Using erfcinv and array indices are being rejected as not positive or valid, despite all being positive in the vector

Hello, I am currently working on some research data analysis and have been trying to fit autocorrelated data to a plot with a 95% confidence interval to check the distribution of the residuals. To do this, I'm working from the mathworks autocorrelation page (https://www.mathworks.com/help/signal/ug/residual-analysis-with-autocorrelation.html). When trying to set the confidence interval, I get the error that all array indices must be positive or logical values. Below is the code that I'm using, as well as residual data set.

erfbatp = erfcinv(a.Residuals.Raw);
erfbatp = erfbatp';
erfbatp = erfbatp(~isnan(erfbatp))';
conf99batp = sqrt(2)*erfbatp(2*.05/2);
lconfbatp = -conf99batp/sqrt(length(a.Residuals.Raw));
upconfbatp = conf99batp/sqrt(length(a.Residuals.Raw));

1.52973772211594

1.58151023710729

1.53641175549635

1.68852456597857

1.36494692501592

2.07737261195390

1.50753906916397

2.02256675790035

1.05155195289689

1.40278856349961

2.16079514557325

1.43199717380622

1.48556496430640

1.42827621898581

1.62133741215299

1.34671652045243

1.62813386205574

1.41104099011439

1.50196551118339

1.34604336518968

1.47733944833881

1.19896422993831

1.72322100428785

Above is the residual values in the erfbatp variable after removing NaNs, any advice or information would be appreciated.

Answer 1

[Torsten]

As far as I understand, "erfbatp" is an array of numbers.

So erfbatp(2*.05/2) = erfbatp(0.05) does not make sense, does it ? Valid array indices are 1,2,3,4,...

[John D'Errico]

I might conjecture this

conf99batp = sqrt(2)*erfbatp(2*.05/2);

could possibly have been intended as

conf99batp = sqrt(2)*erfbatp*(2*.05/2);

[Gregory Lang]

Hello, thanks for the help. It seems that it was a missing multiplication sign. I'll flag the error on the mathworks site

[Torsten]

I don't see it is used on the MATLAB site this way. I could only find

conf99 = sqrt(2)*erfcinv(2*.01/2);
lconf = -conf99/sqrt(length(x));
upconf = conf99/sqrt(length(x));

which looks correct.