Unable to classify the variable 'feq' in the body of the parfor-loop

3 次查看（过去 30 天）
Thierry Rebetez
Thierry Rebetez 2023-4-2
评论： Walter Roberson 2023-4-2
Hi everybody,
I have a code that has the following error: Error: Unable to classify the variable 'feq' in the body of the parfor-loop. I don't understand the issue and how to solve it. Thank you very much for helping me.
% Set simulation parameters
Lx = 50; % length of container in x-direction
Ly = 50; % length of container in y-direction
dx = 1; % grid spacing
dt = 0.1; % time step
c = dx/dt; % speed of sound (?)
tmax = 100; % total simulation time
omega = 1.8; % relaxation parameter
rho0 = 1; % fluid density
mu = 0.1; % fluid viscosity
nu = mu/rho0; % fluid kinematic viscosity
m = 100; % number of floaters
R = 1; % radius of floaters
x0 = rand(m,1)*Lx; % initial x-positions of floaters
y0 = rand(m,1)*Ly; % initial y-positions of floaters
% Set lattice parameters
q = 9; % number of lattice velocities
e = [0,0;0,1;1,0;0,-1;-1,0;1,1;1,-1;-1,-1;-1,1]; % lattice velocities % TODO: check order of arrows
w = [4/9,1/9,1/9,1/9,1/9,1/36,1/36,1/36,1/36]; % lattice weights
% Initialize fluid and floaters
f = zeros(Lx,Ly,q); % distribution function
feq = zeros(Lx,Ly,q); % equilibrium distribution function
rho = ones(Lx,Ly)*rho0; % fluid density
ux = zeros(Lx,Ly); % fluid x-velocity
uy = zeros(Lx,Ly); % fluid y-velocity
x = x0; % floater x-positions
y = y0; % floater y-positions
vx = zeros(m,1); % floater x-velocities
vy = zeros(m,1); % floater y-velocities
% Set up boundary conditions
A = 0.01; % amplitude of Faraday wave perturbation
fFarad = 5; % frequency of Faraday wave perturbation
bc_func = @(t) A*sin(2*pi*fFarad*t); % boundary function
% Set simulation parameters
Lx = 50; % length of container in x-direction
Ly = 50; % length of container in y-direction
dx = 1; % grid spacing
dt = 0.1; % time step
c = dx/dt; % speed of sound (?)
tmax = 100; % total simulation time
omega = 1.8; % relaxation parameter
rho0 = 1; % fluid density
mu = 0.1; % fluid viscosity
nu = mu/rho0; % fluid kinematic viscosity
m = 100; % number of floaters
R = 1; % radius of floaters
x0 = rand(m,1)*Lx; % initial x-positions of floaters
y0 = rand(m,1)*Ly; % initial y-positions of floaters
% Set lattice parameters
q = 9; % number of lattice velocities
e = [0,0;0,1;1,0;0,-1;-1,0;1,1;1,-1;-1,-1;-1,1]; % lattice velocities % TODO: check order of arrows
w = [4/9,1/9,1/9,1/9,1/9,1/36,1/36,1/36,1/36]; % lattice weights
% Initialize fluid and floaters
f = zeros(Lx,Ly,q); % distribution function
feq = zeros(Lx,Ly,q); % equilibrium distribution function
ux = zeros(Lx,Ly); % fluid x-velocity
uy = zeros(Lx,Ly); % fluid y-velocity
x = x0; % floater x-positions
y = y0; % floater y-positions
vx = zeros(m,1); % floater x-velocities
vy = zeros(m,1); % floater y-velocities
% Set up boundary conditions
A = 0.01; % amplitude of Faraday wave perturbation
fFarad = 5; % frequency of Faraday wave perturbation
bc_func = @(t) A*sin(2*pi*fFarad*t); % boundary function
% delete(gcp('nocreate')); % delete any existing parallel pool
% parpool; % start a parallel pool from the Parallel Computing Toolbox
% parpool('local', 12); % 12 is the number of workers (= number of cores)
poolobj = gcp('nocreate'); % get the pool object, and do it avoiding creating a new one.
if isempty(poolobj) % check if there is not a pool.
poolsize = 1;
else
poolsize = poolobj.NumWorkers;
delete( gcp('nocreate')); % delete the current pool object.
end
parpool( poolsize, 'IdleTimeout',Inf); % create a new pool with the previous poolsize and new specs.
