal[1]:11,MoebVal[-1]+MoebVal[1]:11,
 MoebVal[-1]-MoebVal[1]:7,MoebVal[0]:11);
 i:= low(MertensValues);
 while MertensValues[i] = 0 do
   inc(i);
 j:= High(MertensValues);
 while MertensValues[j] = 0 do
   dec(j);
 write('Merten min ',i:6,' max ',j:6,' zeros ',MertensValues[0]:8);
 writeln(' zeroCross ',ZeroCross);
 writeln;

end;

procedure Count_x10; var

 MoebCount: tMoebVal;
 pSieve : tpSieve;
 i,lmt,Merten,Moebius,LastMert,ZeroCross: NativeInt;

begin

 writeln('[1 to limit]');
 Writeln('Limit        Moeb. odd   Moeb.even  sqr-free Merten     Zeros');
 pSieve := @sieve[0];
 For i := -1 to 1 do
   MoebCount[i]:=0;
 ZeroCross := 0;
 LastMert :=1;
 Merten :=0;
 lmt := 10;
 i := 1;
 repeat
   while i <= lmt do
   Begin
     Moebius := pSieve[i];
     inc(MoebCount[Moebius]);
     inc(Merten,Moebius);
     inc(MertensValues[Merten]);//MoebCount[1]-MoebCount[-1]]);
     inc(ZeroCross,ORD( (Merten = 0) AND (LastMert <> 0)));
     LastMert := Merten;
     inc(i);
   end;
   OutMerten10(Lmt,ZeroCross,MoebCount);
   IF lmt >= BigLimit then
     BREAK;
   lmt := lmt*10;
   IF lmt >BigLimit then
     lmt := BigLimit;
 until false;
 writeln;

end;

procedure OutMerten(lmt:NativeInt); var

 i,k,m : NativeInt;

Begin

 iF lmt> BigLimit then
   lmt := BigLimit;
 writeln('Mertens numbers from 1 to ',lmt);
 k := 9;
 write(:3);
 m := 0;
 For i := 1 to lmt do
 Begin
   inc(m,sieve[i]);
   write(m:3);
   dec(k);
   IF k = 0 then
   Begin
     writeln;
     k := 10;
   end;
 end;
 writeln;

end;

procedure OutMoebius(lmt:NativeInt); var

 i,k : NativeInt;

Begin

 iF lmt> BigLimit then
   lmt := BigLimit;
 writeln('Möbius numbers from 1 to ',lmt);
 k := 19;
 write(:3);
 For i := 1 to lmt do
 Begin
   write(sieve[i]:3);
   dec(k);
   IF k = 0 then
   Begin
     writeln;
     k := 20;
   end;
 end;
 writeln;

end;

Begin

 setlength(sieve,BigLimit+1);
 InitPrimes;
 SieveSquares;
 Count_x10;
 OutMoebius(199);
 OutMerten(99);
 setlength(primes,0);
 setlength(sieve,0);

end.</lang>

Output:
[1 to limit]
Limit        Moeb. odd   Moeb.even  sqr-free Merten     Zero's
         10          4          3          7      1          3
Merten min     -2 max      1 zero's        1 zeroCross 1

        100         30         31         61     -1         39
Merten min     -4 max      2 zero's        6 zeroCross 5

       1000        303        305        608     -2        392
Merten min    -12 max      7 zero's       92 zeroCross 59

      10000       3053       3030       6083     23       3917
Merten min    -43 max     35 zero's      406 zeroCross 256

     100000      30421      30373      60794     48      39206
Merten min   -132 max     96 zero's     1549 zeroCross 949

    1000000     303857     304069     607926   -212     392074
Merten min   -368 max    311 zero's     5361 zeroCross 3269

   10000000    3039127    3040164    6079291  -1037    3920709
Merten min  -1078 max   1143 zero's    12546 zeroCross 7646

  100000000   30395383   30397311   60792694  -1928   39207306
Merten min  -3448 max   3290 zero's    41908 zeroCross 25525

 1000000000  303963673  303963451  607927124    222  392072876
Merten min  -8565 max  10246 zero's   141121 zeroCross 85652

10000000000 3039652332 3039618610 6079270942  33722 3920729058
Merten min -35517 max  50286 zero's   431822 zeroCross 262605


