MATLAB: How to generate unique random samples for the following scenario

sampling

I have rectangular formation of 208 bolts( 8 along the height and 28 along the length). I want to generate a random sample set of 10000 points which would give me different combinations of bolt failures. There is no restriction on the number of bolts that can fail.

Best Answer

This will do it:

for k = 1 : 1000
  % Get the number of failures for this row.
  numFailures = randi(208);
  % Get the failed indexes.
  failedIndexes{k} = randperm(208, numFailures);
end
% Print to command window.
celldisp(failedIndexes);

Because it's a different number of failures each time, I used a cell array. You could use a regular array if you want - just make it big enough and set non-failures to nan or 0 or something.

Related Solutions

MATLAB: Adding some changes and code rewriting

clc
B=1000; L=2000; Sc=1.1; Sq=1.1; Sgama=.8;
nsamples=10000;
rmem = zeros(3,nsamples);
for i=1:nsamples
  while true
    C = normrnd(620,147.64);
    if C<=450 || C >= 800 || ismember(C,rmem(1,1:K-1)); continue; end
    gama = normrnd(1.96,0.02);
    if gama <= 1.92 || gama >= 1.98 || ismember(gama,rmem(2,1:K-1)); continue; end
    fi = normrnd(3.76,1.1034);
    if fi <= 2.7 || fi >= 16.3 || ismember(fi,rmem(3,1:K-1)); continue; end
    rmem(:,K) = [C; gama; fi];
    break
  end
    Nq = tan((pi/4)+(pi*fi/360)) * tan((pi/4)+(pi*f/360)) * exp(pi*tan(fi*pi/180));
    Nc = (Nq-1)*cot(fi*pi/180);
    Ngama = 2*(Nq+1)*tan(fi*pi/180);
    qult(K) = (C*Nc*Sc)+(384*Nq*Sq)+(980*Ngama*Sgama);
  end

I am not clear as to why you want to check for duplicates: it isn't that duplicates are impossible, but they are relatively rare at 53 bit precision. In the 500000 normrnd(620,147.6) that I generated as a test, the two closest values were a good 2.9899638320785e-11 apart, which was 7 clear bits greater resolution than the largest value generated. I calculate that a 50% chance of a duplicate does not arise until slightly under 112 million samples.

If you eliminate the checks for duplicates or generate a few extra points and discard the duplicates and any extras afterwards, then the code can be made considerably more efficient.

MATLAB: Weibull 0 time failures

I assume you were trying to fit the data using something like

parmhat = wblfit(data);

I tried a few experiments and it doesn't seem to matter whether you replace the zeros by 0.001, 0.1, or just leave them off, as in the example below. I also did a quantile-quantile plot, which shows that the Weibull distribution agrees pretty well with the data except for the zeros and the largest values.

[parmhat,parmci] = wblfit(data(10:end));
syntheticData = wblrnd(parmhat(1),parmhat(2),length(data),1);
qqplot(data,syntheticData)

Best Answer

Related Solutions

MATLAB: Adding some changes and code rewriting

MATLAB: Weibull 0 time failures

Related Question