The Origin Forum

File Exchange

Try Origin for Free

The Origin Forum

All Forums

Origin Forum for Programming

Forum for Origin C

Simple code!

Note: You must be registered in order to post a reply.
To register, click here. Registration is FREE!

Screensize:

UserName:

Password:

Anti-Spam Code:

Format Mode:

Format:

Message:

* HTML is OFF
* Forum Code is ON

Smilies

[quote][i]Originally posted by espenhjo[/i]
[br]Hi,
Because I have been less lucky with the response to some of my postings here I'd like to state a quite simple question related to my C code for use with NLSF.

My function, parplane, has a loop that calculates an array of values that is to be used in the NLSF equation. The parameter I want to optimize is also a part of the array definition so I want the array to vary freely during the fit. Now, the array must be initialized and I have been doing that in the C code for the function. I think this makes a problem for NLSF because the array is reset each time NLSF tries to improve the fit.

My question(s) then is:
Should I move the array initialization loop to "Initialize parameters" in NLSF? Do I need to declare the array also there then? Should the array be declared both for parameter initialization and in the function code?

My code is:
#define JMAX 50
#define xacc 1e-12
#define nr_roots 100	//number of roots to be found

[red]The function:[/red]
double parplane(double a, double q)		//function for use in NLSF
{
	double dSum = 0;					//cumulative output set to zero
	double rtn[nr_roots],x1[nr_roots],x2[nr_roots],fx[nr_roots],df[nr_roots],dx[nr_roots];
	int j;
	int i;

[red]The array initialization:[/red]
	for (i=0; i<nr_roots; i++)
	{
		x1[i] = (i+1)*PI-1;
		x2[i] = (i+1)*PI+1;
		rtn[i] = 0.5*(x1[i]+x2[i]);	
	}
	//these guess for the range in which a root exists could come from a 
	//bracketing and bisection algorithm
	
	i = 0;
	while (i < nr_roots)
	{
		for (j=1;j<JMAX; j++)
		{
			fx[i] = rtn[i] * tan(rtn[i]) / a;				//the function with roots
			df[i] = tan(rtn[i]) / a + rtn[i] * (1+tan(rtn[i]) * tan(rtn[i])) / a;	//first derivative of the function
						
			dx[i] = -fx[i]/df[i];
			rtn[i] += dx[i];

printf("%e\n", dx[i]);

if ((x1[i]-rtn[i])*(rtn[i]-x2[i]) < 0.0){
				printf("error, jumped outside bounds");
				exit(1);}
			if (fabs(dx[i]) < xacc){
				////printf("found root after %d attempts, at %lf\n", j, rtn[i]);
				printf("The value of rtn[%d] is %lf\n", i, rtn[i]);
				break;}
			if (j = JMAX - 1){
				printf("error - exceeded max tries no root");
				i++;
				exit(1);}
		}
		dSum += exp(-(rtn[i]^2) / (a^2))  * 2 * (1 + sin(2*rtn[i])/(2*rtn[i]))^(-1) * ((2*PI*q*a)*sin(2*PI*q*a)*cos(rtn[i])-rtn[i]*cos(2*PI*q*a)*sin(rtn[i]))^2 / ((2*PI*q*a)^2-rtn[i]^2)^2;
		i++;
	}
	return dSum;
}

Thank yoouu!

Espen

[/quote]

Check here to include your profile signature.
Check here to subscribe to this topic.

T O P I C R E V I E W
espenhjo	Posted - 11/19/2003 : 08:28:27 AM Hi, Because I have been less lucky with the response to some of my postings here I'd like to state a quite simple question related to my C code for use with NLSF. My function, parplane, has a loop that calculates an array of values that is to be used in the NLSF equation. The parameter I want to optimize is also a part of the array definition so I want the array to vary freely during the fit. Now, the array must be initialized and I have been doing that in the C code for the function. I think this makes a problem for NLSF because the array is reset each time NLSF tries to improve the fit. My question(s) then is: Should I move the array initialization loop to "Initialize parameters" in NLSF? Do I need to declare the array also there then? Should the array be declared both for parameter initialization and in the function code? My code is: #define JMAX 50 #define xacc 1e-12 #define nr_roots 100 //number of roots to be found The function: double parplane(double a, double q) //function for use in NLSF { double dSum = 0; //cumulative output set to zero double rtn[nr_roots],x1[nr_roots],x2[nr_roots],fx[nr_roots],df[nr_roots],dx[nr_roots]; int j; int i; The array initialization: for (i=0; i<nr_roots; i++) { x1[i] = (i+1)PI-1; x2[i] = (i+1)PI+1; rtn[i] = 0.5(x1[i]+x2[i]); } //these guess for the range in which a root exists could come from a //bracketing and bisection algorithm i = 0; while (i < nr_roots) { for (j=1;j<JMAX; j++) { fx[i] = rtn[i] tan(rtn[i]) / a; //the function with roots df[i] = tan(rtn[i]) / a + rtn[i] * (1+tan(rtn[i]) * tan(rtn[i])) / a; //first derivative of the function dx[i] = -fx[i]/df[i]; rtn[i] += dx[i]; printf("%e\n", dx[i]); if ((x1[i]-rtn[i])(rtn[i]-x2[i]) < 0.0){ printf("error, jumped outside bounds"); exit(1);} if (fabs(dx[i]) < xacc){ ////printf("found root after %d attempts, at %lf\n", j, rtn[i]); printf("The value of rtn[%d] is %lf\n", i, rtn[i]); break;} if (j = JMAX - 1){ printf("error - exceeded max tries no root"); i++; exit(1);} } dSum += exp(-(rtn[i]^2) / (a^2)) 2 * (1 + sin(2rtn[i])/(2rtn[i]))^(-1) * ((2PIqa)sin(2PIqa)cos(rtn[i])-rtn[i]cos(2PIqa)sin(rtn[i]))^2 / ((2PIqa)^2-rtn[i]^2)^2; i++; } return dSum; } Thank yoouu! Espen
2 L A T E S T R E P L I E S (Newest First)
espenhjo	Posted - 11/20/2003 : 02:53:50 AM This was exactly what I ment! Thank you! Espen
yuivanov	Posted - 11/19/2003 : 10:29:09 AM Hi Espen Usually if you want some part of the function to be executed only on first call you may use static BOOL to indicate that function is being executed for the first time. If I understood you correctly you want the following part of the function to be executed only once. for (i=0; i<nr_roots; i++) { x1[i] = (i+1)PI-1; x2[i] = (i+1)PI+1; rtn[i] = 0.5(x1[i]+x2[i]); } so if you do this: static BOOL l_bFirstTime = TRUE; if( l_bFirstTime ) { l_bFirstTime = FALSE; for (i=0; i<nr_roots; i++) { x1[i] = (i+1)PI-1; x2[i] = (i+1)PI+1; rtn[i] = 0.5(x1[i]+x2[i]); } } it should initialize your array only once per fitting session. Hope this helps. Yuri

The Origin Forum

Snitz Forums 2000