Convert decimal and negative values ​​to integers

There are x and y coordinates that I need to convert to integer values ​​while preserving the meaning of the graph

a = 0:pi/100:2*pi;
x=cos(a);
y = sin(x);
plot(x,y) % first graph
% plot(a,y) % second graph

Thanks in advance

[Dyuman Joshi]

By preserve I assume the plot must remain same but the values must be converted to integers? And if so, how do you want to convert values to integers?

[Lev Mihailov]

@Dyuman Joshi yes, the values ​​must be integers. By conversion, I thought about sorting values, replace fractional values ​​with integers

[Rik]

The only non-negative integers you will get out of sin and cos are 0 and 1.

Do you mean you want to preserve the shape of the graph?

Why do you want to only leave positive integers? Are you planning to use these values as indices?

a=linspace(0,2*pi,200);
x=cos(a);
y=sin(a);
plot(x,y)
hold on
plot(a,y)
hold off
daspect([1 1 1])

[Lev Mihailov]

@Rik Yes, I only want to keep the graph shape. Yes this data will be used as matrix indexes

[Dyuman Joshi]

"replace fractional values ​​with integers"

What do you mean by this?

[Lev Mihailov]

@Dyuman Joshi I mean, 0.3 0.5 0.7 is replaced by 1 2 3, and -5.4 1.2 4.5 is replaced by 1 3 4

[Dyuman Joshi]

Only these specific values?

[Lev Mihailov]

@Dyuman Joshi well i would like a generic method

[Dyuman Joshi]

Don't you think you should think about it? Especially when it look like a random method.

[Lev Mihailov]

@Dyuman Joshi I thought for a week and there was no good thought

[Dyuman Joshi]

Ok. So, What is it that you are trying to do in the first place? How did you come up with the conversions you mentioned above?

[Lev Mihailov]

@Dyuman Joshi take min and max values, create an array, from -5.4 ; 1.2 to 4.5, fill array from 1 to n(up to 4), where 1 is the smallest value and 4 is the maximum

Answer 1

[Matt J]

Something like this, perhaps?

a = 0:pi/100:2*pi;
x=cos(a);
y = sin(x);

fcn=@(z)round(rescale(z,0,intmax('uint16')));

plot(fcn(a),fcn(y)) 

Answer 2

[Rik]

Since all your values are between -1 and 1, you need a method to discretize your data to some arbitrary (but repeatable) indices. You have two general options:

1. Have a linear lookup table. This will map all numbers lineraly to the closest integer.
2. Have a non-linear lookup table. This allows you to decide what data ranges should have more precision.

Below I will implement the simplest version of option 1 I can think of. If you want to increase performance, interp1 may be required.

discretize_values([-1 0 1],3)

ans = 1×3

1 2 3

Now we can apply this to your curve. What I show here is how you can set the precision to influence how jagged your graph will look. Set a higher number for a smoother curve. The mapping back to non-integers should of course be removed if you are planning to use this for indexing.

a=linspace(0,2*pi,200);x=cos(a);y=sin(a);
plot(a,y)
hold on
precision=20;
y2=discretize_values(y,precision);
y2=2*((y2-1)/(precision-1))-1; % map back to non-integers
plot(a,y2)
hold off

function out=discretize_values(in,precision)
% map (-1,1) to 1:precision
if any(in>1) || any(in<-1)
    error('data outside expected range')
end
out=1+round((precision-1)*(in+1)/2);
end