| -rw-r--r-- | help.lua | 2 | ||||
| -rw-r--r-- | help/nlfit.lua | 48 | ||||
| -rw-r--r-- | num.lua | 5 |
@@ -1,5 +1,5 @@ -local help_files = {'graphics', 'matrix', 'iter', 'integ', 'ode'} +local help_files = {'graphics', 'matrix', 'iter', 'integ', 'ode', 'nlfit'} local function help_init( ... ) local REG = debug.getregistry() diff --git a/help/nlfit.lua b/help/nlfit.lua new file mode 100644 index 00000000..5fe6153e --- /dev/null +++ b/help/nlfit.lua @@ -0,0 +1,48 @@ + +local REG = debug.getregistry() +NLFIT = REG['GSL.NLINFIT'] + +local M = { + [num.nlinfit] = [[ +num.nlinfit {n= <int>, p= <int>} + + Create a non-linear fit solver object for a system of dimension "n" + with "p" fitting parameters. +]], + + [NLFIT.set] = [[ +<nlfit>:set(fdf, x0) + + Associate the non-linear fit solver with the user-defined function + fdf and set the initial condition for the fit parameters to x0. + The function fdf will be called in the form + + fdf(x, f, J) + + where x is a column matrix with the fit parameters, f is either + nil or a column matrix of dimension n to store the values f(x_i) - + y_i. J is either nil or a matrix of dimension n x p to store the + Jacobian of f versus the fit parameters: + + J[i, j] = d f_i / d x_j + + The function fdf should set all the elements of f and J if they + are not nil. +]], + + [NLFIT.iterate] = [[ +<nlfit>:iterate() + + Advance the solver of a single step. It does return "continue" if + it did not reach the optimal point and "success" otherwise. +]], + + [NLFIT.test] = [[ +<nlfit>:test(eps_abs, eps_err) + + Check if the the search converged for the given absolute error + eps_abs and relative error eps_rel. +]], +} + +return M @@ -1,6 +1,8 @@ local template = require 'template' +local REG = debug.getregistry() + num = {} function num.ode(spec) @@ -23,7 +25,6 @@ function num.ode(spec) local ode = template.load(method, spec) - local REG = debug.getregistry() REG['GSL.help_hook'].ODE = ode local mt = { @@ -49,6 +50,8 @@ local NLINFIT = { end } +REG['GSL.NLINFIT'] = NLINFIT_METHODS + function num.nlinfit(spec) if not spec.n then error 'number of points "n" not specified' end if not spec.p then error 'number of parameters "p" not specified' end |