Conventions regarding the C-version of the function dtrnlsp_solve

zuliani — Wed, 27 Feb 2008 20:41:02 GMT

Hi there.

I am reading the documentation of the C-version of the function dtrnlsp_solve and I would like some help since I find some unconsistencies with the notation.

Here are my questions:

1] fvec is really defined as ( y - f(x) ) (as indicated in page 2649 of the manual) or is it just f(x) ?

2] Let f : R^N ---> R^M, 0 <= m <= M-1and 0 <= n <= N-1. What is the Jacobianfjac supposed to contain? Is the stuff stored in column major order or row major order? Which one of the following conventions is the correct one?

- fjac[ m + n*M ] = frac{ partial f(x) }{ partial x }

- fjac[n + m*N ] = frac{ partial f(x) }{ partial x }

- fjac[ m + n*M ] = frac{ y - partial f(x) }{ partial x }

- fjac[n + m*N ] = frac{ y - partial f(x) }{ partial x }

Thanks a lot for any insight that could avoid me a trial and error boring session ;)

Marco

topic Conventions regarding the C-version of the function dtrnlsp_solve in Intel® oneAPI Math Kernel Library

Conventions regarding the C-version of the function dtrnlsp_solve