Homec4science
Diffusion Instrument Control (TekRSA-Freq-vs-Time) Browse

rINSTCONTROL/Fit classes/@MyExponentialFit/MyExponentialFit.m
cc82e2b47ac1TekRSA-Freq-vs-Time

Last Change
Raw File

MyExponentialFit.m
View Options

classdef MyExponentialFit < MyFitParamScaling
methods (Access = public)
function this = MyExponentialFit(varargin)
this@MyFitParamScaling( ...
'fit_name', 'Exponential',...
'fit_function', 'a*exp(b*x)+c',...
'fit_tex', '$$ae^{bx}+c$$',...
'fit_params', {'a','b','c'},...
'fit_param_names', {'Amplitude','Rate','Offset'},...
varargin{:});
end
end
methods (Access = protected)
function calcInitParams(this)
ind=this.data_selection;
x=this.Data.x(ind);
y=this.Data.y(ind);
%Setting upper and lower limits
[amp_max,ind_max]=max(y);
[amp_min,ind_min]=min(y);
this.lim_upper=[Inf,Inf,Inf];
this.lim_lower=-this.lim_upper;
%Fix to avoid unphysical offsets on data where all y values
%exceed 0.
if all(y>0)
this.lim_lower(3)=0;
end
if abs(amp_max)>abs(amp_min)
this.lim_upper(1)=Inf;
this.lim_lower(1)=0;
else
this.lim_upper(1)=0;
this.lim_lower(1)=-Inf;
end
if (ind_max>ind_min && abs(amp_max)>abs(amp_min))...
|| (ind_max<ind_min && abs(amp_max)<abs(amp_min))
this.lim_upper(2)=Inf;
this.lim_lower(2)=0;
else
this.lim_upper(2)=0;
this.lim_lower(2)=-Inf;
end
%Method for estimating initial parameters taken from
%http://www.matrixlab-examples.com/exponential-regression.html.
%Some modifications required to account for negative y values
% y=y-amp_min;
y=y-amp_min+eps;
n=length(x);
y2=log(y);
j=sum(x);
k=sum(y2);
l=sum(x.^2);
r2=sum(x .* y2);
p_in(2)=(n * r2 - k * j)/(n * l - j^2);
p_in(1)=exp((k-p_in(2)*j)/n);
if abs(amp_max)>abs(amp_min)
p_in(3)=amp_min;
else
p_in(3)=amp_max;
end
this.param_vals = p_in;
end
function sc_vals = scaleFitParams(~, vals, scaling_coeffs)
[mean_x,std_x,mean_y,std_y]=scaling_coeffs{:};
sc_vals(2)=vals(2)*std_x;
sc_vals(1)=vals(1)*exp(sc_vals(2)*mean_x/std_x)/std_y;
sc_vals(3)=(vals(3)-mean_y)/std_y;
end
function vals = unscaleFitParams(~, sc_vals, scaling_coeffs)
[mean_x,std_x,mean_y,std_y]=scaling_coeffs{:};
vals(1)=exp(-sc_vals(2)*mean_x/std_x)*sc_vals(1)*std_y;
%Edge case for limits
if isnan(vals(1))
vals(1)=0;
end
vals(2)=sc_vals(2)/std_x;
vals(3)=std_y*sc_vals(3)+mean_y;
end
end
end
c4science · Help