Möbius numbers from 1 to 199
     1 -1 -1  0 -1  1 -1  0  0  1 -1  0 -1  1  1  0 -1  0 -1
  0  1  1 -1  0  0  1  0  0 -1 -1 -1  0  1  1  1  0 -1  1  1
  0 -1 -1 -1  0  0  1 -1  0  0  0  1  0 -1  0  1  0  1  1 -1
  0 -1  1  0  0  1 -1 -1  0  1 -1 -1  0 -1  1  0  0  1 -1 -1
  0  0  1 -1  0  1  1  1  0 -1  0  1  0  1  1  1  0 -1  0  0
  0 -1 -1 -1  0 -1  1 -1  0 -1 -1  1  0 -1 -1  1  0  0  1  1
  0  0  1  1  0  0  0 -1  0  1 -1 -1  0  1  1  0  0 -1 -1 -1
  0  1  1  1  0  1  1  0  0 -1  0 -1  0  0 -1  1  0 -1  1  1
  0  1  0 -1  0 -1  1 -1  0  0 -1  0  0 -1 -1  0  0  1  1 -1
  0 -1 -1  1  0  1 -1  1  0  0 -1 -1  0 -1  1 -1  0 -1  0 -1

Mertens numbers from 1 to 99
     1  0 -1 -1 -2 -1 -2 -2 -2
 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2
 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3
 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1
 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2
 -2 -1 -1  0  1  2  2  1  1  1

real    3m54,249s = 234s //BigLimit = 100*1000*1000;takes 2.017s

Perl

<lang perl>use utf8; use strict; use warnings; use feature 'say'; use List::Util 'uniq';

sub prime_factors {

   my ($n, $d, @factors) = (shift, 1);
   while ($n > 1 and $d++) {
       $n /= $d, push @factors, $d until $n % $d;
   }
   @factors

}

sub μ {

   my @p = prime_factors(shift);
   @p == uniq(@p) ? 0 == @p%2 ? 1 : -1 : 0

}

sub progressive_sum {

   my @sum = shift @_;
   push @sum, $sum[-1] + $_ for @_;
   @sum

}

my($upto, $show, @möebius) = (1000, 199, ()); push @möebius, μ($_) for 1..$upto; my @mertens = progressive_sum @möebius;

say "Mertens sequence - First $show terms:\n" .

   (' 'x4 . sprintf "@{['%4d' x $show]}", @mertens[0..$show-1]) =~ s/((.){80})/$1\n/gr .
   sprintf("\nEquals zero %3d times between 1 and $upto", scalar grep { ! $_ } @mertens) .
   sprintf "\nCrosses zero%3d times between 1 and $upto", scalar grep { ! $mertens[$_-1] and $mertens[$_] } 1 .. @mertens;</lang>
Output:
Mertens sequence - First 199 terms:
       1   0  -1  -1  -2  -1  -2  -2  -2  -1  -2  -2  -3  -2  -1  -1  -2  -2  -3
  -3  -2  -1  -2  -2  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1  -1  -2  -1   0
   0  -1  -2  -3  -3  -3  -2  -3  -3  -3  -3  -2  -2  -3  -3  -2  -2  -1   0  -1
  -1  -2  -1  -1  -1   0  -1  -2  -2  -1  -2  -3  -3  -4  -3  -3  -3  -2  -3  -4
  -4  -4  -3  -4  -4  -3  -2  -1  -1  -2  -2  -1  -1   0   1   2   2   1   1   1
   1   0  -1  -2  -2  -3  -2  -3  -3  -4  -5  -4  -4  -5  -6  -5  -5  -5  -4  -3
  -3  -3  -2  -1  -1  -1  -1  -2  -2  -1  -2  -3  -3  -2  -1  -1  -1  -2  -3  -4
  -4  -3  -2  -1  -1   0   1   1   1   0   0  -1  -1  -1  -2  -1  -1  -2  -1   0
   0   1   1   0   0  -1   0  -1  -1  -1  -2  -2  -2  -3  -4  -4  -4  -3  -2  -3
  -3  -4  -5  -4  -4  -3  -4  -3  -3  -3  -4  -5  -5  -6  -5  -6  -6  -7  -7  -8

Equals zero  92 times between 1 and 1000
Crosses zero 59 times between 1 and 1000

