Preallocate varible before run (the pc runs out of memory )

Hello everyone I hope you can help me.

This code evaluates two variables: and . The length of each variable is 300.

When I run the code my PC runs out of memory and hangs. Can I optimize the code or free up memory while it runs?

phiH = (0:0.0033:1);% axis x
muZ = (0:0.0033:1);% axis y

%phiH = linspace(1.e-14,1,300);%eje x
%muZ = linspace(1.e-14,1,300);%eje y

LphiH = length(phiH);
LmuZ = length(muZ);
%
dotarray=cell(LphiH, LmuZ, 1);
parfor i= 1:LphiH
   for j = 1:LmuZ
        [H, ZH_H] = func_phiH_muZ(phiH(i),muZ(j));
        if H(end-10:end)>2
            if ZH_H(end-10:end)>10
                dot=45; %suvirval
            else
            dot=20; %e out zoospore
            end
        else
        dot=1; % total extinc
        end
        dotarray(j,i)={dot};
    end
end
dotmatrix=cell2mat(dotarray(1:length(phiH), 1:length(muZ)));
%job1 = batch('val_phiH_phiH_22jul2021','CaptureDiary', true, 'Pool',4);
%load(job1)
%colormap(hsv);
image(dotmatrix, 'XData', phiH, 'YData', muZ);

I thank you for the help.

[Walter Roberson]

Why are you creating dotarray as a cell, instead of as a numeric array?

[Alvaro Sepulveda]

Hi, I want to create an image like this. I don't know how to improve memory filling

[Cris LaPierre]

Please share your code for func_phiH_muZ

[Alvaro Sepulveda]

Hi @Cris LaPierre the code for function is:

function [H, ZH_H] = func_phiH_muZ(phiH,muZ)
%function main
    options = odeset('RelTol',1e-6,'AbsTol',1.0e-6,'Stats','on'); 
    tspan = (0:0.1:2000);
%================================================================================
%-------------------- Numeric solution ode15 --------------
%================================================================================
    x0 = [50 50 100 100 500]; % intitials conditions
    [t,xa] = ode15s(@bd_host_ode,tspan, x0, options);
%================================================      
%----- Solutions of model -----
%================================================
     H = xa(:,1); % 
     h = xa(:,2); % 
     %
     ZH = xa(:,3); % 
     Zh = xa(:,4); % 
     Z = xa(:,5); % 
     %
     ZH_H = xa(:,3)'./xa(:,1)';
     Zh_h = xa(:,4)'./xa(:,2)';
%================================================

     
%end
% Function that describe of ODES model
function dxdt = bd_host_ode(t, x)
    %==================================
    %**********************************
    % ---- first set of parameters ----
    %**********************************
    %==================================
    %
    % 
    ah = 0.36; % 
    %
    % 
    bH = 0.085; % 
    bh = 0.85; %  
    %
    % 
    Kbh = 9.7e-04;%
    %
    %
    Kh = 9.7e-04;%

    %=====================================
    %*************************************
    % ----second set of parameters ----
    %*************************************
    %=====================================
    %
    % 
    alphaH = 1.e-5; % 
    alphah = 1.e-5; % 
    %
    % 
    lambdaH = 8.e-1; %
    lambdah = 8.e-1; %
    %
    % 
    muH = 7.e-3; %
    muh = 7.e-3; %
    %
    %muZ = 0.80; % 
    %
    betaH = 0.0125; %  
    betah = 0.0125;  %  
    %
    %phiH = 1.e-14; % 
    phih = 69.e-4; % 
    %
    %===============================================
    %***********************************************
    % ---- other values  ----
    %***********************************************
    %===============================================
    % 
    tmax=27;    tmin=4;    plumax=200;     plumin=10;
    %
    S = 75; % 
    %=======================================
    % 
    %=======================================
    %
    aHmin = 3.8; % 
    %
    aHmax = 48.0; % 
    %
    %===================================================
    %====================================================
    %
    rHmin = 0.0; % 
    %
    rHmax = 10.0;  % 
    %
    rhmin = 0.0; % 
    %
    rhmax = 10.0;  % 
    %
    % 
    pluv = plumax-((plumax-plumin)*(1-cos((2*pi/52)*(t-1.2)+pi))./2); 
    % 
    temp = tmax-((tmax-tmin)*(1-cos((2*pi./52)*(t-1./2)+pi))./2); 
    % 
    rh=rhmax-((rhmax-rhmin)/(tmax-tmin))*(temp-tmin); 
    %
    rH=rHmax-((rHmax-rHmin)/(tmax-tmin))*(temp-tmin);
    %
    aH = aHmax-((aHmax-aHmin)/(S-plumin))*(pluv-plumin); 
    aH = max(aH,0);
    %
    dxdt = zeros(5,1);
    dxdt(1) = ah * exp(-Kh * x(2)) * x(2) - bH * x(1) - alphaH * x(3);
    dxdt(2) = aH * x(1) - (ah * exp(-Kh * x(2)) + bh - bh * exp(-Kbh * x(2))) * x(2) - alphah * x(4);
    dxdt(3) = x(5) * betaH * x(1)./(x(2)+x(1)) + x(3) * (rH - bH - muH - alphaH * (1 + x(3)/x(1) * (phiH + 1)./ phiH));
    dxdt(4) = x(5) * betah * x(2)./(x(2)+x(1)) + x(4) * (rh - ah * exp(-Kh * x(2)) - bh + bh * exp(-Kbh * x(2)) - muh - alphah * (1 + x(4)/x(2) * (phih + 1)./ phih));
    dxdt(5) = lambdaH * x(3) + lambdah * x(4) - x(5)*( muZ + betaH * x(1)./(x(2)+x(1)) + betah * x(2)./(x(2)+x(1)));
end

end

I'm using matlab version R2016b

[Cris LaPierre]

The script has been running for over 24 hours on my computer, but I have not gotten an out-of-memory error yet. I did make some modifications to the code. I moved your calculation of dot inside the function func_phiH_muZ, and changed what it returns to just dot. I de get a lot of warnings about failure to meet integration tolerances.

