question about indexing of logic matrix

Hi:

I read through the blog presented in link:

https://blogs.mathworks.com/graphics/2015/07/22/implicit-surface-intersections/

there is an operation using command:

mask(hel.faces)

Where hel.faces is a m*3 matrix, and mask is a n*1 logic matrix. I could not understand how could this operation happens because it can not reproduce in my side. my test commands are:

A=rand(20,3);
B=rand(20,1);
C=B>0.5;
C(A)

could anyone give me some suggestions?

Thanks!

Yu

Answer 1

[Stephen23]

Your mistake is on this line:

A=rand(20,3);

You correctly wrote that "hel.faces is a m*3 matrix", but you seem to have missed that its values are all indices, i.e. positive integers, with values from 1 to numel(mask):

>> numel(mask)
ans =
      121368
>> max(hel.faces(:))
ans =
      121368
>> min(hel.faces(:))
ans =
     1
>> find(mod(hel.faces(:),1)) % only integers
ans =
   Empty matrix: 0-by-1

That is why it can be used as an index into another array (e.g. mask): because it it contains indices.

But you generated A with random numbers between 0 and 1, which are not indices (indices must be positive integers). Once you generate A with positive integers, with values from 1 to numel(B), then your code will work:

>> B = rand(20,1);
>> A = randi([1,20],20,3);
>> C = B>0.5;
>> C(A)
ans =
     0     1     0
     1     1     0
     0     0     0
     1     0     0
     0     1     1
     0     1     1
     1     1     0
     1     1     0
     0     1     0
     1     0     0
     1     0     0
     1     0     0
     0     0     0
     1     1     1
     1     0     1
     0     1     0
     0     1     1
     0     0     0
     1     1     1
     1     1     1

[Guillaume]

So, specifically, A must be valid indices, i.e. integers between 1 and numel(C):

B=rand(20,1);
C=B>0.5;

A = randi(numel(C), 20, 3);
D = C(A)  %returns a matrix the same shape and size as A where D(r, c) is C(A(r, c))

[Adam Danz]

My answer moved here as a supplement to Stephen's answer.

" Where hel.faces is a m*3 matrix, and mask is a n*1 logic matrix. I could not understand how could this operation happens"

In that blog article, mask is a [121368 x 1] logical vector and hel.faces is a [239778, 3] matrix of integers. Note that the range of values in hel.faces is

min(hel.faces(:))  % = 1
max(hel.faces(:))  % = 121368    = numel(mask)

These values are linear indices of the mask vector. Here's a demo with a smaller amount of data to illustrate this point:

a = -3:6; 
b = [4     7     7
     5     6     3
     6     3     1];
a(b)
ans =
     0     3     3      % a(4) a(7) a(7)
     1     2    -1      % a(5) a(6) a(3)
     2    -1    -3      % a(6) a(3) a(1)

[Yu Li]

thank you all very much!

Bests,

Yu

Answer 2

[Steven Lord]

This is linear indexing into a logical array.

A = [true false]
B = randi(2, 5, 5)
C = A(B)

The elements of C that are true (A(1)) correspond to elements in B that are 1.

The elements of C that are false (A(2)) correspond to elements in B that are 2.

Perhaps a different example, one that has a few more unique values from which to select, would illustrate the technique more clearly.

D = ["Ace of spades"; "Queen of hearts"; "7 of diamonds"; "Jack of clubs"]
selection = randi(numel(D), [4 4])
selectedCards = D(selection)

Element 11 of selectedCards is the element in D whose index is stored in element 11 of selection.

[Yu Li]

Hi Steven:

thank you all the same!

Bests,

Yu