Phix

Based on the stackexchange link, short and sweet but not very fast: 1.4s just for the first 1000... <lang Phix>function Mertens(integer n)

   integer res = 1
   for k=2 to n do
       res -= Mertens(floor(n/k))
   end for
   return res

end function sequence s = {" ."} for i=1 to 143 do s = append(s,sprintf("%3d",Mertens(i))) end for puts(1,join_by(s,1,12," "))

integer prev = 1, zeroes = 0, crosses = 0 for n=2 to 1000 do

   integer m = Mertens(n)
   if m=0 then
       zeroes += 1
       crosses += prev!=0 
   end if
   prev = m

end for printf(1,"\nMertens[1..1000] equals zero %d times and crosses zero %d times\n",{zeroes,crosses})</lang>

Output:

Matches the wp table:

  .   1   0  -1  -1  -2  -1  -2  -2  -2  -1  -2
 -2  -3  -2  -1  -1  -2  -2  -3  -3  -2  -1  -2
 -2  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1
 -1  -2  -1   0   0  -1  -2  -3  -3  -3  -2  -3
 -3  -3  -3  -2  -2  -3  -3  -2  -2  -1   0  -1
 -1  -2  -1  -1  -1   0  -1  -2  -2  -1  -2  -3
 -3  -4  -3  -3  -3  -2  -3  -4  -4  -4  -3  -4
 -4  -3  -2  -1  -1  -2  -2  -1  -1   0   1   2
  2   1   1   1   1   0  -1  -2  -2  -3  -2  -3
 -3  -4  -5  -4  -4  -5  -6  -5  -5  -5  -4  -3
 -3  -3  -2  -1  -1  -1  -1  -2  -2  -1  -2  -3
 -3  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1

Mertens[1..1000] equals zero 92 times and crosses zero 59 times

Prolog

Works with: SWI Prolog

<lang prolog>:- dynamic mertens_number_cache/2.

mertens_number(1, 1):- !. mertens_number(N, M):-

   mertens_number_cache(N, M),
   !.

mertens_number(N, M):-

   N >= 2,
   mertens_number(N, 2, M, 0),
   assertz(mertens_number_cache(N, M)).

mertens_number(N, N, M, M):- !. mertens_number(N, K, M, S):-

   N1 is N // K,
   mertens_number(N1, M1),
   K1 is K + 1,
   S1 is S - M1,
   mertens_number(N, K1, M, S1).

print_mertens_numbers(Count):-

   print_mertens_numbers(Count, 0).

print_mertens_numbers(Count, Count):-!. print_mertens_numbers(Count, N):-

   (N == 0 ->
       write('   ')
       ;
       mertens_number(N, M),
       writef('%3r', [M])
   ),
   N1 is N + 1,
   Column is N1 mod 20,
   (N > 0, Column == 0 ->
       nl
       ;
       true
   ),
   print_mertens_numbers(Count, N1).

count_zeros(From, To, Z, C):-

   count_zeros(From, To, Z, C, 0, 0, 0).

count_zeros(From, To, Z, C, Z, C, _):-

   From > To,
   !.

count_zeros(From, To, Z, C, Z1, C1, P):-

   mertens_number(From, M),
   (M == 0 -> Z2 is Z1 + 1 ; Z2 = Z1),
   (M == 0, P \= 0 -> C2 is C1 + 1 ; C2 = C1),
   Next is From + 1,
   count_zeros(Next, To, Z, C, Z2, C2, M).

main:-

   writeln('First 199 Mertens numbers:'),
   print_mertens_numbers(200),
   count_zeros(1, 1000, Z, C),
   writef('M(n) is zero %t times for 1 <= n <= 1000.\n', [Z]),
   writef('M(n) crosses zero %t times for 1 <= n <= 1000.\n', [C]).</lang>
Output:
First 199 Mertens numbers:
     1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
 -3 -2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
  0 -1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
 -1 -2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
 -4 -4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1
  1  0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3
 -3 -3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4
 -4 -3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0
  0  1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3
 -3 -4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8
M(n) is zero 92 times for 1 <= n <= 1000.
M(n) crosses zero 59 times for 1 <= n <= 1000.

Raku

(formerly Perl 6)

Works with: Rakudo version 2019.11

Mertens number is not defined for n == 0. Raku arrays are indexed from 0 so store a blank value at position zero to keep x and M(x) aligned.