Note that your if conditional statements (e.g. H(end-10:end)>2) will only evaluate as true if all 10 values meet the criteria. Depending what you want, you might want to make this conditional more explicit by using the all or any functions.

phiH = (0:0.0033:1);% axis x
muZ = (0:0.0033:1);% axis y

LphiH = length(phiH);
LmuZ = length(muZ);

dotmatrix=ones(LphiH, LmuZ);
parfor i= 1:LphiH
    for j = 1:LmuZ
        dot = func_phiH_muZ(phiH(i),muZ(j));
        dotmatrix(j,i)=dot;
    end
end

image(dotmatrix, 'XData', phiH, 'YData', muZ);


function dot = func_phiH_muZ(phiH,muZ)

options = odeset('RelTol',1e-4,'AbsTol',1.0e-6,'Stats','off');
tspan = (0:2000);

%  solve ode15 --------------
x0 = [50 50 100 100 500];  % intitials conditions
[~,xa] = ode15s(@bd_host_ode,tspan, x0, options);

%----- Solutions of model -----
H = xa(:,1);
ZH_H = xa(:,3)'./xa(:,1)';

if H(end-10:end)>2
    if ZH_H(end-10:end)>10
        dot=45; % suvirval
    else
        dot=20; % e out zoospore
    end
else
    dot=1;  % total extinc
end

%Function that describe of ODES model
    function dxdt = bd_host_ode(t, x)
        %---- first set of parameters ----
        ah = 0.36;
        bH = 0.085;
        bh = 0.85;
        Kbh = 9.7e-04;
        Kh = 9.7e-04;

        %----second set of parameters ----
        alphaH = 1.e-5;
        alphah = 1.e-5;

        lambdaH = 8.e-1;
        lambdah = 8.e-1;

        muH = 7.e-3;
        muh = 7.e-3;

        % muZ = 0.80;

        betaH = 0.0125;
        betah = 0.0125;

        % phiH = 1.e-14;
        phih = 69.e-4;

        %---- other values  ----
        tmax=27;    tmin=4;    plumax=200;     plumin=10;
        S = 75;

        aHmin = 3.8;
        aHmax = 48.0;

        rHmin = 0.0;
        rHmax = 10.0;
        rhmin = 0.0;
        rhmax = 10.0;

        pluv = plumax-((plumax-plumin)*(1-cos((2*pi/52)*(t-1.2)+pi))./2);
        temp = tmax-((tmax-tmin)*(1-cos((2*pi./52)*(t-1./2)+pi))./2);

        rh=rhmax-((rhmax-rhmin)/(tmax-tmin))*(temp-tmin);
        rH=rHmax-((rHmax-rHmin)/(tmax-tmin))*(temp-tmin);

        aH = aHmax-((aHmax-aHmin)/(S-plumin))*(pluv-plumin);
        aH = max(aH,0);

        dxdt = zeros(5,1);
        dxdt(1) = ah * exp(-Kh * x(2)) * x(2) - bH * x(1) - alphaH * x(3);
        dxdt(2) = aH * x(1) - (ah * exp(-Kh * x(2)) + bh - bh * exp(-Kbh * x(2))) * x(2) - alphah * x(4);
        dxdt(3) = x(5) * betaH * x(1)./(x(2)+x(1)) + x(3) * (rH - bH - muH - alphaH * (1 + x(3)/x(1) * (phiH + 1)./ phiH));
        dxdt(4) = x(5) * betah * x(2)./(x(2)+x(1)) + x(4) * (rh - ah * exp(-Kh * x(2)) - bh + bh * exp(-Kbh * x(2)) - muh - alphah * (1 + x(4)/x(2) * (phih + 1)./ phih));
        dxdt(5) = lambdaH * x(3) + lambdah * x(4) - x(5)*( muZ + betaH * x(1)./(x(2)+x(1)) + betah * x(2)./(x(2)+x(1)));
    end
end

[Walter Roberson]

Each element of a cell array has over 100 bytes of overhead, so creating a cell array and storing a single number in each location needs much more memory than using a numeric array.

[Alvaro Sepulveda]

Hi @Walter Roberson and @Cris LaPierre if indeed the pc out without memory. The code tries to check a condition (existence or extinction) on the phase space of the system ode . If anyone knows another way to see it graphically, I am open to suggestions.

Greetings

[Image Analyst]

Maybe you don't need all that resolution. I'm not sure why a 300x300 array ends up using a bazillion terabytes but maybe you can get essentially the same answer with less. What happens if you use, like 20 points? Can you try this:

numPoints = 20;
phiH = linspace(0, 1, numPoints);  % axis x
muZ = linspace(0, 1, numPoints);  % axis y

Does that finish? How long does it take?

[Alvaro Sepulveda]

Hi @Image Analyst Well, 20 points not works for me because the pixel are very big, maybe 200 points can works fine since I need to see smooth frontiers between each state (extiction or survival) in the image.