Unique function with rows and 'first' as arguments

Hello guys!

I have a 90002x6 matrix called "timeVec" with the following format:

2023 2 6 0 0 0

2023 2 6 0 0 0.01

2023 2 6 0 0 0.02

...

...

2023 2 6 0 15 0.01

I am trying to understand the following command:

[~,tInd,~] = unique(timVec(:,4:5),'rows','first')

The resulting tIndi s a 16x1 matrix with the following values:

1

6001

12001

18001

24001

...

...

90001

Thanks!

Answer 1

[Walter Roberson]

The code is locating the first row in which each combination of hour and minute occurs.

There are other implementations, such as looking at whether diff(timVec(:,4)*60+timVec(:,5)) is non-zero . Unless, that is, the entries are out of order -- if they are potentially out of order you should probably also be using the 'stable' option of unique()

[Akis Drosinos]

thanks for the reply, much appreciated. Do you understand where the 6001, 12001, etc. come from?

[Steven Lord]

Row 1 is the first row of timeVec whose 4th and 5th elements is not the same as any previous row.

Row 6001 is the second row of timeVec whose 4th and 5th elements is not the same as any previous row. If I had to guess based on your pattern I'd say it contains [2023 2 6 0 1 0].

Row 12001 of timeVec is likely [2023 2 6 0 2 0].

[Akis Drosinos]

spot on!

Each timestamp is 10 ms, so we need 6000 rows for one second. Column 5 shows the minutes and, as you very well said, row 1 shows the first unique input, so row 1 + 6000 will show the next and then 6001 + 6000 the next and so on!

thanks, it all makes sense now!