<lang perl6>use Prime::Factor;

sub μ (Int \n) {

   return 0 if n %% 4 or n %% 9 or n %% 25 or n %% 49 or n %% 121;
   my @p = prime-factors(n);
   +@p == +@p.unique ?? +@p %% 2 ?? 1 !! -1 !! 0

}

my @mertens = lazy [\+] flat , 1, (2..*).hyper.map: -> \n { μ(n) };

put "Mertens sequence - First 199 terms:\n",

   @mertens[^200]».fmt('%3s').batch(20).join("\n"),
   "\n\nEquals zero ", +@mertens[1..1000].grep( !* ),
   ' times between 1 and 1000', "\n\nCrosses zero ",
   +@mertens[1..1000].kv.grep( {!$^v and @mertens[$^k]} ),
   " times between 1 and 1000\n\nFirst Mertens equal to:";

for 10, 20, 30 … 100 -> $threshold {

   printf "%4d: M(%d)\n", -$threshold, @mertens.first: * == -$threshold, :k;
   printf "%4d: M(%d)\n",  $threshold, @mertens.first: * ==  $threshold, :k;

}</lang>

Output:
Mertens sequence - First 199 terms:
      1   0  -1  -1  -2  -1  -2  -2  -2  -1  -2  -2  -3  -2  -1  -1  -2  -2  -3
 -3  -2  -1  -2  -2  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1  -1  -2  -1   0
  0  -1  -2  -3  -3  -3  -2  -3  -3  -3  -3  -2  -2  -3  -3  -2  -2  -1   0  -1
 -1  -2  -1  -1  -1   0  -1  -2  -2  -1  -2  -3  -3  -4  -3  -3  -3  -2  -3  -4
 -4  -4  -3  -4  -4  -3  -2  -1  -1  -2  -2  -1  -1   0   1   2   2   1   1   1
  1   0  -1  -2  -2  -3  -2  -3  -3  -4  -5  -4  -4  -5  -6  -5  -5  -5  -4  -3
 -3  -3  -2  -1  -1  -1  -1  -2  -2  -1  -2  -3  -3  -2  -1  -1  -1  -2  -3  -4
 -4  -3  -2  -1  -1   0   1   1   1   0   0  -1  -1  -1  -2  -1  -1  -2  -1   0
  0   1   1   0   0  -1   0  -1  -1  -1  -2  -2  -2  -3  -4  -4  -4  -3  -2  -3
 -3  -4  -5  -4  -4  -3  -4  -3  -3  -3  -4  -5  -5  -6  -5  -6  -6  -7  -7  -8

Equals zero 92 times between 1 and 1000

Crosses zero 59 times between 1 and 1000

First Mertens equal to:
 -10: M(659)
  10: M(1393)
 -20: M(2791)
  20: M(3277)
 -30: M(9717)
  30: M(8503)
 -40: M(9831)
  40: M(11770)
 -50: M(24018)
  50: M(19119)
 -60: M(24105)
  60: M(31841)
 -70: M(24170)
  70: M(31962)
 -80: M(42789)
  80: M(48202)
 -90: M(59026)
  90: M(48405)
-100: M(59426)
 100: M(114717)

REXX

The   Mertens   function is named after Franz Mertens.

Programming note:   This REXX version supports the specifying of the low and high values to be generated,
as well as the "group" size for the grid   (it can be specified as   1   which will show a vertical list).

A null value will be shown as a bullet (•) when showing the Möbius and/or Mertens value of for zero   (which can be changed easily).

The above "feature" was added to make the grid to be aligned with other solutions. <lang rexx>/*REXX pgm computes & shows a value grid of the Mertens function for a range of integers*/ parse arg LO HI grp eqZ xZ . /*obtain optional arguments from the CL*/ if LO== | LO=="," then LO= 0 /*Not specified? Then use the default.*/ if HI== | HI=="," then HI= 199 /* " " " " " " */ if grp== | grp=="," then grp= 20 /* " " " " " " */ if eqZ== | eqZ=="," then eqZ= 1000 /* " " " " " " */ if xZ== | xZ=="," then xZ= 1000 /* " " " " " " */ call genP /*generate primes up to max √ HIHI */

