Thread Subject:

finding distances that satisfy the equation ? (confused :'( , please help)

Hi all, i've generated a code as below:
basically its 20 randomly placed nodes in a 1x1 square. i've then found a matrix of angles of each node and every other node (i.e. node 1 + node 2,3,4,..20)

i then wrote: i want to find angles that fall within
sect1=((0<thetasort) & (thetasort<(0.25*pi)));
i.e. 0rad to pi/4 rad.
returning me a matrix of 0 and 1's, (1's if they satisfy the eqn between 0-pi/4rad)
[sect1sort,R1]=sort(sect1,1)
to sort...
now:
for those 1's (that satisfy), i want to find the distance between e.g. node 1 + those that satisfy, node 2 + those that satisfy etc)

apologies for the confusion, its so hard to put it into words :(

but i hope i can find a solution to this

any other ideas are welcome!!

cheers
natalie


========Script: ================
a=20; %number of nodes

% Randomly selecting values for x,y and z-coordinates
x=rand(1,a); %generate 1xa sized matrix of random numbers between 0 and 1
y=rand(1,a);

theta=zeros(a,a);
for node=1:a;
    for u=1:a;
        theta(node,u)=atan2(imag((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]),...
                        real((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]));
    end
end

[thetasort,R]=sort(theta,1); % increasing angles downward (column)

sect1=((0<thetasort) & (thetasort<(0.25*pi)));

"Natalie Sin Hwee " <sin.ng09@imperial.ac.uk> wrote in message <hoo666$8u3$1@fred.mathworks.com>...
> Hi all, i've generated a code as below:
> basically its 20 randomly placed nodes in a 1x1 square. i've then found a matrix of angles of each node and every other node (i.e. node 1 + node 2,3,4,..20)
>
> i then wrote: i want to find angles that fall within
> sect1=((0<thetasort) & (thetasort<(0.25*pi)));
> i.e. 0rad to pi/4 rad.
> returning me a matrix of 0 and 1's, (1's if they satisfy the eqn between 0-pi/4rad)
> [sect1sort,R1]=sort(sect1,1)
> to sort...
> now:
> for those 1's (that satisfy), i want to find the distance between e.g. node 1 + those that satisfy, node 2 + those that satisfy etc)
>
> apologies for the confusion, its so hard to put it into words :(
>
> but i hope i can find a solution to this
>
> any other ideas are welcome!!
>
> cheers
> natalie
>
>
> ========Script: ================
> a=20; %number of nodes
>
> % Randomly selecting values for x,y and z-coordinates
> x=rand(1,a); %generate 1xa sized matrix of random numbers between 0 and 1
> y=rand(1,a);
>
> theta=zeros(a,a);
> for node=1:a;
> for u=1:a;
> theta(node,u)=atan2(imag((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]),...
> real((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]));
> end
> end
>
> [thetasort,R]=sort(theta,1); % increasing angles downward (column)
>
> sect1=((0<thetasort) & (thetasort<(0.25*pi)));

a hint:
- create a list [IR,IC] of those nodes only that satisfy your condition...
- then, compute their distance in a simple loop...

     [ir,ic]=find(sect1);
     lst=[ir,ic];
% now, loop through your list LST...

just a thought...
us

"Natalie Sin Hwee " <sin.ng09@imperial.ac.uk> wrote in message <hoo666$8u3$1@fred.mathworks.com>...
> ........
> for node=1:a;
> for u=1:a;
> theta(node,u)=atan2(imag((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]),...
> real((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]));
> end
> end
>
> [thetasort,R]=sort(theta,1); % increasing angles downward (column)
>
> sect1=((0<thetasort) & (thetasort<(0.25*pi)));
----------
  This looks like a very awkward way to compute your angles. Why take the trouble to create complex-valued quantities only to convert them back to real values again? Why not just say this:

 theta(node,u)=atan2(y(u)-y(node),x(u)-x(node));

  With either method you will be obtaining angles measured counterclockwise from the x-axis to the directed line between each pair of points and which range from -pi to +pi. You will be selecting only those between 0 and +pi/4, and in particular rejecting those between -pi/4 and 0. In the cases where node equals u along the diagonal your results will be meaningless. Is this what you want?

  Why do you sort the angles before subjecting them to your range tests? What you will get is a series of falses, then trues, and finally falses. As your code stands there is no way to associate this with the original data in x and y.

