% This is the main function of my graphical user interface program
%
%
%
% Get some parameters from the user's screen
set(h,'Pointer','watch');
datafilesnum = str2num(get(filenum2,'String'));	%get parameters from screen
datapatho = get(filepatho2,'String');
datapathc = get(filepathc2,'String');
datafiles = get(filedata2,'String');

if strcmp(get(o3000,'Checked'),'on'),
	NPTS = 3000;
else
	NPTS = 6000;
end

LAGMAX = 1000;		%This is the number of data points in the resulting df files (ie 10sec worth)
	X = zeros(1,LAGMAX +2); 
	Y = zeros(1,LAGMAX +2);
	data1 = strtok(datafiles);
for i=1:datafilesnum,	%Loop through all datafiles: load, calculate correlations, average all trials
		data = strtok(datafiles);
		load([datapatho,data,'.cp'])
		datafiles = strrep(datafiles,strtok(datafiles),' ');
			

	temp = eval(data);			
	
	[a b] = size(temp);
	if a ~= NPTS,
		error('Your files do not contain the data points specified');
	end

	tempX = temp(1:NPTS,1);			
	tempY = temp(1:NPTS,2);				
	tempX = tempX';				
	tempY = tempY';
	tempX = tempX - mean(tempX);
	tempY = tempY - mean(tempY);
	

%	corrX = acf(tempX,NPTS,LAGMAX);		%This acf function is a c program written with the help of Michael Lauk
%	corrY = acf(tempY,NPTS,LAGMAX);		%This c function does not run as fast anymore and has been commented from this program

	corrX = xcorr(tempX,tempX,'biased');
	fix = corrX;	%calculate correlations
	corrX = corrX(NPTS:NPTS+LAGMAX +1);	%take last half of symmetric corrX
	corrX = 2*var(tempX) - 2.*corrX;

	corrY = xcorr(tempY,tempY,'biased');	%Repeat for Y
	corrY = corrY(NPTS:NPTS+LAGMAX+1);
	corrY = 2*var(tempY) - 2.*corrY;

%	corrX = corrX(LAGMAX + 1:2*LAGMAX +2);	%Only needed if acf c program used
%	corrY = corrY(LAGMAX + 1:2*LAGMAX +2);
%	corrX = 2*var(tempX) - 2.*corrX;
%	corrY = 2*var(tempY) - 2.*corrY;


	X = X + corrX;				%Add all correlated files
	Y = Y + corrY;
	if i == 5,
		fprintf('\n\nAre we having fun yet????\n');
	end
end
	X = X./datafilesnum;			%average all files
	Y = Y./datafilesnum;
	R = X + Y;
	
%save .df data
	fprintf('\nWriting data to file: %s\n',[data1,'.df']);
	fid = fopen([datapathc,data1,'.df'],'wt');
	Z = [X; Y];
	fprintf(fid,'%f %f\n',Z);
	fclose(fid);

%find fit times
	filt = firls(40,[0 1],[1 1],'differentiator');	%find coefficients for filter and filter data twice for second derivativ
	derdata = filter(filt,1, X);			
	der2data = filter(filt,1,derdata);		
	[fitTimeX,critpointX] = findtime(der2data);     %Use my findtime.m function to find the critical points and times used to fit the regression lines
	
	derdata = filter(filt,1, Y);			%Repeat for Y	
	der2data = filter(filt,1,derdata);
	[fitTimeY, critpointY] = findtime(der2data);

	derdata = filter(filt,1, R);			%Repeat for R		
	der2data = filter(filt,1,derdata);
	[fitTimeR, critpointR] = findtime(der2data);	 
	
%fit linear lines
	tau  = 0.01:.01:(LAGMAX +2)*.01;     

	tau1Xplot = fitTimeX(1):.01:fitTimeX(2)+1.2;		%These extra variables are to plot the lines longer than the times used to fit the curves
	tau2Xplot = fitTimeX(3)-1.2:.01:fitTimeX(4);		
	fitcoeffx1 = polyfit(fitTimeX(1):.01:fitTimeX(2),X(fitTimeX(1)*100:fitTimeX(2)*100),1);
	linex1 = polyval(fitcoeffx1,tau1Xplot);

	fitcoeffx2 = polyfit(fitTimeX(3):.01:fitTimeX(4),X(fitTimeX(3)*100:fitTimeX(4)*100),1);
	linex2 = polyval(fitcoeffx2,tau2Xplot);