              call Franz LO, HI

if eqZ>0 then call Franz 1,-eqZ if xZ>0 then call Franz -1, xZ exit 0 /*stick a fork in it, we're all done. */ /*──────────────────────────────────────────────────────────────────────────────────────*/ Franz: parse arg a 1 oa,b 1 ob; @Mertens=' The Mertens sequence from ' a= abs(a); b= abs(b); grid= oa>=0 & ob>=0 /*semaphore used to show a grid title. */ if grid then say center(@Mertens LO " ──► " HI" ", max(50, grp*3), '═') /*show title*/

        else say

zeros= 0 /*# of 0's found for Mertens function.*/ Xzero= 0 /*number of times that zero was crossed*/ prev= $= /*$ holds output grid of GRP numbers. */

  do j=a  to b;     _= Mertens(j)               /*process some numbers from  LO ──► HI.*/
  if _==0  then zeros= zeros + 1                /*Is Zero?  Then bump the zeros counter*/
  if _==0  then if prev\==0 then Xzero= Xzero+1 /*prev ¬=0?   "   "    "  Xzero    "   */
  prev= _
  if grid  then $= $ right(_, 2)                /*build grid if A & B are non─negative.*/
  if words($)==grp  then do;  say substr($, 2);  $=    /*show grid if fully populated,*/
                         end                           /*  and nullify it for more #s.*/
  end   /*j*/                                   /*for small grids, using wordCnt is OK.*/

if $\== then say substr($, 2) /*handle any residual numbers not shown*/ if oa<0 then say @Mertens a " to " b ' has crossed zero ' Xzero " times." if ob<0 then say @Mertens a " to " b ' has ' zeros " zeros." return /*──────────────────────────────────────────────────────────────────────────────────────*/ Mertens: procedure expose @. !. M.; parse arg n /*obtain a integer to be tested for mu.*/

        if M.n\==.  then return M.n             /*was computed before?  Then return it.*/
        if n<1      then return '∙'             /*handle special cases of non─positive#*/
        m= 0                                    /*the sum of all the  MU's  (so far).  */
             do k=1  for n;   m= m + mobius(k)  /*sum the  MU's  up to  N.             */
             end   /*k*/                        /* [↑] mobius function uses memoization*/
        M.n= m;          return m               /*return the sum of all the  MU's.     */

/*──────────────────────────────────────────────────────────────────────────────────────*/ mobius: procedure expose @. !.; parse arg x 1 ox /*obtain a integer to be tested for mu.*/

       if !.x\==.  then return !.x              /*X computed before?  Return that value*/
       if x<1      then return '∙'              /*handle special case of non-positive #*/
       #= 0                                     /*start with a  mu  value of zero.     */
            do k=1;  p= @.k                     /*get the  Kth  (pre─generated)  prime.*/
            if p>x  then leave                  /*prime (P)   > X?    Then we're done. */
            if p*p>x  then do;   #= #+1;  leave /*prime (P**2 > X?    Bump # and leave.*/
                           end
            if x//p==0  then do; #= #+1         /*X divisible by P?   Bump mu number.  */
                                 x= x % p       /*                    Divide by prime. */
                                 if x//p==0  then return 0  /*X÷by P?  Then return zero*/
                             end
            end   /*k*/                         /*#  (below) is almost always small, <9*/
       !.ox=  -1 ** #;        return !.ox       /*raise -1 to the mu power, memoize it.*/

/*──────────────────────────────────────────────────────────────────────────────────────*/ genP: !.=.; M.=!.; hihi= max(HI, eqZ, xZ) /*initialize 2 arrays for memoization. */

     @.1=2;  @.2=3; @.3=5; @.4=7; @.5=11; @.6= 13; nP=6 /*assign low primes; # primes. */
      do lim=nP  until lim*lim>=hihi;  end      /*only keep primes up to the sqrt(HI). */
      do j=@.nP+4  by 2  to hihi                /*only find odd primes from here on.   */
         do k=3  while k*k<=j                   /*divide by some generated odd primes. */
         if j // @.k==0  then iterate j         /*Is J divisible by  P?  Then not prime*/
         end   /*k*/                            /* [↓]  a prime  (J)  has been found.  */
      nP= nP+1;  if nP<=HI  then @.nP=j         /*bump prime count; assign prime to  @.*/
      end      /*j*/;            return</lang>
output   when using the default inputs:

Output note:   note the use of a bullet (•) to signify that a "null" is being shown (for the 0th entry).

