Services & ResourcesWolfram ForumsMathGroup Archive

[Date Index] [Thread Index] [Author Index]

variables in packages

• To: mathgroup at smc.vnet.net
• Subject: [mg25877] variables in packages
• From: Andre Heinemann <a.heinemann at ifw-dresden.de>
• Date: Sat, 4 Nov 2000 02:04:19 -0500 (EST)
• Sender: owner-wri-mathgroup at wolfram.com

Hi mathgroup,

I wrote a small package for my special purposes in
Nonlinear Regression. The code as a function in a notebook
works well, but it dosn't work as a package.
The rules: (BestFitParameters /. nonlinregress) and
(ParameterCITable /. nonlinregress) did not work in the same
way, because NewRegress`Private`BestFitParameters does not exist.
This was the answer of mathematica during a test of my package.
How can I declare BestFitParameters that mathematica knows that
it is a variable in NonlinearRegress usable also from outside
of the package ?

I send the whole code, not because I like to bore you, but may
be the problem is situated in a othe part of the code.

I hope you can help me because I tried a lot of different thinks,
but nothink works.

Thanks,

Andre

__________snip, snip________________________________________



BeginPackage["NewRegress`",
    (* needed for NonlinearRegress *)
    "Statistics`NonlinearFit`",
    (* needed for BooleanSelect*)
"Statistics`DataManipulation`"];

NewRegress::usage = "NewRegress[data, weights,
model_function,
{x,xmin,xmax},allparameter,fitparameter,precisition,options]
returns a list of a NonlinearRegression of the data with the
model_function.
and weights. output[[1]],output[[2]] are for compatibility. output[[3]]
is the full report 3."

Begin["`Private`"]

NewRegress[data_List, gewichte_List, modelFunction_, Xvariable_List,
    parameter_List, fit_List, precision_Integer, opt___] :=
  Block[{errlist = {},
      fitselect = {},
      fitinput = {},
      fixselect = {},
      fixinput = {},
      nonlinregress = {},
      out1 = {},
      useddata = {},
      usedgewichte = {},
      xfree = Xvariable[[1]],
      xmin = Xvariable[[2]],
      xmax = Xvariable[[3]],
      pll, j, i
      },

    For[j = 1; useddata = {}; usedgewichte = {}, j <= Length[data], j++,

      If[data[[j]][[1]] >= xmin && data[[j]][[1]] <= xmax,
        useddata = Join[useddata, {data[[j]]}];
        usedgewichte = Join[usedgewichte, {gewichte[[j]]}]]];

    pll = Length[parameter];

    fitselect = Table[MemberQ[fit, parameter[[i, 1]]], {i, 1, pll}];

    fitinput = BooleanSelect[parameter, fitselect];

    fixselect = Table[Not[fitselect[[i]]], {i, 1, pll}];

    fixinput =
      Table[(BooleanSelect[parameter, fixselect])[[i,
            1]] -> (BooleanSelect[parameter, fixselect])[[i, 2]], {i, 1,

          Length[(BooleanSelect[parameter, fixselect])]}];

    nonlinregress =
      NonlinearRegress[useddata, Evaluate[modelFunction] /. fixinput,
        Xvariable[[1]], fitinput, ShowProgress -> False,
        Weights -> usedgewichte, opt];

    out1 = Join[fixinput, (BestFitParameters /. nonlinregress)];

    For[j = 1; outlist = {}, j <= pll, j++,
      For[i = 1, i <= pll, i++,
        If[parameter[[j]][[1]] === out1[[i]][[1]],
          outlist =
            Join[outlist, {parameter[[j]][[1]] ->
                  SetPrecision[out1[[i]][[2]], precision]}]]]];

    errlist =
      Table[{fitinput[[j, 1]],
          SetPrecision[(ParameterCITable /. nonlinregress)[[1]][[j, 1]],

            precision],
          SetPrecision[
            Max[Abs[(ParameterCITable /. nonlinregress)[[1]][[j,
                    3]] - (ParameterCITable /. nonlinregress)[[1]][[j,
1]]]],
            precision]}, {j, 1, Length[fitinput]}];

    {outlist, errlist, nonlinregress}
    ]

End[]

EndPackage[]

______________snip * snip__________________________




--
+-----------------------------------+-----------------------------+
| email: a.heinemann at ifw-dresden.de |      Andre Heinemann        |
+-----------------------------------+-----------------------------+