How to store the output of multiple for loops in an empty matrix in a particular order?

Hi everyone,

My script consists of 2 loops with multiple conditions. For each iteration, I get an output that consists of 3 rows and 4 columns. I want to save this output in an empty matrix in sequential order. May someone suggest how I can do this?

Here is my attempt

T=[]

for kk=1:2
    s2=t_d(t_d(:,1)>= CE_L(kk) & t_d(:,1)<= CE_M(kk),:);
    s1=v_d(v_d(:,1)>= CE_L(kk) & v_d(:,1)<= CE_M(kk),:);

    for i=0:1:3;
            t1_f= addtodate(CE_L(kk), i, 'day');
            % multiple missing step to make the probelm simple 
            ww=1:length(tzcd);
            ph=abs(phz);
            phase=mean(ph);
            med=median(ph);
            st=std(ph);
            obb=i;
            res=[i phase med st];
            TT=[TT,res];

    end
    
end 

However this save outpout as an array, whereas i need as a matrix. My output is looks like this

 0.0	 166.1	 167.0	 2.3	
 1.0	 169.5	 166.4	 11.0	
 2.0	 160.8	 162.9	 5.7	
 3.0	 157.6	 151.7	 21.1

for each iteration of kk loop, where as each row represent the output of i iteration.

Answer 1

[Alex Hanes]

After each iteration of your for loop over i (for i=0:1:3 ... ), you replace the array, TT, that stores everything sequentially. Consider pre-allocating an array for TT and update after each loop, as you are already doing:

nk = 2; % number of iterations in your kk-loop
ni = 4; % number of iterations in your i-loop
nc = 4; % number of columns from your 'res' each loop

TT = zeros(nk*ni,nc); % pre-allocate

Now, instead of having TT = [TT,res]; in your code, replace it with:

for kk = 1:nk 
    for ii = 0:1:ni-1
        % Stuff
        
        idx = (kk-1)*ni + ii + 1; % calculate index
        TT(idx,:) = res; 
    end
end

[Andi]

@Alex Hanes thanks for help, but still there is this error

Unrecognized function or variable 'linInd'.

Error in untitled7 (line 113)
                    TT(linInd,:) = res;
                    
                    

Here is how i modified the script

clear all
clc

% BELOW SCRIPT IS FOR MULTIPLE DAYS CALCULATION
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

cd_ev=readmatrix('T_time.csv'); % 
ev_time=datenum(cd_ev(:,1),cd_ev(:,2),cd_ev(:,3),cd_ev(:,4),cd_ev(:,5),cd_ev(:,6));
CE_M=ev_time';
for jj=1:29
     b=CE_M(:,jj);
     aa(jj)= addtodate(b, 5, 'day');
     bb(jj)= addtodate(b, -5, 'day');
end 
CE_U=aa;
CE_L=bb;
% % ------------ Data Set 2:  -------------%
s_tid=dlmread('TID.csv', ',', 1, 1);
t1=datenum(s_tid(:,1),s_tid(:,2),s_tid(:,3),s_tid(:,4),s_tid(:,5),s_tid(:,6)); 
t_d=[t1 s_tid(:,7)];

%------------- Data Set 3:  --------------% 
s_ven=dlmread('VENT.csv', ',', 1, 1);

ts=datenum(s_ven(:,1),s_ven(:,2),s_ven(:,3),s_ven(:,4),s_ven(:,5),s_ven(:,6)); 
v_d=[ts s_ven(:,7)];
% ---------- Data selection for a particular event ---------%
%tt=[];


nk = 2; % number of iterations in your kk-loop
ni = 4; % number of iterations in your i-loop
nc = 4; % number of columns from your 'res' each loop
TT = zeros(nk*ni,nc); % pre-allocate