══════════ The Mertens sequence from  0  ──►  199 ══════════
 ∙  1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3
-3 -2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0
 0 -1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1
-1 -2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4
-4 -4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1
 1  0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3
-3 -3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4
-4 -3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0
 0  1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3
-3 -4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8

 The Mertens sequence from  1  to  1000  has  92  zeros.

 The Mertens sequence from  1  to  1000  has crossed zero  59  times.

Sidef

Built-in: <lang ruby>say mertens(123456789) #=> 1170 say mertens(1234567890) #=> 9163</lang>

Algorithm for computing M(n) in sublinear time:

<lang ruby>func mertens(n) is cached {

   var lookup_size    = (2 * n.iroot(3)**2)
   var mertens_lookup = [0]
   for k in (1..lookup_size) {
       mertens_lookup[k] = (mertens_lookup[k-1] + k.moebius)
   }
   static cache = Hash()
   func (n) {
       if (n <= lookup_size) {
           return mertens_lookup[n]
       }
       if (cache.has(n)) {
           return cache{n}
       }
       var M = 1
       var s = n.isqrt
       for k in (2 .. floor(n/(s+1))) {
           M -= __FUNC__(floor(n/k))
       }
       for k in (1..s) {
           M -= (mertens_lookup[k] * (floor(n/k) - floor(n/(k+1))))
       }
       cache{n} = M
   }(n)

}</lang>

Task: <lang ruby>with (200) {|n|

   say "Mertens function in the range 1..#{n}:"
   (1..n).map { mertens(_) }.slices(20).each {|line|
       say line.map{ "%2s" % _ }.join(' ')
   }

}

with (1000) {|n|

   say "\nIn the range 1..#{n}, there are:"
   say (1..n->count_by { mertens(_)==0 }, " zeros")
   say (1..n->count_by { mertens(_)==0 && mertens(_-1)!=0 }, " zero crossings")

}</lang>

Output:
Mertens function in the range 1..200:
 1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3 -3
-2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0  0
-1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1 -1
-2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4 -4
-4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1  1
 0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3 -3
-3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4 -4
-3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0  0
 1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3 -3
-4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8 -8

In the range 1..1000, there are:
92 zeros
59 zero crossings

Wren

Library: Wren-fmt
Library: Wren-math

<lang ecmascript>import "/fmt" for Fmt import "/math" for Int

var isSquareFree = Fn.new { |n|

   var i = 2
   while (i * i <= n) {
       if (n%(i*i) == 0) return false
       i = (i > 2) ? i + 2 : i + 1
   }
   return true

}

var mu = Fn.new { |n|

   if (n < 1) Fiber.abort("Argument must be a positive integer")
   if (n == 1) return 1
   var sqFree = isSquareFree.call(n)
   var factors = Int.primeFactors(n)
   if (sqFree && factors.count % 2 == 0) return 1
   if (sqFree) return -1
   return 0

}

var M = Fn.new { |x| (1..x).reduce { |sum, n| sum + mu.call(n) } }

System.print("The first 199 Mertens numbers are:") for (i in 0..9) {

   for (j in 0..19) {
       if (i == 0 && j == 0) {
           System.write("    ")
       } else {
           System.write("%(Fmt.dm(3, M.call(i*20 + j))) ")
       }
   }
   System.print()

}

// use the recurrence relationship for the last 2 parts rather than calling M directly var count = 0 var mertens = M.call(1) for (i in 2..1000) {

   mertens = mertens + mu.call(i)
   if (mertens == 0) count = count + 1

} System.print("\nThe Mertens function is zero %(count) times in the range [1, 1000].")

count = 0 var prev = M.call(1) for (i in 2..1000) {

   var next = prev + mu.call(i)
   if (next == 0 && prev != 0) count = count + 1
   prev = next

} System.print("\nThe Mertens function crosses zero %(count) times in the range [1, 1000].")</lang>

