How to access elements with same field names in an array of structures without using a loop?

Hello,

Assume we have a cell array of structures:

a.name = 'first';
b.name = 'second';
c.name = 'third';
s = {a; b; c};

Now I want to change the values of fields "name" in some specific structures, for example, in 1 and 3. I know how to do it using a "for"-loop:

ind = [1,3];
replacement = {'whatever', 'word'};
for i = 1:length(ind)
    s{ind(i)}.name = replacement{i};
end

Is there a way to do it without a loop? I have found a similar question, but there the author was interested in extracting the data, not in replacing them. Frankly, I am bad in both writing and reading handles' syntaxis, so maybe this answer can be applied to my problem as well, and I just do not see how. Plus, as was mentioned in the comments, cellfun is not faster than a loop - and, at least for me, not easier to read either.

[Stephen23]

What are "handles" ?

[Paul]

Is there an objection to using a for loop?

[Wassili Dimitriew]

@Stephen23, function handles.

https://nl.mathworks.com/help/matlab/function-handles.html

@Paul, no, I am just curious.

Answer 1

[Stephen23]

"How to access elements with same field names in an array of structures without using a loop?"

There is no way using your data, because you actually showed a cell array of scalar structures (which is not very good data design if all of the scalar structures have the same fields). Much better would be to use a non-scalar structure array:

s(1).name = 'first';
s(2).name = 'second';
s(3).name = 'third';
s.name % checking

ans = 'first'

ans = 'second'

ans = 'third'

in which case your request is certainly possible:

idx = [1,3];
rpl = {'whatever', 'word'};
[s(idx).name] = deal(rpl{:});
s.name % checking

ans = 'whatever'

ans = 'second'

ans = 'word'

For more information:

https://www.mathworks.com/matlabcentral/discussions/tips/847976-tutorial-comma-separated-lists-and-how-to-use-them

https://www.mathworks.com/help/matlab/matlab_prog/access-multiple-elements-of-a-nonscalar-struct-array.html

[Stephen23]

However a cell array of scalar structures (like you showed) will not work:

a.name = 'first';
b.name = 'second';
c.name = 'third';
ThisIsACellArrayNotAStructureArray = {a; b; c};

idx = [1,3];
rpl = {'whatever', 'word'};
[ThisIsACellArrayNotAStructureArray{idx}.name] = deal(rpl{:});

Intermediate brace '{}' indexing produced a comma-separated list with 2 values, but it must produce a single value when followed by subsequent indexing operations.

for the reasons explained here:

https://www.mathworks.com/matlabcentral/discussions/tips/847976-tutorial-comma-separated-lists-and-how-to-use-them/2563711#reply_2563711

(the explanation is generally valid for any nested container arrays, not just structure arrays).

[Wassili Dimitriew]

Thank you for the answer. Unfortunately, the structures in my cell array (the one I am actually working with) are the scalar ones. On top of that, I cannot be sure that all the structures have the same number/positioning of fields; as far as I know, that means I cannot convert my structures into non-scalar ones.

[Stephen23]

"I cannot be sure that all the structures have the same number/positioning of fields"

Only the fieldnames are really relevant.

If you prefer to stick with scalar structures in a cell array then you will have to use an implict or explicit loop.

See also:

https://www.mathworks.com/matlabcentral/answers/12912-how-to-create-an-empty-array-of-structs

[Wassili Dimitriew]

Thank you for the links, I will give them a look.

[dpb]

"I cannot be sure that all the structures have the same number/positioning of fields"

Perhaps instead of the cell array of disparate but apparently somehow related scalar struct, use a nested struct that incorporates the alternate data fields by what identifies the particular quantities? <Address Nested Structures>

Answer 2

[Walter Roberson]

You could possibly work something up using cellfun with a subsasgn call, but it would be ugly and needlessly complex.

[Rik]

And also this suggestion comes with the caveat that it will be slower than a loop, have more overhead, and will be harder to maintain/update or debug.