for kk=1:2
    s2=t_d(t_d(:,1)>= CE_L(kk) & t_d(:,1)<= CE_M(kk),:);
    s1=v_d(v_d(:,1)>= CE_L(kk) & v_d(:,1)<= CE_M(kk),:);
    ser1=s1(:, 2)-mean(s1(:,2)); % series 1 minus its mean
    ser2=s2(:, 2)-mean(s2(:,2)); % series 2 minus its mean
    ta=s1(:, 1);
    ta1=(ta-ta(1));
    tb=s2(:, 1);
    tb1=(tb-tb(1));
%     figure(1); 
%     subplot(211); 
%     plot(ta1,ser1,'-b'); 
%     legend('Vent Fluid')
%     title('Vent Fluid Temp'); grid on
%     subplot(212); 
%     plot(tb1,ser2,'r-'); 
%     title('Tidal Height'); grid on
%     xlabel('Time (days)')
%     legend('Tidal')
%     obb=zeros(1,20);        %allocate array for phase angles
%     phase=zeros(1,20);        %allocate array for estimates of dominant frequency


    for i=0:1:3;
             t1_f= addtodate(CE_L(kk), i, 'day');
             t2_f= addtodate(CE_L(kk), i+2, 'day');
             sf2=t_d(t_d(:,1)>= t1_f & t_d(:,1)<= t2_f,:);
             sf1=v_d(v_d(:,1)>= t1_f & v_d(:,1)<= t2_f,:);
             if length(sf1) ~= length(sf2)
                continue;
            end
            S=[sf1 sf2];
            tfa=sf1(:, 1);
            tfa=(tfa-tfa(1));  %time (days) (initial offset removed)
            sff1=S(:, 2)-mean(S(:,2)); % series 1 minus its mean
            sff2=S(:, 4)-mean(S(:,4)); % series 2 minus its mean

            N=length(tfa);
            dt=(tfa(end)-tfa(1))/(N-1); %sampling interval (days)
            fs=1/dt;  fNyq=fs/2;    %sampling frequency, Nyquist frequency (cycles/day)
            f1=1; f2=4;             %bandpass corner frequencies (cycles/day)
            [z,p,k]=butter(4,[f1,f2]/fNyq,'bandpass');  %4th order Butterworth bandpass filter coefficients
            [sos,g]=zp2sos(z,p,k); %sos representation of the filter
            y1=filtfilt(sos,g,sff1);  %apply zero phase filter to ser1
            y2=filtfilt(sos,g,sff2);  %apply zero phase filter to ser2


            %     % %--------- Zero crossing time --------%
% 
            zci = @(v) find(diff(sign(v))>0);  %define function: returns indices of +ZCs
            izc1=zci(y1);                      %find indices of + zero crossings of y1
            izc2=zci(y2);                      %find indices of + zero crossings of y2

            ZeroX = @(x0,y0,x1,y1) x0 - (y0.*(x0 - x1))./(y0 - y1); % Interpolated x value for Zero-Crossing 
            ZT1 = ZeroX(tfa(izc1),y1(izc1),tfa(izc1+1),y1(izc1+1));
            ZT2 = ZeroX(tfa(izc2),y2(izc2),tfa(izc2+1),y2(izc2+1));

                if length(ZT1)==length(ZT2)
            tzcd=ZT1-ZT2;          %zero crossing time differences
                elseif length(ZT1)==length(ZT2)-1
            tzcd=ZT1-ZT2(1:end-1); %zero crossing time differences
                elseif length(ZT1)-1==length(ZT2)
            tzcd=ZT1(2:end)-ZT2;   %zero crossing time differences
                end
            fdom=(length(ZT2)-1)/(ZT2(end)-ZT2(1));
            phz=360*fdom*tzcd;          %phase lag of y1 relative to y2 (degrees)
            ww=1:length(tzcd);
            ph=abs(phz);
            phase=mean(ph);
            med=median(ph);
            st=std(ph);
            obb=i;
            res=[i phase med st];
            for kk = 1:nk 
                for ii = 0:1:ni-1
                    idx = (kk-1)*ni + ii + 1; % calculate index
                    TT(linInd,:) = res; 
                end
            end

    end
    
end 

[Alex Hanes]

I updated my original response just now. linInd should have been idx.

[Andi]

@Alex Hanes thank you! but it still only stored results from last iteration.