% Main simulation loop
% parfor t=0;dt;tmax
% feq = zeros(1,q);
% f = zeros(1,q);
% ux = zeros(Lx,Ly); % fluid x-velocity
% uy = zeros(Lx,Ly); % fluid y-velocity
% x = x0; % floater x-positions
% y = y0; % floater y-positions
% vx = zeros(m,1); % floater x-velocities
% vy = zeros(m,1); % floater y-velocities
parfor t=0:dt:tmax
% Collision step
for i=1:q
rho = ones(Lx,Ly)*rho0; % fluid density
feq(:,:,i) = w(i)*rho.*(1 + 3.*(e(i,1).*ux + e(i,2).*uy)./c + 9./2.*(e(i,1).*ux + ...
e(i,2).*uy).^2./c.^2 - 3/2.*(ux.^2 + uy.^2)./c.^2); % TODO check ^2
end
for i=1:q
f(:,:,i) = (1-omega)*f(:,:,i) + omega*feq(:,:,i) + omega*w(i)*(3*(e(i,1)*ux + e(i,2)*uy)./c + ...
9/2*(e(i,1)*ux + e(i,2)*uy).^2./c^2 - 3/2*(ux.^2 + uy.^2)./c^2);
end
% External forcing
% bc = [1,2,3,4]; % boundary nodes
% ux(bc,:) = bc_func(t); % TODO: What is this supposed to do?
% Streaming step
for i=1:q
f(:,:,i) = circshift(f(:,:,i), e(i,:));
end
% Compute macroscopic quantities
rho = sum(f,3);
ux = sum(f.*reshape(e(:,1),[1 1 q]),3)./rho;
uy = sum(f.*reshape(e(:,2),[1 1 q]),3)./rho;
% Update floater positions and velocities
for i=1:m
ix = max(1, min(floor(x(i)/dx) + 1, Lx));
iy = max(1, min(floor(y(i)/dx) + 1, Ly));
vx(i) = (ux(ix,iy) - vx(i));
vy(i) = (uy(ix,iy) - vy(i));
% ix = floor(x(i)/dx) + 1;
% iy = floor(y(i)/dx) + 1;
% vx(i) = (ux(ix,iy) - vx(i));
% vy(i) = (uy(ix,iy) - vy(i));
x(i) = x(i) + vx(i)*dt;
y(i) = y(i) + vy(i)*dt;
% Check for collisions with walls
if x(i) < R
x(i) = R;
vx(i) = -vx(i);
elseif x(i) + R > Lx
x(i) = Lx - R;
vx(i) = -vx(i);
end
if y(i) < R
y(i) = R;
vy(i) = -vy(i);
elseif y(i) + R > Ly
y(i) = Ly - R;
vy(i) = -vy(i);
end
% Check for collisions with other floaters
for j=1:m
if i ~= j && norm([x(i) y(i)] - [x(j) y(j)]) < 2*R % TODO überprüfen
theta = atan2(y(i)-y(j),x(i)-x(j));
v1 = [vx(i)*cos(theta) + vy(i)*sin(theta) vx(i)*-sin(theta) + vy(i)*cos(theta)];
v2 = [vx(j)*cos(theta) + vy(j)*sin(theta) vx(j)*-sin(theta) + vy(j)*cos(theta)];
v1f = (v1*(R-R) + 2*R*v2)./(R+R);
v2f = (v2*(R-R) + 2*R*v1)./(R+R);
vx(i) = v1f(1)*cos(theta) + v1f(2)*-sin(theta);
vy(i) = v1f(1)*sin(theta) + v1f(2)*cos(theta);
vx(j) = v2f(1)*cos(theta) + v2f(2)*-sin(theta);
vy(j) = v2f(1)*sin(theta) + v2f(2)*cos(theta);
x(i) = x(i) + vx(i)*dt;
y(i) = y(i) + vy(i)*dt;
x(j) = x(j) + vx(j)*dt;
y(j) = y(j) + vy(j)*dt;
end
end
end
% Apply boundary conditions
% Boundary condition for left wall
%% f(1,:,2) = f(2,:,2) - 2*w(2)*rho(1,:).*e(2,1)*feval(bc_func,t); % Option A
f(1,:,2) = f(2,:,2) - 2*w(2)*f(1,:,2).*e(2,1)*feval(bc_func,t); % Option B
f(1,:,2) = f(2,:,2) - 2*w(2)*f(1,:,2).*e(2,1)*feval(bc_func,t);
f(1,:,5) = f(2,:,5) - 2*w(5)*f(1,:,5).*e(5,1)*feval(bc_func,t);
f(1,:,6) = f(2,:,6) - 2*w(6)*f(1,:,6).*(e(6,1)*feval(bc_func,t) + e(6,2)*feval(bc_func,t));
f(1,:,7) = f(2,:,7) - 2*w(7)*f(1,:,7).*(-e(7,1)*feval(bc_func,t) + e(7,2)*feval(bc_func,t));
