Matlab: reshape 3-dimensional array into 2-dimensional array - arrays

I have a 3x3x2000 array of rotation matrices that I need to transform into a 2000x9 array.
I think I have to use a combination of permute() and reshape(), but I don't get the correct output order.
This is what I need:
First row of 3x3 array needs to be columns 1:3 in the output
Second row of 3x3 array needs to be columns 4:6 in the output
Third row of 3x3 array needs to be columns 7:9 in the output
I have tried all possible combinations of numbers 1 2 3 in the following code:
out1 = permute(input, [2 3 1]);
out2 = reshape(out1, [2000 9]);
But I always end up with the wrong order. Any tips for a Matlab newbie?

How about a simple for-loop?
for i=1:size(myinput,3)
myoutput(i,:)=[myinput(1,:,i) myinput(2,:,i) myinput(3,:,i)];
% or
% myoutput(i,:)=reshape(myinput(:,:,i),[],9);
end
It's not simple as using permute and reshape, but it is transparent and easier for debugging. Once everything in your program runs perfectly, you can consider to rewrite such for-loops in your code...

You had a mix-up in your permute
a = reshape(1:9*6, 3, 3, []);
a is a 3x3x6 matrix, each
a(:,:,i) = 9*(i-1) + [1 4 7
2 5 8
3 6 9];
So
out1 = permute(a, [3,1,2]);
out2 = reshape(out1, [], 9);
Or in one line
out3 = reshape(permute(a, [3,1,2]), [], 9);
So
out2 = out3 =
1 2 3 4 5 6 7 8 9
10 11 12 13 14 15 16 17 18
19 20 21 22 23 24 25 26 27
28 29 30 31 32 33 34 35 36
37 38 39 40 41 42 43 44 45
46 47 48 49 50 51 52 53 54

Related

2D matrix to 3D matrix with row to [row, col] mapping

I have a 2D matrix with in the 1st dimension different channels, and in the 2nd dimension time samples. I want to rearrange this to a 3D matrix, with in the 1st and 2nd dimension channels, and in the 3rd time samples.
The channels have to mapped according to a certain mapping. Right now I am using a for-loop to do so, but what would be a no-loop solution?
N_samples = 1000;
N_channels = 64;
channel_mapping = reshape(1:64, [8 8]).';
% Results in mapping: (can also be random)
% 1 2 3 4 5 6 7 8
% 9 10 11 12 13 14 15 16
% 17 18 19 20 21 22 23 24
% 25 26 27 28 29 30 31 32
% 33 34 35 36 37 38 39 40
% 41 42 43 44 45 46 47 48
% 49 50 51 52 53 55 55 56
% 57 58 59 60 61 62 63 64
data = rand(N_channels, N_samples);
data_grid = NaN(8,8, N_samples);
for k = 1:N_samples
tmp = data(:, k);
data_grid(:, :, k) = tmp(channel_mapping);
end
You can do it in one go as follows:
data_grid = reshape(data(channel_mapping, :), 8, 8, []);

