Serial.print ovf

Projects Programming
1

Don't hate me. I am not good with c, so seek for your advise.
trying to implement multilateration, found this algorithm in one paper.

Can you suggest me what should I do/ how should I define my parameters to avoid this ovf problem? I think numbers I get are too big, and because of this I can't find final value.
tried to make it long double, but it didn't work/ all of the Serial.prints just to follow where the code fails.

Here is my code:

Serial.begin(9600);

long TOAi=millis();
long TOAj=TOAi+3, TOAk=TOAi+7, TOAl=TOAi+18;

int CurrC=340;

int xi=0, yi=0, zi=0;
int xj=0, yj=50, zj=0;
int xk=50, yk=0, zk=0;
int xl=50, yl=50, zl=50;

float Rij=(float) CurrC*(TOAi-TOAj);
float Rik=(float) CurrC*(TOAi-TOAk);
float Ril=(float) CurrC*(TOAi-TOAl);

double Rkl=(double) CurrC*(TOAk-TOAl);
double Rkj=(double) CurrC*(TOAk-TOAj);

//A=Xikx/Xijy
double A=(Rik*(xj-xi)-Rij*(xk-xi))/(Rij*(yk-yi)-Rik*(yj-yi));
Serial.println(A);
//B=Xikz/Xijy
double B=(Rik*(zj-zi)-Rij*(zk-zi))/(Rij*(yk-yi)-Rik*(yj-yi));
Serial.println(B);
//C=Xklx/Xkjy
double C=(Rkl*(xj-xk)-Rkj*(xl-xk))/(Rkj*(yl-yk)-Rkl*(yj-yk));
Serial.println(C);
//D=Xklz/Xkjy
double D=(Rkl*(zj-zk)-Rkj*(zl-zk))/(Rkj*(yl-yk)-Rkl*(yj-yk));
Serial.println(D);
//E=(Rik*Rij2xyz-Rij*Rik2xyz)/2Xijy
double E=(Rik*(Rij*Rij+(xi*xi+yi*yi+zi*zi) - (xj*xj+yj*yj+zj*zj))-Rij*(Rik*Rik + (xi*xi+yi*yi+zi*zi) - (xk*xk+yk*yk+zk*zk)))/(2*(Rij*(yk-yi)-Rik*(yj-yi)));
Serial.println(E);
//F=(Rkl*Rkj2xyz - Rkj*Rkl2xyz)/2Xkjy
double F=((Rkl*(Rkj*Rkj+(xk*xk+yk*yk+zk*zk) - (xj*xj+yj*yj+zj*zj))-Rkj*(Rkl*Rkl + (xk*xk+yk*yk+zk*zk) - (xl*xl+yl*yl+zl*zl))))/(2*(Rkj*(yl-yk)-Rkl*(yj-yk)));
Serial.println(F);
double G=(D-B)/(A-C);
double H=(F-E)/(A-C);
Serial.println(G);
Serial.println(H);
double I=A*G+B;
double J=A*H+E;
Serial.println(I);
Serial.println(J);
double K=(Rik*Rik + (xi*xi+yi*yi+zi*zi) - (xk*xk+yk*yk+zk*zk) +2*(xk-xi)*H +2*(yk-yi)*J);
double L=2*((xk-xi)*G+(yk-yi)*I+(zk-zi));
double M=4*Rik*Rik*(G*G+I*I+1)-L*L;
double N=8*Rik*Rik*(G*(xi-H)+I*(yi-J)+zi)+2*L*K;
double O=4*Rik*Rik*((xi-H)*(xi-H)+(yi-J)*(yi-J)+zi*zi)-K*K;
Serial.println(K);
Serial.println(L);
Serial.println(M);
Serial.println(N);
Serial.println(O);

double Q;
//sign() function...
if (N>0) Q=N+sqrt(N*N-4*M*O);
else if(N<0) Q=N-sqrt(N*N-4*M*O);
else Q=N;
Serial.println(Q);
long double pos[6]={G*Q/2/M+H, I*Q/2/M+J, Q/2/M, G*2*O/Q+H, I*2*O/Q+J, 2*O/Q};
2

here is what I get:

0.00
0.73
-0.27
13886.28
-50850.67
0.88
212418.12
0.38
104922.62
26903712.29
87.50
43183394.39
ovf
ovf
nan

3 
float Rij=(float) CurrC*(TOAi-TOAj);

The (float) cast is pointless!. The compiler will do this for you AFTER you have done 3405050 which will not fit in an int ....

Mark

4

well, I understand. as I told you I am numb in these variables types... so did it just in case!

5

so did it just in case

Perhaps you should read holmes4's complete message. Particularly this part:

which will not fit in an int ....

Then, perhaps you should consider why holmes4 would have even mentioned an int.

Look at the types of ALL the variables involved in that statement. Then, consider making CurrC a long, along with the other ints.