Generate comma separated list in single line of code?

7 vues (au cours des 30 derniers jours)
Evan
Evan le 13 Juin 2017
Modifié(e) : Stephen23 le 14 Avr 2023
I would like to put a piece of code within an expression which generates a comma separated list, without having to create a dummy variable. Is this possible? I was trying to use subsref(), like this:
subsref(repmat({'A'},1,4),struct('type','{}','subs',{{':'}}))
But this only returns a single output. It seems like subsref can't actually behave like A{:} . Is there another way?
The specific reason I would like to do this is for indexing an array where the dimension is unknown prior to runtime. So instead of (for example)
idx = [{1} repmat({':'},1,ndims(inputData)-1)];
output = A(idx{:});
I could do something like
output = A(<expression>);
  8 commentaires
Stephen23
Stephen23 le 13 Juin 2017
Modifié(e) : Stephen23 le 13 Juin 2017
Don't worry about others. They will cope just fine :)
This is a relatively uncommon topic to deal with, so regarding this thread:
  1. beginners are unlikely to stumble across it.
  2. it documents particular concepts that otherwise do not get discussed much.
Adam
Adam le 13 Juin 2017
Modifié(e) : Adam le 13 Juin 2017
It matters more whether there is a clear and useful answer than whether the initial question might confuse. If it were a question where the answers just add to the confusion then it would be bad, but if the accepted answer (accepting Stephen's answer would be useful!) clears up the confusion then that is fine.
Hence, I removed the flag from the question.

Connectez-vous pour commenter.

Réponse acceptée

Stephen23
Stephen23 le 13 Juin 2017
Modifié(e) : Stephen23 le 13 Juin 2017
"I would like to put a piece of code within an expression which generates a comma separated list, without having to create a dummy variable. Is this possible?"
No.
Read this answer and the comments below for the reasons why.
"But this only returns a single output."
No, actually your code works perfectly and returns all of the elements of the cell array, just as we would expect. You just didn't define output variables for all of the outputs of the function:
>> Z = {'A','B','C','D'};
>> [a,b,c,d] = subsref(Z,struct('type','{}','subs',{{':'}}))
a =
A
b =
B
c =
C
d =
D
Because subsref is a perfectly normal function its outputs are only allocated to multiple output variables when those variables are defined in a comma-separated list as output arguments. Exactly like any other function, we would not expect
max(X)
to return both of its output arguments until we have defined both of them in a comma-separated list:
[val,idx] = max(X)
Exactly the same applies to subsref, because it is a normal function as well, and is consistent with what the documentation states regarding returning multiple function outputs:
This all begs the question: what is {:} if not a function? Answer: I have no idea.
Summary: your question is basically the wrong way around. You asked about why subsref behaves strangely, whereas in fact subsref behaves perfectly normally for a MATLAB function. It is actually {:} that behaves strangely: you should be asking about that!
  3 commentaires
Stephen23
Stephen23 le 13 Juin 2017
Modifié(e) : Stephen23 le 13 Juin 2017
"but I don't know why this couldn't have been implemented."
Because, for the reasons that I gave in my answer, how many outputs a MATLAB function returns is demand driven, i.e. determined by how many output arguments have been specified, and is not driven by how many outputs the function potentially has. This also gives predictable results, for example max can accept two inputs and two outputs, yet because the output is demand driven this
max(max(X))
means the the inner max simply provides one output to the outer max, exactly as we would want (and sometimes use with a matrix). What you are suggesting is that the inner max should force the outer max to accept both of its outputs, giving a mathematically unusual maximum of some indices and some values of X. What you want to do would force MATLAB to make the above equivalent to:
[val,idx] = max(X)
max(val,idx)
Lets have a look at another simple example, finding the highest index of non-zeros in a vector:
>> X = [9,0,5];
>> max(find(X))
ans = 3
But note that find can have one, two, or three outputs (and these outputs also change depending on how many there are... but that is another layer of complexity). Following your suggestion all three of them should be provided to max, which would then be an error. How would that be useful? Even with just two of them would be equivalent to:
>> X = [9,0,5];
>> [R,C] = find(X);
>> max(R,C)
ans =
1 3
I do not see how this would be an improvement on MATLAB. It makes little sense (respectfully).
Evan
Evan le 13 Juin 2017
Modifié(e) : Evan le 13 Juin 2017
"but I don't know why this couldn't have been implemented."
Now I do. Thanks! BTW, no sarcasm intended in the earlier comment.

Connectez-vous pour commenter.

Plus de réponses (3)

Wonsang You
Wonsang You le 13 Juin 2017
Please try the following command.
subsref(repmat({'A'},1,4),struct('type','()','subs',{{':'}}))
  1 commentaire
Stephen23
Stephen23 le 13 Juin 2017
Modifié(e) : Stephen23 le 13 Juin 2017
This answer is irrelevant to the topic at hand: it does not return a comma-separated list as the original question asks about, and even the title clearly states: "Generate comma separated list.."
This code simply returns a cell array, and is equivalent to (:).
The original question asks about generating a comma-separated list, equivalent to {:}.
Someone apparently does not know the difference, and voted for it. Perhaps they liked the pretty output.

Connectez-vous pour commenter.


Walter Roberson
Walter Roberson le 13 Juin 2017
Expand = @(cell) cell{:};
Expand(repmat({'A'},1,4))

Stephen23
Stephen23 le 14 Avr 2023
Modifié(e) : Stephen23 le 14 Avr 2023
A one-line approach that has been possible since R2019b:
struct('x',{'A','B','C','D'}).x
ans = 'A'
ans = 'B'
ans = 'C'
ans = 'D'

Catégories

En savoir plus sur Logical dans Help Center et File Exchange

Tags

Produits

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by