Filter löschen
Filter löschen

Max/Min of nonzero rows/cols

11 Ansichten (letzte 30 Tage)
Edward Umpfenbach
Edward Umpfenbach am 12 Apr. 2012
Kommentiert: Dan am 28 Dez. 2015
I have a Matrix, A.
I want to find the max and min for every row and column in A, excluding the zero entries.
I have coded it using For loops, but it is too slow. I need it to be all vectorized. Any help is greatly appreciated.
  1 Kommentar
Dan
Dan am 28 Dez. 2015
A = matrix with some zero entries . Indzero = find( A== 0); B=A; B(Indzero) = max(max(A))+1; Min(B,[],1) B(Indzero) = min(min(A))-1; Max(B,[],1)
U can use this sort of logic maybe . u can also go at it by gpu processing or mexing for faster results ( this is exactly the type of probelms where u get the best rests with gpus Nd mexing )

Melden Sie sich an, um zu kommentieren.

Akzeptierte Antwort

Richard Brown
Richard Brown am 12 Apr. 2012
B = A;
B(B == 0) = NaN;
colMax = max(B);
colMin = min(B);
rowMax = max(B, [], 2);
rowMin = min(B, [], 2);
edit: missed the bit about the zero entries
  1 Kommentar
Edward Umpfenbach
Edward Umpfenbach am 12 Apr. 2012
Thank you. I didn't think to change the zeros to NaN so that they didn't mess up the min operation.

Melden Sie sich an, um zu kommentieren.

Weitere Antworten (1)

Geoff
Geoff am 12 Apr. 2012
When you say 'for every row and column', do you mean a single max and min for the entire matrix? This would do it:
nzmin = min(A(A ~= 0));
nzmax = max(A(A ~= 0));
[edit]
Note that if you have float data, you might want to consider values 'near' zero:
ztol = eps(max(A(:)));
nz = A < -ztol | A > ztol; % select non-zeros
nzmin = min(A(nz));
nzmax = max(A(nz));
It's refreshing to use a proliferation of 'nz', since it's my nationality =)
  3 Kommentare
Geoff
Geoff am 12 Apr. 2012
My heart's in Dunedin. But I moved up to Auckland to get away from Animation Research Ltd ;-) You're a Canterbury lad, right?
Richard Brown
Richard Brown am 12 Apr. 2012
I am indeed, wobbles and all :)

Melden Sie sich an, um zu kommentieren.

Kategorien

Mehr zu Parallel Computing finden Sie in Help Center und File Exchange

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by