How to find two local slopes in an arbitrary curve?

3 views (last 30 days)
I am used Matlab to find a model from curves. So i am processing curvesand i have some difficult to find two local slopes?
For information i am not an expert in math.
Thanks for help!
Best regards
  2 Comments
Jan
Jan on 3 Jun 2013
What does "processing curves" exactly mean? Do youz have two vector of x- and y_values? Or a sheet of paper with a hand-drawn line?

Sign in to comment.

Answers (5)

Iain
Iain on 3 Jun 2013
The general expression to calculate "slope" is (y2 - y1) / (x2 - x1)
If you have a vector of "y" values and "x" values of the same length, you can:
slope = diff(y) ./ diff(x);
slope_x_vals = (x(2:end) + x(1:end-1))/2;
plot(slope_x_vals,slope_y_vals)
If your y value is in metres, and your x value is in seconds, the units of "slope" is metres per second.

Body_sea
Body_sea on 3 Jun 2013
Hello,
I see my question is not clear...
I have a curve stored in a vector X and Y as the following picture :
in an arbitraru curve i know there a two local slopes. So my question is how to find the two local slopes in the curve?
With the picture i hope it is clear.
Best regards.

Jan
Jan on 3 Jun 2013
The gradient function calculates the slope in a fast and stable way.

Body_sea
Body_sea on 3 Jun 2013
I know the gradient function in Matlab. But you can see, my problem is not here, i am looking for the two local slope. i mean after to have the gradient of the curve, i want to find the local linear slope and to make sure this is the good of the two slopes i want.
An idea?
Regards

Jules Ray
Jules Ray on 16 Jun 2015
Use ginput (four in this case) to the boundaries or the limits that enclose both areas, for each area use for polyfit, and lineal interpolations, this allows estimates the slope but also the associated 2 sigma error of your measurements

Categories

Find more on Curve Fitting Toolbox in Help Center and File Exchange

Community Treasure Hunt

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

Start Hunting!