MATLAB Answers

How to code sparse diagonal matrix of increasing dimensions

28 views (last 30 days)
Leone Dragone
Leone Dragone on 16 Jan 2021
Commented: Leone Dragone on 28 Jan 2021 at 14:35
I need to code a growing matrix via for loop and evaluate the time fo some computations, but when i plot the time implied related to the computation it doesn't respect the intervals in which i want to calculate it. Please help!
I'd be very grateful!
clc
clear all
close all
for n= 500:50:1000
e = ones(n,1);
A = spdiags([-2*e 0*e 6*e 0*e -2*e],-2:2,n,n);
for i = 1:n
bi(i) = -(i+1)^2;
b = bi';
end
tic;
L1 = chol(A,'lower');
y1 = L1\b;
x1 = L1'\y1;
toc;
t1(n) = toc;
A2 = full(A);
tic;
L2 = chol(A2,'lower');
y2 = L2\b;
x2 = L2'\y2;
toc;
t2(n) = toc;
end
abs_plot = linspace(500,1000,1000);
plot(abs_plot,t1,'ro','markerfacecolor','r')
hold on
plot(abs_plot,t2,'bo','markerfacecolor','b')
axis tight

  1 Comment

Jan
Jan on 16 Jan 2021
I do not understand this sentene: " it doesn't respect the intervals in which i want to calculate it"

Sign in to comment.

Answers (1)

Jan
Jan on 16 Jan 2021
Edited: Jan on 16 Jan 2021
Maybe this is not, what you want:
for n = 500:50:1000
tic
t1(n) = toc;
end
Setting t(500) in the 1st step ist not useful. Better:
nList = 500:50:1000
for k = 1:numel(nList)
n = nList(k);
tic
...
t1(k) = toc; % Not t(n)
end
Or the full code without the time wasting "brute clearing header":
function YourFcn % Smarter than "clear all"
nList = 500:50:1000;
for k = 1:numel(nList)
n = nList(k);
e = ones(n,1);
A = spdiags([-2*e 0*e 6*e 0*e -2*e],-2:2,n,n);
b = -(2:n+1).' .^ 2; % Vectorized instead of a for loop
tic;
L1 = chol(A,'lower');
y1 = L1 \ b;
x1 = L1' \ y1;
t1(k) = toc;
A2 = full(A);
tic;
L2 = chol(A2,'lower');
y2 = L2 \ b;
x2 = L2' \ y2;
t2(k) = toc;
end
plot(nList, t1, 'ro', 'markerfacecolor', 'r')
hold on
plot(nList, t2, 'bo', 'markerfacecolor', 'b')
axis tight
end

  1 Comment

Leone Dragone
Leone Dragone on 28 Jan 2021 at 14:35
Thank you very much! This was really helpful, thanks a lot!!

Sign in to comment.

Community Treasure Hunt

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

Start Hunting!

Translated by