Output:
The first 199 Mertens numbers are:
      1   0  -1  -1  -2  -1  -2  -2  -2  -1  -2  -2  -3  -2  -1  -1  -2  -2  -3 
 -3  -2  -1  -2  -2  -2  -1  -1  -1  -2  -3  -4  -4  -3  -2  -1  -1  -2  -1   0 
  0  -1  -2  -3  -3  -3  -2  -3  -3  -3  -3  -2  -2  -3  -3  -2  -2  -1   0  -1 
 -1  -2  -1  -1  -1   0  -1  -2  -2  -1  -2  -3  -3  -4  -3  -3  -3  -2  -3  -4 
 -4  -4  -3  -4  -4  -3  -2  -1  -1  -2  -2  -1  -1   0   1   2   2   1   1   1 
  1   0  -1  -2  -2  -3  -2  -3  -3  -4  -5  -4  -4  -5  -6  -5  -5  -5  -4  -3 
 -3  -3  -2  -1  -1  -1  -1  -2  -2  -1  -2  -3  -3  -2  -1  -1  -1  -2  -3  -4 
 -4  -3  -2  -1  -1   0   1   1   1   0   0  -1  -1  -1  -2  -1  -1  -2  -1   0 
  0   1   1   0   0  -1   0  -1  -1  -1  -2  -2  -2  -3  -4  -4  -4  -3  -2  -3 
 -3  -4  -5  -4  -4  -3  -4  -3  -3  -3  -4  -5  -5  -6  -5  -6  -6  -7  -7  -8 

The Mertens function is zero 92 times in the range [1, 1000].

The Mertens function crosses zero 59 times in the range [1, 1000].

zkl

<lang zkl>fcn mertensW(n){

  [1..].tweak(fcn(n,pm){
     pm.incN(mobius(n));
     pm.value
  }.fp1(Ref(0)))

} fcn mobius(n){

  pf:=primeFactors(n);
  sq:=pf.filter1('wrap(f){ (n % (f*f))==0 });  // False if square free
  if(sq==False){ if(pf.len().isEven) 1 else -1 }
  else 0

} fcn primeFactors(n){ // Return a list of prime factors of n

  acc:=fcn(n,k,acc,maxD){  // k is 2,3,5,7,9,... not optimum
     if(n==1 or k>maxD) acc.close();
     else{

q,r:=n.divr(k); // divr-->(quotient,remainder) if(r==0) return(self.fcn(q,k,acc.write(k),q.toFloat().sqrt())); return(self.fcn(n,k+1+k.isOdd,acc,maxD)) # both are tail recursion

     }
  }(n,2,Sink(List),n.toFloat().sqrt());
  m:=acc.reduce('*,1);      // mulitply factors
  if(n!=m) acc.append(n/m); // opps, missed last factor
  else acc;

}</lang> <lang zkl>mertensW().walk(199) .pump(Console.println, T(Void.Read,19,False), fcn{ vm.arglist.pump(String,"%3d".fmt) });

println("\nIn the first 1,000 terms of the Mertens sequence there are:"); otm:=mertensW().pump(1_000,List); otm.reduce(fcn(s,m){ s + (m==0) },0) : println(_," zeros"); otm.reduce(fcn(p,m,rs){ rs.incN(m==0 and p!=0); m }.fp2( s:=Ref(0) )); println(s.value," zero crossings");</lang>

Output:
  1  0 -1 -1 -2 -1 -2 -2 -2 -1 -2 -2 -3 -2 -1 -1 -2 -2 -3 -3
 -2 -1 -2 -2 -2 -1 -1 -1 -2 -3 -4 -4 -3 -2 -1 -1 -2 -1  0  0
 -1 -2 -3 -3 -3 -2 -3 -3 -3 -3 -2 -2 -3 -3 -2 -2 -1  0 -1 -1
 -2 -1 -1 -1  0 -1 -2 -2 -1 -2 -3 -3 -4 -3 -3 -3 -2 -3 -4 -4
 -4 -3 -4 -4 -3 -2 -1 -1 -2 -2 -1 -1  0  1  2  2  1  1  1  1
  0 -1 -2 -2 -3 -2 -3 -3 -4 -5 -4 -4 -5 -6 -5 -5 -5 -4 -3 -3
 -3 -2 -1 -1 -1 -1 -2 -2 -1 -2 -3 -3 -2 -1 -1 -1 -2 -3 -4 -4
 -3 -2 -1 -1  0  1  1  1  0  0 -1 -1 -1 -2 -1 -1 -2 -1  0  0
  1  1  0  0 -1  0 -1 -1 -1 -2 -2 -2 -3 -4 -4 -4 -3 -2 -3 -3
 -4 -5 -4 -4 -3 -4 -3 -3 -3 -4 -5 -5 -6 -5 -6 -6 -7 -7 -8

In the first 1,000 terms of the Mertens sequence there are:
92 zeros
59 zero crossings