central difference method

How can I calculate the central difference for set of data using matlab If I have big data.Could any one help me to do it for this small data so I can I apply to my data X 0.225 0.30 0.35 0.40 0.45 0.475 Y 0.251 0.90 2.02 3.63 7.2 9.800

Answer 1

[Jan]

You can simply use:

gradient(X, Y)

Then the first and the last element are forward and backward differences respectively. But all interior elements are central differences.

For large data sets FEX: DGradient is faster (10 to 16 times) than Matlab's gradient. In addition it can calulate the 2nd order approximation, when X is not uniformly distributed.

[Muhammad Hammad Malik]

@Jan i am asking here because already a post here, if you want i can have a new post.

i want to calculate gradient (central difference) from my filtered data. Want to calculate with window size=10samples and 50%overlap to previous window, how can do this.

[Jan]

t  = linspace(0, 2*pi, 1000);
x  = sin(t);
dx = (x(11:5:end) - x(1:5:end-10)) / (10 * diff(t(1:2)));
plot(t(1:5:end-5), cos(t(1:5:end-5)), 'ro');
hold on
plot(t(5:5:end-10),dx)

[Muhammad Hammad Malik]

@Jan thanks for the kind response. i have checked your code, but unable to understand well. why you use 11:5 and cos.

pls see what i have done, i am attaching here, the issue is i am unable to add window overlap.I also can change the window size to bigger or smaller, not restricted to use just 10.

The one you mentioned is working but how i can plot that with my time vector "IMU_ULISS.time" (1x25805), and i changed this time vector to plot mine as you can see.

Grad =gradient(Data_filtered(1:window:end),window)

figure;

plot(IMU_ULISS.time(1:window:end),Grad)

time_modified=IMU_ULISS.time(1:window:end)

hold onplot(time_modified(ceil(find(IMU_ULISS.step_instant_target(1,:)>0)/window)),Grad(ceil(find(IMU_ULISS.step_instant_target(1,:)>0)/window)),'r');

hold on

Answer 2

[abdulhadi khalifa]

I want to calculate the slope by this method for y and X how I can do it

[Jan]

The output of GRADIENT *is* the slope. See my former answer.

It would be more helpful if you reply to the different ideas you got in your other postings concerning this problem. Simply posting the problem repeatedly in new threads wastes the time of all who want to assist.