Help with OLD code

[juanirias]

Hi, I need to compile a math unit, but I'm having trouble.

The unit NUMLIB code is here:
https://github.com/alrieckert/freepascal/tree/master/packages/numlib

And the function is "determinant", but I'm not getting it to work.
https://wiki.lazarus.freepascal.org/NumLib_Documentation#Determinant_of_a_matrix_(unit_det)


Ragards, JUAN

[Thaddy]

numlib is a standard fpc code library, unless there is a naming conflict, so will port without issue if it is the same "numlib".
It is in .\packages\numlib
It is also maintained. (when necessary)

Don't use third party links when it is a standard package, always use the package FPC provides.....

Since you are using the same numlib, only much older, if it does not port when using the correct one, plz report back.
The author, marcov, is a core developer and will help you in that case.

In a standard installation numlib compiles out-of-the-box and is in the path.
To my knowledge there are no interface changes apart for 32->64 bit.

[juanirias]

I cant run well

Code: Pascal  [Select][+][-]

1. program determinant_spd_matrix;   // "spd" = symmetric positive definite
2.  
3. uses
4.   numlib;
5.  
6. const
7.   n = 4;
8.  
9. const
10.   a: array[1..n, 1..n] of ArbFloat = (
11.     ( 4,  2,  4,  1)
12.     (30, 20, 45, 12)
13.     (20, 15, 36, 10)
14.     (35, 28, 70, 20)
15.   );
16.  
17. var
18.   f: ArbFloat;
19.   k: Integer;
20.   term: Integer = 0;
21.   i, j: Integer;
22.  
23. begin
24.   // write input matrix
25.   WriteLn('A = ');
26.   for i := 1 to n do begin
27.     for j := 1 to n do
28.       Write(a[i, j]:10:0);
29.     WriteLn;
30.   end;
31.   WriteLn;
32.  
33.   // Calculate determinant of symmetric positive definite matrix
34.   detgpd(n, n, a[1, 1], f, k, term);
35.  
36.   // or uncomment to use one of the other methods
37.   // detgen(n, n, a[1, 1], f, k, term);
38.   // detgsy(n, n, a[1, 1], f, k, term);
39.  
40.   if term = 1 then
41.     WriteLn('Determinant of A = ', f * intpower(8, k):0:9, ' (k = ', k, ')')
42.   else
43.     WriteLn('Incorrect input');
44. end.
45.  
46.  

determinant_spd_matrix.pas(15,4) Fatal: Syntax error, "," expected but ";" found

[Thaddy]

Corrected code, Your code would never have worked but this code does:

Code: Pascal  [Select][+][-]

1. program determinant_spd_matrix;   // "spd" = symmetric positive definite
2. {$mode delphi}
3. uses
4.   math, typ, numlib; //  2/3 forgotten???
5.  
6. const
7.   n = 4;
8.  
9. const
10.   // here all comma's forgotten???
11.   a: array[1..n, 1..n] of ArbFloat = (
12.     ( 4,  2,  4,  1),
13.     (30, 20, 45, 12),
14.     (20, 15, 36, 10),
15.     (35, 28, 70, 20)
16.   );
17.  
18. var
19.   f: ArbFloat;
20.   k: Integer;
21.   term: Integer = 0;
22.   i, j: Integer;
23.  
24. begin
25.   // write input matrix
26.   WriteLn('A = ');
27.   for i := 1 to n do begin
28.     for j := 1 to n do
29.       Write(a[i, j]:10:0);
30.     WriteLn;
31.   end;
32.   WriteLn;
33.  
34.   // Calculate determinant of symmetric positive definite matrix
35.   detgpd(n, n, a[1, 1], f, k, term);
36.  
37.   // or uncomment to use one of the other methods
38.   // detgen(n, n, a[1, 1], f, k, term);
39.   // detgsy(n, n, a[1, 1], f, k, term);
40.  
41.   if term = 1 then
42.     WriteLn('Determinant of A = ', f * intpower(8, k):0:9, ' (k = ', k, ')')
43.   else
44.     WriteLn('Incorrect input');
45. end.

It wouldn't even have worked back in the old days: too many mistakes....
Note it does not work if the correct library isn't available. but that code is also included: the shared library is tpnumlib.pas which you should compile first.

[MathMan]

@Thaddy,

Totally correct, no question. However you missed the point that the code is a direct copy of the Numlib wiki webpage 

Cheers,
MathMan

[Thaddy]

Then change the wiki page: I just spend a little time testing it.... 
And it works. Also 64 bit:

Code: Text  [Select][+][-]

1. A =
2.          4         2         4         1
3.         30        20        45        12
4.         20        15        36        10
5.         35        28        70        20
6.  
7. Determinant of A = 0.000000000 (k = 2)

I have no opinion about the correctness of the result. Compiled win64-11-fpc-trunk. It probably is correct because of the check.

[MathMan]

I did just that after my first reply - just took a while to create account, etc.

Should be correct now.

[juanirias]

Must be the linux distro...


sorry

I still stuck

[Thaddy]

That can be the case , but give me a few seconds....
Yes the code is windows oriented. No direct UNIX support.
It should work, though, but I have to check the exports.
One or some of the parts already wants libtpnumlib.so. with a wrong name or place.
That is fixable. <logging off for the night, though>

[wp]

I had done a major reconstruction of the determinants chapter in April 2017, and it seems that unfortunately some sub-chapters were mixed up on this occasion. Looked through the history and recreated the original parts. Tested the determinant demos in Linux - working fine.

Use the uses clause as given in the wiki samples. I did not test "uses numlib".

[Thaddy]

@wp
Thanks. Yes it works. Was too late for me yesterday.

@juanirias
It is not the Linux distro, that is agnostic for this code.
Did you get it to work? Otherwise ask again.

Just make sure the library is compiled before the program and make sure the uses clause is correct.