Creating a more complex data structure in Matlab?

问题:

I need some help with creating a data structure in Matlab. Until now I needed the following:

string1 value1
string2 value2
string3 value3

and so on. I used a structure for this:

mystruct = struct('mystrings', {}, 'myvalues', {});

Now additional to string and values I now need to assign several arrays (column vectors, containing only numbers) to my string-value pairs. It can be different, how many arrays are assigned to a string-value pair, e.g.

string1 value1 [1;2] [1;3]
string2 value2 [9;10]
string3 value3 [3;4] [2;9] [0;3]

I don't know how to create such a data structure. It's probably not a problem for me to get rid of a structure at all if it's the wrong data structure now. I need help on creating the new data structure, though. Thanks for any help :-)

回答1:

There are always a lot of options when considering a "right" data structure. Some options:

1. You could assign a cell array to each field of your structure.
   You would write to such a structure like this:

   data.field1{1} = 'a string';
   data.field1{2} = [1 2 3 4];
   data.field1{3} = [5:2:10];
   data.field2{1} = 'another string'
   

   and you would read it like this:

   allValueInACellArray = data.field1;
   onlyTheThirdValue = data.field1{3};
   

2. You could use a nested structure.
   Then a write looks like this:

   data.field1.name = 'some name';
   data.field1.firstarray = [1 2 3 4];
   data.field1.secondarray = [5:2:10];
   data.field2.name = 'another name';
   

   and reads look like this:

   justTheFirstName = data.field1.name;
   onlyTheSecondArray = data.field1.firstArray;
   

3. Another construct I use a lot is an array of structures.
   Combining this with a cell-valued field, a write looks like:

   data(1).name = 'some name';
   data(1).arrays = {[1 2 3 4]  5:2:10};
   data(2).name = 'another name';
   data(2).arrays = {[5 6 7 8]  6:3:12};
   

There is usually not a right answer to this. For small programs it doesn't really matter, you should just choose whatever feels most natural to you. For performance-limited applications you need to consider things like efficient memory allocation, and fast access to data in the manner which you usually need to access it.