Evaluate gradient function in the for loop.

is a function dependent on time. It is n the function and need to be differentiated to get and the differentiation should pass through the for loop. But for loop picks one value at time t and differentiation failed. Code attached.
function S = Get_Vel(t)
ts=0.0001;
x(t)=cos(2*pi*t)
y(t)=sin(2*pi*t)
vx(t)=gradient(x,ts)
vy(t)=gradient(y,ts)
S=[vx;vy]
end
function A = Compute(t,ts,~,~,~)
S=Get_Vel(t)
end
function solve = solver(F,t0,tf,y0,~)
for t=t0:ts:tf-ts
A =F(t,ts,~,~)
end
solve =A
end
%% MAIN
Result=solver(@Compute,t0,tf,y0,~)
But, since solver used a for loop gradient failed.
Any help is apperciated.
Thank you

4 Kommentare

Sara Boznik
Sara Boznik am 14 Aug. 2020
What happens if you write A(t)?
HN
HN am 14 Aug. 2020
Always zero. Gradient is not calculated.
Sara Boznik
Sara Boznik am 14 Aug. 2020
It looks like you have only constant.
HN
HN am 14 Aug. 2020
No, x and y changes over time. but gradient is not because of lost previous information while passing through for loop.
Thanks

Melden Sie sich an, um zu kommentieren.

Antworten (1)

KSSV
KSSV am 14 Aug. 2020

0 Stimmen

ts=0.0001;
x(t)=cos(2*pi*t) % index of x is t, it cannot be, it shows error
y(t)=sin(2*pi*t)
vx(t)=gradient(x,ts) % index cannot be fraction and to use gradient you need to have x as vector
vy(t)=gradient(y,ts)
S=[vx;vy]
You may rather use:
ts=0.0001 ;
vx = sin(2*pi*t) ;
vy = cos(2*pi*t) ;
S=[vx;vy]

10 Kommentare

HN
HN am 14 Aug. 2020
Bearbeitet: HN am 14 Aug. 2020
Actually x is different from what is shown in the code. It is a bit complicated to simply differentiate it. I just tried to show my problem. if any other tip maybe,
Thank you
KSSV
KSSV am 14 Aug. 2020
Bearbeitet: KSSV am 14 Aug. 2020
USe the vector version, don't call the function for each step. You proceed like this:
t0 = 0 ; t1 = 10 ;
dt = 0.0001 ;
t=t0:dt:t1 ;
x = cos(2*pi*t)
y = sin(2*pi*t)
vx = gradient(x,t)
vy = gradient(y,t)
S=[vx;vy]
HN
HN am 14 Aug. 2020
Well, how bout the for loop ? The function solver has for loop and the vector must pass through that.
How it can be handled ?
Thanks
HN
HN am 14 Aug. 2020
Bearbeitet: HN am 14 Aug. 2020
Is there any technique to save previous value of S outside the function and calculate the gradient from that ?
KSSV
KSSV am 14 Aug. 2020
What is F in the solver?
HN
HN am 14 Aug. 2020
F is RK4 integrator, to inegrate some function that is made from S and other variables .
Thank you
KSSV
KSSV am 14 Aug. 2020
F is a function, it can be evaluated at once. Loop is not required.
HN
HN am 14 Aug. 2020
Can you brief me a bit mote, Dr,
I didnot get it very well. If needed I can post the code here.
KSSV
KSSV am 14 Aug. 2020
It is suggested to post the code here..so that if not me others also can help you.
HN
HN am 22 Aug. 2020
Bearbeitet: HN am 22 Aug. 2020
KSSV ,
Here is the program. I tried to use persistant for t but the dimension exceeded. If t= not a time vector, S become zero all the time.
% time function is to create a time vector
function time(t)
persistent n
n=t;
if isempty(n)
n = 0;
end
n = n+1
end
function [Pose, S] = get_St3PhRS(t)
ts=1/1000;
tt=time(t);
rp=1000; % Radius of the base plate in mm
th=-0.2*cos(2*pi*tt);
psi=0.2*sin(2*pi*tt);
z=707.1068;
phi=atan2(sin(psi)*sin(th),(cos(psi)+cos(th)));
T=Rot('y',th)*Rot('x',psi)*Rot('z',phi)
x=1/2*rp*(-cos(phi)*cos(psi)+cos(phi)*cos(th)+sin(phi)*sin(psi)*sin(th));
y=-rp*cos(psi)*sin(phi);
Pose=[x;y;z;th;psi;phi]
% derivative of moving plate rotation angles
vx=gradient(x,ts);
vy=gradient(y,ts);
vz=gradient(z,ts);
dth=gradient(th,ts); %(2*pi*sin(2*pi*t))/5;
dpsi=gradient(psi,ts); %(2*pi*cos(2*pi*t))/5;
dphi=gradient(phi,ts);%(cos(psi)*sigma1*dpsi+cos(th)*dpsi-cos(psi)*sin(psi)*sin(th)*dth)/(-sigma2*sigma1+sigma2+2*cos(psi)*cos(th)+2*sigma1);
JT=[0, cos(th), cos(psi)*sin(th);1, 0, -sin(psi);0, -sin(th), cos(psi)*cos(th)];
dTh=[dth; dpsi;dphi]
w=JT*dTh;
S=[vx;vy;vz;w(1);w(2);w(3)];
end
function A = InverseVelocityPhRs20200820(T,th,~,q)
:
:
:
S=get_St3PhRS(T) % Here is where the function is called
:
:
A=~~;
end
function [thout,Stm_out,MM_out,Pout,G_out,FRK_OUT,R_out,q_out,Pd_out] = RK4_RhPRS(F,t0,h,tfinal,y0,p0,q0)
% ODE4 Classical Runge-Kutta ODE solver.
th = y0;
R = eye(3);
P = p0;
q = q0;
q_out=q;
thout = th;
Pout = P;
Pd=[-9.9667;0;707.1068];
Pd_out=Pd;
Stm=[0;0;0;1.1080;0; 0.8593];
Stm_out = Stm;
MM_out = [];
G_out = [];
R_out=R;
FRK=[0;0;0;1.1080;0; 0.8593];
FRK_OUT = FRK;
for t = t0 : h : tfinal-h
[k1,~,~,P1,~,~,~,q1] = F(t,th,P,q); % Error in RK4_RhPRS (line 20) : Runge-Kutta
:
:
end
end
% Main function call
[J,S,M,P,G,F,R,q_out,Pd]= RK4_RhPRS(@InverseVelocityPhRs20200820,0,ts,1,th,P,q);
%% Error
Error using time
Too many output arguments.
Error in get_St3PhRS (line 3)
tt=time(t); % t contains all t values but
Error in InverseVelocityPhRs20200820 (line 19)
S=get_St3PhRS(T)
Error in RK4_RhPRS (line 20)
[k1,~,~,P1,~,~,~,q1] = F(t,th,P,q);
Error in Main3PhRS_20200820 (line 103)
[J,S,M,P,G,F,R,q_out,Pd]= RK4_RhPRS(@InverseVelocityPhRs20200820,0,ts,1,th,P,q);

Melden Sie sich an, um zu kommentieren.

Kategorien

Mehr zu Mathematics finden Sie in Hilfe-Center und File Exchange

Gefragt:

HN
am 14 Aug. 2020

Bearbeitet:

HN
am 22 Aug. 2020

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by