I have a system of n equations with m variables. Can SMath solve it approximately? Like, for example, MathCAD can with MinErr function.

Wrote

I have a system of n equations with m variables.
Can SMath solve it approximately ?
Like, for example, MathCAD can with MinErr function.

Answer is YES as long as your "m variables" are "m parameters"
and that you set a system of "m equalities" among the 'n' data sets
That's how it works in Smath 6179 previous to 6671.

1. The pure parabolic example can be made slightly noisy to reflect
experimental data collected. In that case, al_nleqsolve will rougly
get passage at the support points. Further, Genfit Conjugate Gradient
will refine on SSD.

2. The other project concludes a long collaboration in this Forum.
From a 7 parameters model, it like averages a set of 7 parameters
that will satisfy the 9 sets of data. Unfortunately, the data sets
don't belong to an homogeneous family.

All in all: Smath has done same as Mathcad 11 for > 100 fitting
sessions. In many of these sessions Smath is declared superior to MCD.

Jean

Genfit al_ nleqsolve DIDACTIC.sm (78 KiB) downloaded 48 time(s).
Solve al_nleqsolve Ber7_Nicolas [6179] RECONCILIATION Copy.sm (80 KiB) downloaded 52 time(s).

Wrote

1. The pure parabolic example can be made slightly noisy to reflect
experimental data collected. In that case, al_nleqsolve will rougly
get passage at the support points. Further, Genfit Conjugate Gradient
will refine on SSD.

This is not true: al_nleqsolve will not roughly find solutions.
Probably caused by the hyper-reflexive model.
Genfit will find adequate parameters to the noisy data set,
but diverges at the 3rd iterate.

Jean

Genfit al_ nleqsolve DIDACTIC.sm (84 KiB) downloaded 47 time(s). ... updated