Initializing multidimensional array and dynamic expansion
57 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
Sandeep Parameshwara
el 14 de En. de 2020
Hello, I am trying to initialize array 'A' with unspecified dimension and sizes. Then, I proceed to specify the size of each dimension thereby also specifying how many dimensions it has. It looks something like this.
A=[]; %This initialization is wrong, but I want to initialize it somehow
Next, I would like to specify size of dimension and also 'fill entire array with zeros' for preallocation purposes(I have not done that below)
size(A,1)=5; %This means my array A has second dimension of size 5.
size(A,2)=5; %This means my array A has third dimension of size 5.
I have a pattern after this. So I would like to create a 'for' loop like this because number of dimension of my array depends on length of some object 'rho{i}' which is again contained in cell array 'rho' of size 1 by n.
for i=3:(2+n)
size(A,i)=length(rho{i}))
end
So, after this I would like to have 5 dimensional array A(assuming I had 3 objects in my cell array rho) completely filled with zero (pre-allocated)
What I want to learn is how can I automate the indexing here? For ex, if I have a 5D array i dont want to replace all indexing by A(i,i,:,:,:). The code should be generated by just specifying number of objects in my cell array rho. Then I would like to fill my array with numbers. This is just example of 4D array, here I have hardcoded entries for all the dimension. I would like to automate this in the code. Pls note that n, d, k's are all known.
A(1,:,:,:)=[-2*d d zeros(1,n-2) -(k(1)+k(2)) k(2) zeros(1,n-2)];
for i=2:n-1
A(i,i-1,:,:)=d;
A(i,i,:,:)=-3*d;
A(i,i+1,:,:)=d;
A(i,n+i-1,:,:)=k(i);
A(i,n+i,:,:)=-(2*k(i)+k(i+1));
A(i,n+i+1,:,:)=k(i+1);
end
A(n,:,:,:)=[zeros(1,n-2) d -2*d zeros(1,n-2) k(n) -2*k(n)];
Any hint or answer would be hugely helpful. Thank you. If you are unclear about the question, pls ask me!
0 comentarios
Respuesta aceptada
Steven Lord
el 15 de En. de 2020
Build your size vector first. Ideally you have an upper bound on the number of dimensions:
upperBoundOnNdims = 10;
sizeVec = ones(1, upperBoundOnNdims);
Fill in the elements of your size vector however you want.
sizeVec(2) = 3;
sizeVec(3) = input('Enter the size in dimension 3:');
sizeVec(5) = randi([2 10]);
Once your size vector contains the correct values, make an array of the size specified by your size vector. I'm going to make this one with all elements equal to 0, but you could use other functions (ones, rand, etc.) to create it.
A = zeros(sizeVec);
Note that all the elements of sizeVec after the fifth are 1 and that A has five dimensions. Arrays in MATLAB usually have a very large number of trailing singleton dimensions that can be queried but are not displayed.
ndims(A) % 5
size(A, 42) % 1
size(A, 1e6) % 1
3 comentarios
Stephen23
el 15 de En. de 2020
Editada: Stephen23
el 15 de En. de 2020
"So how can I make this automatic? I would like to avoid adding extra colons..."
You can do that very simply using this syntax:
A(1,1,:) = 1;
This works because the last explicit index does not refer to that specific dimension, but in fact refers to all trailing dimensions, collapsed into one. See the blog discussion here:
Más respuestas (1)
Meg Noah
el 14 de En. de 2020
Editada: Meg Noah
el 14 de En. de 2020
If you know the size you want:
A = zeros(nrow,ncol);
or
A = zeros(nrow, ncol1, ncol2,ncol3);
etc.
Can also initialize to all ones.
A = ones(nrow,ncol);
If you want all the initial values to be 5, then
A = 5*ones(nrow,ncol);
If you want all the intial values to be undefined:
A = nan(nrow,ncol);
And there is a function
padarray
that can be used to expand arrays that have already been partially populated.
You can also use vertcat, horzcat, and simply:
A = [A ones(nmorrows,1)]
and so on
2 comentarios
Meg Noah
el 15 de En. de 2020
padarray can be used to extend your array sizes
or you can append or prepend arrays using, for example,
A = [A zeros(1,ncol)]
maybe if you posted your code, and not just a snippet, it woudld be easier to figure out what you're missing.
Ver también
Categorías
Más información sobre Matrix Indexing en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!