f(1,:,8) = f(2,:,8) - 2*w(8)*f(1,:,8).*(-e(8,1)*feval(bc_func,t) - e(8,2)*feval(bc_func,t));
f(1,:,9) = f(2,:,9) - 2*w(9)*f(1,:,9).*(e(9,1)*feval(bc_func,t) - e(9,2)*feval(bc_func,t));
% Boundary condition for right wall
f(Lx,:,4) = f(Lx-1,:,4) - 2*w(4)*f(Lx,:,4).*e(4,1)*feval(bc_func,t);
f(Lx,:,8) = f(Lx-1,:,8) - 2*w(8)*f(Lx,:,8).*(-e(8,1)*feval(bc_func,t) - e(8,2)*feval(bc_func,t));
f(Lx,:,9) = f(Lx-1,:,9) - 2*w(9)*f(Lx,:,9).*(e(9,1)*feval(bc_func,t) - e(9,2)*feval(bc_func,t));
f(Lx,:,6) = f(Lx-1,:,6) - 2*w(6)*f(Lx,:,6).*(e(6,1)*feval(bc_func,t) + e(6,2)*feval(bc_func,t));
f(Lx,:,3) = f(Lx-1,:,3) - 2*w(3)*f(Lx,:,3).*(e(3,1)*feval(bc_func,t) + e(3,2)*feval(bc_func,t));
% Boundary condition for bottom wall
f(:,1,3) = f(:,2,3) - 2*w(3)*f(:,1,3).*e(3,2)*feval(bc_func,t);
f(:,1,6) = f(:,2,6) - 2*w(6)*f(:,1,6).*(e(6,1)*feval(bc_func,t) + e(6,2)*feval(bc_func,t));
f(:,1,7) = f(:,2,7) - 2*w(7)*f(:,1,7).*(-e(7,1)*feval(bc_func,t) + e(7,2)*feval(bc_func,t));
f(:,1,2) = f(:,2,2) - 2*w(2)*f(:,1,2).*e(2,2)*feval(bc_func,t);
f(:,1,9) = f(:,2,9) - 2*w(9)*f(:,1,9).*(e(9,1)*feval(bc_func,t) - e(9,2)*feval(bc_func,t));
% Boundary condition for top wall
f(:,Ly,5) = f(:,Ly-1,5) - 2*w(5)*f(:,Ly,5).*e(5,2)*feval(bc_func,t);
f(:,Ly,8) = f(:,Ly-1,8) - 2*w(8)*f(:,Ly,8).*(-e(8,1)*feval(bc_func,t) - e(8,2)*feval(bc_func,t));
f(:,Ly,6) = f(:,Ly-1,6) - 2*w(6)*f(:,Ly,6).*(e(6,1)*feval(bc_func,t) + e(6,2)*feval(bc_func,t));
f(:,Ly,7) = f(:,Ly-1,7) - 2*w(7)*f(:,Ly,7).*(-e(7,1)*feval(bc_func,t) + e(7,2)*feval(bc_func,t));
f(:,Ly,1) = f(:,Ly-1,1) - 2*w(1)*f(:,Ly,1).*e(1,2)*feval(bc_func,t);
f(:,Ly,9) = f(:,Ly-1,9) - 2*w(9)*f(:,Ly,9).*(e(9,1)*feval(bc_func,t) - e(9,2)*feval(bc_func,t));
% Plot fluid and floaters
if mod(t,10) == 0
figure(1)
clf
subplot(2,2,1)
imagesc(ux')
colorbar
title('Fluid x-velocity')
subplot(2,2,2)
imagesc(uy')
colorbar
title('Fluid y-velocity')
subplot(2,2,3)
scatter(x,y,10,'filled')
axis equal
xlim([0 Lx])
ylim([0 Ly])
title('Floaters')
subplot(2,2,4)
quiver(ux',uy')
axis equal
title('Fluid velocity field')
drawnow
end
end

请先登录，再回答此问题。

采纳的回答

Thierry Rebetez
Thierry Rebetez 2023-4-2
编辑：Walter Roberson 2023-4-2
Thank you Walter Robertson for your answer. So I modified the code as follows: I put the initialization of feq, f, etc. after the parfor loop as follows:
parfor t=0:tmax
% Initialize fluid and floaters
f = zeros(Lx,Ly,q); % distribution function
feq = zeros(Lx,Ly,q); % equilibrium distribution function
ux = zeros(Lx,Ly); % fluid x-velocity
uy = zeros(Lx,Ly); % fluid y-velocity
x = x0; % floater x-positions
y = y0; % floater y-positions
vx = zeros(m,1); % floater x-velocities
vy = zeros(m,1); % floater y-velocities
% Collision step
for i=1:q
rho = ones(Lx,Ly)*rho0; % fluid density
feq(:,:,i) = w(i)*rho.*(1 + 3.*(e(i,1).*ux + e(i,2).*uy)./c + 9./2.*(e(i,1).*ux + ...
e(i,2).*uy).^2./c.^2 - 3/2.*(ux.^2 + uy.^2)./c.^2); % TODO check ^2
end
...