Julia : Cartesian product of multiple arrays

I would like to compute a product iterator using Iterators.jl.
Let's say I have an array of UnitRanges tab with a priori unknown size.
I would like to compute the cartesian product of the elements of tab.
For example if tab length is 2 and tab[1] = a and tab[2] = b I want to compute product(a,b) from Iterators.jl.
I want to make a generic function that compute the cartesian product of every component in tab.
I tried something like this
prod = tab[1]
for i in tab[2:end]
prod = product(prod,i)
end
However if tab is length 3, components a,b and c, I obtain in prod elements under the form (1,(3,2)) and not (1,3,2). With 1 element of c, 3 element of b and 2 element of a.
In v0.5, there is now Base.product, which is much better than Iterators.product.
It can handle as many arrays as needed, and it even has a shape:
julia> collect(Base.product([1, 2], [3, 4]))
2×2 Array{Tuple{Int64,Int64},2}:
(1,3) (1,4)
(2,3) (2,4)
julia> collect(Base.product(1:5, 1:3, 1:2, 1:2))
5×3×2×2 Array{NTuple{4,Int64},4}:
[:, :, 1, 1] =
(1,1,1,1) (1,2,1,1) (1,3,1,1)
(2,1,1,1) (2,2,1,1) (2,3,1,1)
(3,1,1,1) (3,2,1,1) (3,3,1,1)
(4,1,1,1) (4,2,1,1) (4,3,1,1)
(5,1,1,1) (5,2,1,1) (5,3,1,1)
[:, :, 2, 1] =
(1,1,2,1) (1,2,2,1) (1,3,2,1)
(2,1,2,1) (2,2,2,1) (2,3,2,1)
(3,1,2,1) (3,2,2,1) (3,3,2,1)
(4,1,2,1) (4,2,2,1) (4,3,2,1)
(5,1,2,1) (5,2,2,1) (5,3,2,1)
[:, :, 1, 2] =
(1,1,1,2) (1,2,1,2) (1,3,1,2)
(2,1,1,2) (2,2,1,2) (2,3,1,2)
(3,1,1,2) (3,2,1,2) (3,3,1,2)
(4,1,1,2) (4,2,1,2) (4,3,1,2)
(5,1,1,2) (5,2,1,2) (5,3,1,2)
[:, :, 2, 2] =
(1,1,2,2) (1,2,2,2) (1,3,2,2)
(2,1,2,2) (2,2,2,2) (2,3,2,2)
(3,1,2,2) (3,2,2,2) (3,3,2,2)
(4,1,2,2) (4,2,2,2) (4,3,2,2)
(5,1,2,2) (5,2,2,2) (5,3,2,2)
The shape is extremely useful for map. For instance, here's how to create a multiplication table using Base.product:
julia> map(prod, Base.product(1:9, 1:9))
9×9 Array{Int64,2}:
1 2 3 4 5 6 7 8 9
2 4 6 8 10 12 14 16 18
3 6 9 12 15 18 21 24 27
4 8 12 16 20 24 28 32 36
5 10 15 20 25 30 35 40 45
6 12 18 24 30 36 42 48 54
7 14 21 28 35 42 49 56 63
8 16 24 32 40 48 56 64 72
9 18 27 36 45 54 63 72 81
Of course, if you don't need the shape, then you are free to ignore it — it will still iterate properly.
And Base.product is fast too!

Retrieving specific elements in matlab from arrays in MATLAB

I have an array in MATLAB
For example
a = 1:100;
I want to select the first 4 element in every successive 10 elements.
In this example I want to b will be
b = [1,2,3,4,11,12,13,14, ...]
can I do it without for loop?
I read in the internet that i can select the element for each step:
b = a(1:10:end);
but this is not working for me.
Can you help me?
With reshape
%// reshaping your matrix to nx10 so that it has successive 10 elements in each row
temp = reshape(a,10,[]).'; %//'
%// taking first 4 columns and reshaping them back to a row vector
b = reshape(temp(:,1:4).',1,[]); %//'
Sample Run for smaller size (although this works for your actual dimensions)
a = 1:20;
>> b
b =
1 2 3 4 11 12 13 14
To vectorize the operation you must generate the indices you wish to extract:
a = 1:100;
b = a(reshape(bsxfun(#plus,(1:4)',0:10:length(a)-1),[],1));
Let's break down how this works. First, the bsxfun function. This performs a function, here it is addition (#plus) on each element of a vector. Since you want elements 1:4 we make this one dimension and the other dimension increases by tens. this will lead a Nx4 matrix where N is the number of groups of 4 we wish to extract.
The reshape function simply vectorizes this matrix so that we can use it to index the vector a. To better understand this line, try taking a look at the output of each function.
Sample Output:
>> b = a(reshape(bsxfun(#plus,(1:4)',0:10:length(a)-1),[],1))
b =
Columns 1 through 19
1 2 3 4 11 12 13 14 21 22 23 24 31 32 33 34 41 42 43
Columns 20 through 38
44 51 52 53 54 61 62 63 64 71 72 73 74 81 82 83 84 91 92
Columns 39 through 40
93 94

