joining various points when using a for loop.

for p=0:0.05:1
     h=p*log(1-p)
     plot(p,h)
end

For the above code i am trying to join the discrete points but i am unable to do so.i used line command but still i get discrete points only i am unable to connect them.

Answer 1

[Star Strider]

I would eliminate the loop and take advantage of MATLAB’s vectorisation capabilities:

p = 0:0.05:1;
h = p.*log(1-p);

figure(1)
plot(p,h)

This uses element-wise operations (denoted by the dot operator in (.*)). See the documenataion for Array vs. Matrix Operations for details.

[s]

It does not solve my problem.still the plot looks like this.I need those points to be connected so that i have a smooth curve.

[Star Strider]

Use my code just as I posted it without the loop (that is, replace your code with mine) to get this plot:

[s]

h(p) = -p.*log2(p)-(1-p).*log2(1-p) instead i replace with this and the plot should start from 0 and end at 1.how can i do that? i mean p does not take index 0 in matlab

[Star Strider]

You have to make h into an anonymous function, then call it just as you would any other function:

p = 0:0.05:1;
h = @(p) -p.*log2(p)-(1-p).*log2(1-p);

figure(1)
plot(p,h(p))

See the documentation on Anonymous Functions for details.

[Image Analyst]

Though this seems like an equivalent but unnecessarily complicated way of doing it, as compared the the first way Star showed you, which was a simple vectorized equation. I mean, why make a function when you don't have to?

p = 0 : 0.05 : 1;
h = -p.*log2(p) - (1-p).*log2(1-p);
plot(p,h, 'r*-', 'LineWidth', 2, 'MarkerSize', 8);
grid on;
xlabel('p', 'FontSize', 20);
ylabel('h', 'FontSize', 20);

[Star Strider]

I created a function because OP wanted a function!

[Image Analyst]

I didn't think so. I think he wrote h(p) = -p.*log2(p)-(1-p).*log2(1-p) just because that's the way he saw it in an assignment or book and didn't realize that h alone without the (p) is also a numerical version of a function -- but I could be wrong. If he wants to use h over and over for different sets of p, then it should be a function. No offense - +1 vote.

[s]

the equation is given in a TB.Using MATLAB, plot the entropy of the binary data source as a function of p = Pr[1], when p = 0, 0.05, 0.10, . . . , 0.95, 1.0. Note that Pr[0] = 1- p. Show that the maximum entropy is achieved when p = 0.5. Make sure that the entropies of p = 0 and 1 are plotted.So i need the graph to be plotted such that it touches 0 and 1.but i am not allowed to use 0 as a index in matlab.pls help me

[Star Strider]

You can either do the function form I provided, or the straight vectorised equation for h Image Analyst provided. Both give you exactly what you want, with the plot. Considering you want h(0), I would use the function form to get it.

Notice that the function form is not an array, so h(0) is the function of h at 0, not the 0 index of an array.

To get the value of the array h where p=0, do this:

p = 0:0.05:1;
h = -p.*log2(p)-(1-p).*log2(1-p);
p0 = find(p == 0);
hp0 = h(p0)

However hp0 is NaN because log2(0) does not exist, and will not plot at that point.

If the equation you provided is correct, this instruction ‘Make sure that the entropies of p = 0 and 1 are plotted.’ is in error, because the entropy at p=0 does not exist.