0.1 : initial release
0.2:
1. implement correction function for A-method (corr_a_method)
2. Auto_scale: implement correction stage (stage 1)
               implement scaling another tscan data set with
	           already scaled scan stage (stage 4)
3. exp_convolve:
   Handling of NAN is user side problem
   Recommandation
   Set NAN to zero
4. fit_static: Make initial guess more reliable
5. fit_static: provide 
               v: voigt
               g: gaussian
               l: lorenzian
               options for static spectrum fitting
6. fit_tscan: re implement good limit scheme for time constant
7. fit_static, fit_tscan: change marker filling 
7. API change: solve_model does not get time zero parameter
               anymore.
0.3:
API change: solve_model splits to solve_model, compute_model
	       and compute_signal for computation efficiency
	       Let y' = Ay be n dim first order linear ODE with
	       y[0] = y_0.
	       First, solve_model finds eigenvalue, eigenvector and
	       coefficients c for such Initial Value Problem.
	       2-a, compute_model uses eigenvalue,
	       eigenvector and coefficients c obtained from
	       solve_model to compute y
	       2-b, compute_signal evaluates
	       y_signal = convolve(irf, y)

0.3.1:
exp_convolve: bug fix 

0.4:
fit_tscan: Now you can use gaussian, cauchy(lorenzian),
           pseudo voigt profile for irf function
fit_static: improve fitting process
API change: mathfun
Change licence to GPL3 to LGPL3
exe: improve parsing and document

	       
