How to do indexing without moving to a new variable ?
5 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Hi,
I have an variable x=[1 2 3 4 5]
I need to find the mean of the worst half ie, (3+4+5)/2. For that I am sorting x
sort(x,'descend') ----> [5 4 3 2 1]
I need to find the mean of the first 3 numbers with only single statement
mean(sort(x,'descend')(1:3,1)) - This is not working, getting the following error
??? Error: ()-indexing must appear last in an index expression.
Appreciate the help
Karthik
0 Kommentare
Antworten (6)
Matt Fig
am 27 Okt. 2012
As far as memory goes, if you are o.k. in making the temporary variable through the other methods suggested here then you are o.k. making the temporary variable and overwriting it immediately. This is how I approach the problem when I don't want memory filled with a bunch of non-useful variables. There is no real advantage memory-wise in doing it in one line, and there may even be a penalty if one is forced to create several intermediate structures like padded arrays or cell arrays.
clear
A = randi(bitmax,1,1e6);
M = sort(A,'descend'); % This will die immediately below.
M = mean(M(1:3)) % your workspace is 'clean' again!
0 Kommentare
Azzi Abdelmalek
am 27 Okt. 2012
sum(sort(x,'descend').*[ones(1,3) zeros(1,2)])/3
1 Kommentar
Azzi Abdelmalek
am 27 Okt. 2012
or using cellfun
cellfun(@(x) mean(x(1:3)),{sort(x,'descend')},'un',0)
Star Strider
am 27 Okt. 2012
My contribution:
mx = max(mean(reshape([sort(x,'descend') 0],3,2)).*[1 0]);
0 Kommentare
per isakson
am 27 Okt. 2012
AFAIK: It is not possible. Why do you need a single statement?
3 Kommentare
Walter Roberson
am 27 Okt. 2012
The array that is produced from the sort() occupies the same amount of memory whether it is passed on or given a name, except for a small number of bytes in the name table. The data block produced in the expression is internally pointed to both ways; when you assign an expression, a copy of the data is not made if instead the data block can be internally pointed to.
Matt J
am 27 Okt. 2012
A slightly more exotic solution using this class
result = mean(flib.S.sort{x,'descend'}(1:3));
Although, I guess if you were allowed to use additional mfiles, there are much more trivial ways...
0 Kommentare
Chris A
am 27 Okt. 2012
Does this work?
m = [5 4 3 2 7 1 6 8 9 10];
sum(sort(m)'.*vertcat(ones(numel(m)-numel(m)/2,1),zeros(numel(m)/2,1))),
It does create another matrix,but it is a single line of code.
Chris
1 Kommentar
Siehe auch
Kategorien
Mehr zu Data Type Conversion finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!