check if ALL elements of a vector are in another vector

I need to loop through coloumn 1 of a matrix and return (i) when I have come across ALL of the elements of another vector which i can predefine.
check_vector = [1:43] %% I dont actually need to predefine this - i know I am looking for the numbers 1 to 43.
matrix_a coloumn 1 (which is the only coloumn i am interested in looks like this for example
1
4
3
5
6
7
8
9
10
11
12
13
14
16
15
18
17
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
1
3
4
2
6
7
8
We want to loop through matrix_a and return the value of (i) when we have hit all of the numbers in the range 1 to 43.
In the above example we are looking for all the numbers from 1 to 43 and the iteration will end round about position 47 in matrix_a because it is at this point that we hit number '2' which is the last number to complete all numbers in the sequence 1 to 43.
It doesnt matter if we hit several of one number on the way, we count all those - we just want to know when we have reached all the numbers from the check vector or in this example in the sequence 1 to 43.
Ive tried something like:
completed = []
for i = 1:43
complete(i) = find(matrix_a(:,1) == i,1,'first')
end
but not working.
Assuming A as the input column vector, two approaches could be suggested here.
Approach #1
With arrayfun -
check_vector = [1:43]
idx = find(arrayfun(#(n) all(ismember(check_vector,A(1:n))),1:numel(A)),1)+1
gives -
idx =
47
Approach #2
With customary bsxfun -
check_vector = [1:43]
idx = find(all(cumsum(bsxfun(#eq,A(:),check_vector),1)~=0,2),1)+1
To find the first entry at which all unique values of matrix_a have already appeared (that is, if check_vector consists of all unique values of matrix_a): the unique function almost gives the answer:
[~, ind] = unique(matrix_a, 'first');
result = max(ind);
Someone might have a more compact answer but is this what your after?
maxIndex = 0;
for ii=1:length(a)
[f,index] = ismember(ii,a);
maxIndex=max(maxIndex,max(index));
end
maxIndex
Here is one solution without a loop and without any conditions on the vectors to be compared. Given two vectors a and b, this code will find the smallest index idx where a(1:idx) contains all elements of b. idx will be 0 when b is not contained in a.
a = [ 1 4 3 5 6 7 8 9 10 11 12 13 14 16 15 18 17 19 20 21 22 23 24 25 26 ...
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 1 3 4 2 6 7 8 50];
b = 1:43;
[~, Loca] = ismember(b,a);
idx = max(Loca) * all(Loca);
Some details:
ismember(b,a) checks if all elements of b can be found in a and the output Loca lists the indices of these elements within a. The index will be 0, if the element cannot be found in a.
idx = max(Loca) then is the highest index in this list of indices, so the smallest one where all elements of b are found within a(1:idx).
all(Loca) finally checks if all indices in Loca are nonzero, i.e. if all elements of b have been found in a.

Generating matlab array from two arrays

I have two large arrays which I will illustrate using the following examples.
The first array A is:
[ 1 21;
3 4;
4 12;
5 65 ];
The second array B is:
[ 1 26;
31 56;
4 121;
5 54 ];
I want to obtain the final array C as following:
[ 1 21 26;
4 12 121;
5 65 54];
i.e. use the common elements of first column from A and B to sieve out the rows I want to extract from arrays A and B and generate C.
Use the two-output vesion of ismember:
[ii jj] = ismember(A(:,1), B(:,1));
C = [ A(ii,:) B(jj(ii),2) ];
Note that in the second line ii is a logical index, whereas jj(ii) is a standard (integer) index.
bsxfun approach -
%// Slightly bigger and not-so-straightforward example to avoid any confusion
A =[ 1 21;
3 4;
4 12;
8 10
5 65]
B = [ 1 26;
31 56;
4 121;
5 54
9 99
8 88]
binmat = bsxfun(#eq,A(:,1),B(:,1).');%//'
[v1,v2] = max(binmat,[],2);
C = [A(any(binmat,2),:) B(nonzeros(v1.*v2),2:end)]
Output -
A =
1 21
3 4
4 12
8 10
5 65
B =
1 26
31 56
4 121
5 54
9 99
8 88
C =
1 21 26
4 12 121
8 10 88
5 65 54

Resources