Posts: 60
Threads: 8
Joined: Jan 2005
Hello,
I have just added a solver for linear equations (Gauss-Jordan method):
http://www.unilim.fr/pages_perso/jean.de...fbmath.zip
Solvers for non-linear equations (Newton-Raphson) will follow.
Question: is it necessary to add $DYNAMIC in a procedure which uses local dynamic arrays, to ensure that these arrays are correctly deallocated at the end of the procedure? I have never used this directive in my QB programs, nor in this solver.
Thanks in advance,
Jean Debord
Posts: 6,419
Threads: 74
Joined: Mar 2002
If you have dynamic arrays, and you allocate them at the beginning of a SUB procedure, you have to deallocate them at the end with ERASE.
Posts: 922
Threads: 15
Joined: Jun 2003
Deallocation is done automatically with local dynamic arrays.
But they don't need to be dynamic, keep using static arrays (DIM'ed), access is much faster, as they are allocated on stack - that is by default 1MB, what can be changed using the command-line option "-t 4096" for a 4MB stack, for example.
Posts: 6,419
Threads: 74
Joined: Mar 2002
Oh, that's cool. QB ran out of memory if you didn't take care of deallocating Dynamic stuff... Not really sure about it, but when I added those ERASEs my games worked
Posts: 60
Threads: 8
Joined: Jan 2005
Quote:Deallocation is done automatically with local dynamic arrays
Thanks. So, I was right by not using $DYNAMIC
Quote: keep using static arrays (DIM'ed), access is much faster
OK. But it is difficult in scientific applications. In most cases the array size is known only at run time.
Posts: 60
Threads: 8
Joined: Jan 2005
Hello,
I have added some non-linear optimization routines:
* golden search
* minimization along a line
* Newton-Raphson
More routines will be added soon.
Best regards,
Jean Debord
Posts: 60
Threads: 8
Joined: Jan 2005
I am experiencing difficulties with FB 0.11. The compilations go well, but the PRINT USING statement does not work anymore, so that the output of my programs are very ugly !
There was no problem with FB 0.10
Thanks in advance for correcting this bug.
Jean Debord
Posts: 1,407
Threads: 117
Joined: Dec 2002
AFAIK PRINT USING has never worked. Is in the unsupported keywords list.
V0.11 is using more digits to print DOUBLES than V0.10...
Antoni
Posts: 60
Threads: 8
Joined: Jan 2005
The following program works in QB but not in FB 0.11
Code:
DIM X AS DOUBLE
X = 1.23456789
PRINT USING "##.########"; X ' Should print 1.23456789
PRINT USING "##.####"; X ' Should print 1.2346
PRINT USING "##.###"; X ' Should print 1.235
DO: LOOP UNTIL INKEY$ <> ""
With FB 0.11, the program prints the whole number of digits, regardless of the format string.
It seems that FB does not round the number before printing it.
Posts: 73
Threads: 6
Joined: Jan 2005
Maybe I have not written QB program for ages, now I find it is difficult to remember and use Basic's 'print using' than C's 'printf format string'. Perhaps it is not easy to achieve printf-like in fbc since it is written in fbc itself, but who knows :wink:
