How can I vectorize copying one struct to another?

I have

p.Results.events(i)

ans =

struct with fields:

    message: {16×1 cell}
       time: [16×1 uint32]

and

obj.data{1}

ans =

1×70 struct array with fields:

    gx
    gy
    time
    events

where for example

obj.data{1}(1).gx(1:10)

ans =

532.4000  532.3000  532.2000  531.8000  531.1000  530.2000  528.9000  527.6000  527.3000  527.9000

I was able to do this using

                [obj.data{1}.gx] = p.Results.x{:}; 
                [obj.data{1}.gy] = p.Results.y{:};
                [obj.data{1}.time] = p.Results.time{:};
                for i = 1:70
                    obj.data{1}(i).events =  p.Results.events(i);
                end

It seems like in theory, copying the struct should be the easiest to vectorize but I've hit a wall. If it helps, I know exactly what the fields of events should be.

the source code is on github in the constructor

Answer 1

[Guillaume]

events = num2cell(p.Results.events);  %convert array to cell array so that it can then be converted to comma separated list
[obj.data{1}.events] = events{:};  %expand cell array into comma separated list and deal into outputs

However, your hierarchy of non-scalar structure within a cell array within a structure can easily be confusing to the reader of your code. You may want to simplify that a bit.

[Ali1373]

Ah have to convert it, thank you!

And I wish I could simplify, unfortunately I have to make my data a certain shape to work with existing scrips.

[Ali1373]

Follow up question,

what would you recomend for

 for i=1:obj.num_trials
       obj.data{1}(i).num_samples = length(obj.data{1}(i).gx);
 end

considering that each vector has a different length?

[Guillaume]

The for loop is probably the fastest. Another option would be:

l = cellfun('length', {obj.data{1}.gx}, 'UniformOutput', false);  %force cell output
[obj.data{1}.num_samples] = l{:};

But cellfun is just a loop in disguise. Note that the cellfun('length', ...) syntax is deprecated and the modern way of doing the same would be:

l = cellfun(@numel, {obj.data{1}.gx}, 'UniformOutput', false);

which will be even slower.

[Ali1373]

makes sense, thank you for the replies

[Csaba Csupernyák]

@Guillaume, could you please explain more elaborately why do you need to convert the array of a field into a cell array, and then put the target field into square brackets?