Assuming I have 1 x 20 structure arrays, how can obtain the ones that are decreasing element wise?

As mentioned in the title, I have couple of structure arrays, that some of them are

• Increasing, such that [a1, a2, a3......] where a1<a2
• Decreasing, such that [b1, b2, b3......] where b1>b2.

I am only interested in the decreasing arrays and would like to filter out the unnecessary ones. I was thinking of using the function strcmp(), but I am not sure how and where to use it. Please help. Thanks!

[Stephen23]

Are these really structure arrays, or are they actually numeric arrays?

That description them makes it seem like that they are numeric arrays, which are a very different thing from structure arrays. For a start structure arrays do not have any inherent mathematical definition of equality or inequality, so the statement "that some of them are increasing" does not make sense for structures.

[bio lim]

I am sorry for confusion. I have this data which is 1 x 133 struct array with fields f1, f2, and f3. I am only interested in f1 (data.f1), and each array in the first field (f1) is 200 x 1. (For example, data(1).f1 has 200 elements)

Sorry if it is confusing as I have just started learning Matlab. Thank you.

[Stephen23]

Then you need to read this:

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

And in particular you might find this syntax very useful ( data is the structure variable name):

>> [data.field]

is equivalent to

>> [data(1).field, data(2).field,..., data(end).field]

and ditto for a cell array and function calls:

>> {data.field} % creates a cell array
>> some_function(data.field)

This works because for a structure the syntax structure.fieldname creates a comma-separated list, which can be used in many situations.

[bio lim]

Thanks, that was very helpful!

Answer 1

[Stephen23]

It is not very clear what you are trying to achieve. Here are some tools that you might find useful. The function that you should have a play around with is diff: you can use it to locate and identify runs of monotonic increasing/decreasing element values.

To identify monotonic strictly decreasing vectors:

>> vec_dec = [5,4,3,2];
>> all(diff(vec_dec)<0)
ans =
     1

To identify monotonic strictly increasing vectors:

>> vec_inc = [1,2,3,4];
>> all(diff(vec_inc)>0)
ans =
     1

To extract monotonic decreasing elements from a vector:

>> vec_mix = [1,2,3,4,3,2,1];
>> idx = diff(vec_mix)<0;
>> vec_mix([false,idx] | [idx,false])
ans =
   4     3     2     1

[bio lim]

Thank you so much, I will try.

Answer 2

[Walter Roberson]

f1_decreases = arrayfun(@(K) YourStruct(K).f1(1) > YourStruct(K).f1(2), 1:length(YourStruct));
kept_Struct = YourStruct(f1_decreases);

[bio lim]

Thank you very much!