topic how to make a linear polynomial approximation by the least squares method in Intel® oneAPI Math Kernel Library

how to make a linear polynomial approximation by the least squares method

Serjio — Tue, 06 Jul 2021 15:46:08 GMT

Hello.

I have a function f = y (x).

On the interval [a, b] I need to do a linear polynomial approximation of the data.

Linear means that the polynomial is of the first degree.

This means that it is described by two coefficients.

Those. I need to find two coefficients by which I can draw a red line like in the picture.
What MKL library functions can I do this?

Re: how to make a linear polynomial approximation by the least squares method

mecej4 — Tue, 06 Jul 2021 16:50:48 GMT

You asked more or less the same question a month ago:

https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/How-to-make-an-analogue-of-the-polyfit-function-from-Matlab/m-p/1286941

and the question was answered in that thread. In short, use the Lapack routine ?GELS.

Re: how to make a linear polynomial approximation by the least squares method

Serjio — Wed, 07 Jul 2021 15:49:08 GMT

Yes you are right. The problem is I can't figure out how to use the Lapack routine ?GELS.

I have my x[200] and y[200] data.

How do I use my data in LAPACKE_dgels()?

info = LAPACKE_dgels(APACK_ROW_MAJOR,'N',m,n,nrhs,*a,lda,*b,ldb);

And where will the result be? Where will the two coefficients be?

Re: how to make a linear polynomial approximation by the least squares method

mecej4 — Wed, 07 Jul 2021 21:19:13 GMT

See page 221 of this. Change the names of the variables from t to x and b to y.

Re:how to make a linear polynomial approximation by the least squares method

ArpitaP_Intel — Wed, 14 Jul 2021 04:01:27 GMT

Hi,

Glad to know that your issue is resolved. If you need any additional information, please submit a new question as this thread will no longer be monitored.

Thanks!