how to find the closest value in in an array for a set of values

Hi there

I have used the following commented code which has given me an array of 2x100 double for QTcProb.

Then I have picked 5 values at random in COQTcsample.

%%take 300 random samples from controlall
[randomcontrol, idxrandomcontrol] = datasample (Controlall(1,:), 300, 'replace', false);
missIndex = true(1, size(Controlall, 2)); %%get the position of the values of the unsampled data
missValue = Controlall(1, :);
missValue(idxrandomcontrol) = [];  %%get the values of the unsmapled data
close;
[Prob, QTc] = ksdensity(randomcontrol) %%create ksdensity of the 300 samples
QTcProb= vertcat(QTc,Prob); %%create an 2x100 double with probabilty corresponding to the QTc values

%%take 5 samples from the unsampled data
COQTcsample1 = datasample (missValue, 5, 'Replace', false);

%%compare the 5 sampled values against the QTc values in QTcProb to obtain
%%the probability for each of the samples.

...

I now want to perform an operation to take the 5 values in COQTcsample1 and match them to the closest values of QTc in QTcProb so that I can see what the corresponding probability in QTcProb for each of the 5 values is. (I hope this makes sense?)

Many Thanks

Ross

Answer 1

[Adam]

e.g., using just inputs:

QTcProb = rand(2,100);
COQTcsample1 = [0.2, 0.3, 0.5, 0.7, 0.9];

diffs = QTcProb(1,:) - COQTcsample1';
[d, idx] = min( abs( diffs ), [], 2 );

probs = QTcProb(2,idx);

Note: If you are using an older version of Matlab you will have to use:

diffs = bsxfun( @minus, QTcProb(1,:), COQTcsample1' );

instead of

diffs = QTcProb(1,:) - COQTcsample1';

The bsxfun notation will work in all versions of Matlab.

[Ross Johnston]

Thanks Adam this worked perfectly using the bsxfun notation!