Roger Stafford

"Roger Stafford" <ellieandrogerxyzzy@mindspring.com.invalid> wrote in message <hoo9it$r4p$1@fred.mathworks.com>...
> .......
> Why do you sort the angles before subjecting them to your range tests? What you will get is a series of falses, then trues, and finally falses. As your code stands there is no way to associate this with the original data in x and y.
> .......

  Correct that last sentence. You could use your 'R' to relate the results in 'sect1' to the original data, but I still ask, why do the sort?

Roger Stafford

"Roger Stafford" <ellieandrogerxyzzy@mindspring.com.invalid> wrote in message <hoo9it$r4p$1@fred.mathworks.com>...
> "Natalie Sin Hwee " <sin.ng09@imperial.ac.uk> wrote in message <hoo666$8u3$1@fred.mathworks.com>...
> > ........
> > for node=1:a;
> > for u=1:a;
> > theta(node,u)=atan2(imag((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]),...
> > real((x(1,u)-x(1,node))+[(y(1,u)-y(1,node))*i]));
> > end
> > end
> >
> > [thetasort,R]=sort(theta,1); % increasing angles downward (column)
> >
> > sect1=((0<thetasort) & (thetasort<(0.25*pi)));
> ----------
> This looks like a very awkward way to compute your angles. Why take the trouble to create complex-valued quantities only to convert them back to real values again? Why not just say this:
>
> theta(node,u)=atan2(y(u)-y(node),x(u)-x(node));
>
> With either method you will be obtaining angles measured counterclockwise from the x-axis to the directed line between each pair of points and which range from -pi to +pi. You will be selecting only those between 0 and +pi/4, and in particular rejecting those between -pi/4 and 0. In the cases where node equals u along the diagonal your results will be meaningless. Is this what you want?
>
> Why do you sort the angles before subjecting them to your range tests? What you will get is a series of falses, then trues, and finally falses. As your code stands there is no way to associate this with the original data in x and y.
>
> Roger Stafford

hiya,

basically i used a more 'complex' method because im working from 0-pi and 0 - (-pi) hence the atan2. im nore sure why im using the real and imaginary. apparently it will work in 3d (through my previous quetisons on the forum) - which i will later need to make my model into 3d !! ^^

thanks
natalie

"Roger Stafford" <ellieandrogerxyzzy@mindspring.com.invalid> wrote in message <hoobea$nnq$1@fred.mathworks.com>...
> "Roger Stafford" <ellieandrogerxyzzy@mindspring.com.invalid> wrote in message <hoo9it$r4p$1@fred.mathworks.com>...
> > .......
> > Why do you sort the angles before subjecting them to your range tests? What you will get is a series of falses, then trues, and finally falses. As your code stands there is no way to associate this with the original data in x and y.
> > .......
>
> Correct that last sentence. You could use your 'R' to relate the results in 'sect1' to the original data, but I still ask, why do the sort?
>
> Roger Stafford

=========

Oh you are right, i don't need the sorting out!
i was so clueless and stuck so i just did whatever randomly ^^

thanks for reminding me, took it out now!!

> a hint:
> - create a list [IR,IC] of those nodes only that satisfy your condition...
> - then, compute their distance in a simple loop...
>
> [ir,ic]=find(sect1);
> lst=[ir,ic];
> % now, loop through your list LST...
>
> just a thought...
> us

=====

thanks us!! i did that and it works!!!!! awesome!! ive added in:

sect1=((0<theta) & (theta<(0.25*pi)));

[g1,k1]=find(sect1>=1); first=[g1,k1];
first_sort=sort(first,2);

L1=length(k1);
dist1=zeros(L1,1);
for n=1:L1
dist1(n,1)=sqrt((x(1,first_sort(n,2))-x(1,first_sort(n,1))).^2+...
    (y(1,first_sort(n,2))-y(1,first_sort(n,1))).^2);
end


and that returns me a the 'distance between nodes' corresponding to what i've found in 'first' - column1 wrt to column 2 'nodes'

any idea how i can separate them out into the different 'nodes' wrt to column 1. e.g.

node 1 + 2,7,8,9,(random whatever it has given me)

my aim is to find,

the nearest node to each node within that angle (0 to pi/4)...

(will be repeated as applied to the 8 sectors in the circle)

*omg sorry this is getting more and more confusing !! ^^

thanks lots just for reading through it!
Natalie