from th =dot= walter42 =at= gmx =dot= de

On the other hand, I think I found the root cause of the misbehavior of the akima algorithm.
It is located in function akima_calc() in the block

if (NE_next == 0.0)
            {
              tL_ip1 = m[i+1]; /* here '+1' was missing; because after adding, the graph looks much better */
            }

All variables related to the next with indices have a '+1' except 'tl_ip1'.
Replacing 'm[i]' by 'm[i+1]' results in a smooth spline curve.

-----------------

sorry for not being exact enough.
I refer to the demo2.c example living in the sub-directory "interpolation".
If the plotutils are available then visualization can be done via

    ./demo2 | graph -T X

With Kind Regards,
  Thomas