	tau1Yplot = fitTimeY(1):.01:fitTimeY(2)+1.2;
	tau2Yplot = fitTimeY(3)-1.2:.01:fitTimeY(4);
	fitcoeffy1 = polyfit(fitTimeY(1):.01:fitTimeY(2),Y(fitTimeY(1)*100:fitTimeY(2)*100),1);
	liney1 = polyval(fitcoeffy1,tau1Yplot);

	fitcoeffy2 = polyfit(fitTimeY(3):.01:fitTimeY(4),Y(fitTimeY(3)*100:fitTimeY(4)*100),1);
	liney2 = polyval(fitcoeffy2,tau2Yplot);

	tau1Rplot = fitTimeR(1):.01:fitTimeR(2)+1.2;
	tau2Rplot = fitTimeR(3)-1.2:.01:fitTimeR(4);
	fitcoeffr1 = polyfit(fitTimeR(1):.01:fitTimeR(2),R(fitTimeR(1)*100:fitTimeR(2)*100),1);
	liner1 = polyval(fitcoeffr1,tau1Rplot);

	fitcoeffr2 = polyfit(fitTimeR(3):.01:fitTimeR(4),R(fitTimeR(3)*100:fitTimeR(4)*100),1);
	liner2 = polyval(fitcoeffr2,tau2Rplot);

%fit log-log lines
	X(1) = X(1) + .00001;
	Y(1) = Y(1) + .00001;
	R(1) = R(1) + .00001;
	logX = log10(X);
	logY = log10(Y);
	logR = log10(R);
	

	fitcoeffx1log = polyfit(log10(fitTimeX(1):.01:fitTimeX(2)),logX(fitTimeX(1)*100:fitTimeX(2)*100),1);
	linex1log = (tau1Xplot.^fitcoeffx1log(1))*10^fitcoeffx1log(2);

	fitcoeffx2log = polyfit(log10(fitTimeX(3):.01:fitTimeX(4)),logX(fitTimeX(3)*100:fitTimeX(4)*100),1);
	linex2log = (tau2Xplot.^fitcoeffx2log(1))*10^fitcoeffx2log(2); 

	fitcoeffy1log = polyfit(log10(fitTimeY(1):.01:fitTimeY(2)),logY(fitTimeY(1)*100:fitTimeY(2)*100),1);
	liney1log = (tau1Yplot.^fitcoeffy1log(1))*10^fitcoeffy1log(2);

	fitcoeffy2log = polyfit(log10(fitTimeY(3):.01:fitTimeY(4)),logY(fitTimeY(3)*100:fitTimeY(4)*100),1);
	liney2log = (tau2Yplot.^fitcoeffy2log(1))*10^fitcoeffy2log(2);

	fitcoeffr1log = polyfit(log10(fitTimeR(1):.01:fitTimeR(2)),logR(fitTimeR(1)*100:fitTimeR(2)*100),1);
	liner1log = (tau1Rplot.^fitcoeffr1log(1))*10^fitcoeffr1log(2);

	fitcoeffr2log = polyfit(log10(fitTimeR(3):.01:fitTimeR(4)),logR(fitTimeR(3)*100:fitTimeR(4)*100),1);
	liner2log = (tau2Rplot.^fitcoeffr2log(1))*10^fitcoeffr2log(2);

figure;
plotgr;				%This plotgr.m script file graphs what is specified by the user
axis([0 10 0 max(R)+10]);
sparam;				%This sparam.m script file prints the numerical data if requested


%Save parameters in file with .dfp extension (for df parameters)

	fprintf('\nWriting parameters to file: %s\n',[data1,'.dfp']);
	fid = fopen([datapathc,data1,'.dfp'],'wt');
	param = [critpointX*.01; X(critpointX); fitcoeffx1(1)/2; fitcoeffx2(1)/2; fitcoeffx1log(1)/2; 			fitcoeffx2log(1)/2; critpointY*.01; Y(critpointY); fitcoeffy1(1)/2; fitcoeffy2(1)/2; 			fitcoeffy1log(1)/2; fitcoeffy2log(1)/2; critpointR*.01; R(critpointR); fitcoeffr1(1)/2; 			fitcoeffr2(1)/2; fitcoeffr1log(1)/2; fitcoeffr2log(1)/2];  
	fprintf(fid,'%f\n',param);
	fclose(fid);
set(h,'Pointer','arrow');

%This .dfp file is saved as follows: X critical point lag time
%				     The value at this lag time
%				     The slope/2 for the short term linear X
%				     The slope/2 for the long term linear X
%				     The slope/2 for the short term log X
%				     The slope/2 for the long term log X
% This pattern repeats for the Y values and finally for the R values where each
% parameter is on a separate line. (total of 18 parameters)!