The code runs without an error but I don't get any plotting at the end of the code. Why?
Here again the updated code:
% Set simulation parameters
Lx = 50; % length of container in x-direction
Ly = 50; % length of container in y-direction
dx = 1; % grid spacing
dt = 0.1; % time step
c = dx/dt; % speed of sound (?)
tmax = 100; % total simulation time
omega = 1.8; % relaxation parameter
rho0 = 1; % fluid density
mu = 0.1; % fluid viscosity
nu = mu/rho0; % fluid kinematic viscosity
m = 100; % number of floaters
R = 1; % radius of floaters
x0 = rand(m,1)*Lx; % initial x-positions of floaters
y0 = rand(m,1)*Ly; % initial y-positions of floaters
% Set lattice parameters
q = 9; % number of lattice velocities
e = [0,0;0,1;1,0;0,-1;-1,0;1,1;1,-1;-1,-1;-1,1]; % lattice velocities % TODO: check order of arrows
w = [4/9,1/9,1/9,1/9,1/9,1/36,1/36,1/36,1/36]; % lattice weights
% Initialize fluid and floaters
% f = zeros(Lx,Ly,q); % distribution function
% feq = zeros(Lx,Ly,q); % equilibrium distribution function
rho = ones(Lx,Ly)*rho0; % fluid density
% ux = zeros(Lx,Ly); % fluid x-velocity
% uy = zeros(Lx,Ly); % fluid y-velocity
% x = x0; % floater x-positions
% y = y0; % floater y-positions
% vx = zeros(m,1); % floater x-velocities
% vy = zeros(m,1); % floater y-velocities
% Set up boundary conditions
A = 0.01; % amplitude of Faraday wave perturbation
fFarad = 5; % frequency of Faraday wave perturbation
bc_func = @(t) A*sin(2*pi*fFarad*t); % boundary function
% Set lattice parameters
q = 9; % number of lattice velocities
e = [0,0;0,1;1,0;0,-1;-1,0;1,1;1,-1;-1,-1;-1,1]; % lattice velocities % TODO: check order of arrows
w = [4/9,1/9,1/9,1/9,1/9,1/36,1/36,1/36,1/36]; % lattice weights
poolobj = gcp('nocreate'); % get the pool object, and do it avoiding creating a new one.
if isempty(poolobj) % check if there is not a pool.
poolsize = 12;
else
poolsize = poolobj.NumWorkers;
delete( gcp('nocreate')); % delete the current pool object.
end
parpool( poolsize, 'IdleTimeout',Inf); % create a new pool with the previous poolsize and new specs.
parfor t=0:tmax
% Initialize fluid and floaters
f = zeros(Lx,Ly,q); % distribution function
feq = zeros(Lx,Ly,q); % equilibrium distribution function
ux = zeros(Lx,Ly); % fluid x-velocity
uy = zeros(Lx,Ly); % fluid y-velocity
x = x0; % floater x-positions
y = y0; % floater y-positions
vx = zeros(m,1); % floater x-velocities
vy = zeros(m,1); % floater y-velocities
% Collision step
for i=1:q
rho = ones(Lx,Ly)*rho0; % fluid density
feq(:,:,i) = w(i)*rho.*(1 + 3.*(e(i,1).*ux + e(i,2).*uy)./c + 9./2.*(e(i,1).*ux + ...
e(i,2).*uy).^2./c.^2 - 3/2.*(ux.^2 + uy.^2)./c.^2); % TODO check ^2
end
for i=1:q
f(:,:,i) = (1-omega)*f(:,:,i) + omega*feq(:,:,i) + omega*w(i)*(3*(e(i,1)*ux + e(i,2)*uy)./c + ...
9/2*(e(i,1)*ux + e(i,2)*uy).^2./c^2 - 3/2*(ux.^2 + uy.^2)./c^2);
end
% External forcing
% bc = [1,2,3,4]; % boundary nodes
% ux(bc,:) = bc_func(t); % TODO: What is this supposed to do?
% Streaming step
for i=1:q
f(:,:,i) = circshift(f(:,:,i), e(i,:));
end
% Compute macroscopic quantities
rho = sum(f,3);
ux = sum(f.*reshape(e(:,1),[1 1 q]),3)./rho;
uy = sum(f.*reshape(e(:,2),[1 1 q]),3)./rho;
% Update floater positions and velocities
for i=1:m
ix = max(1, min(floor(x(i)/dx) + 1, Lx));
iy = max(1, min(floor(y(i)/dx) + 1, Ly));
vx(i) = (ux(ix,iy) - vx(i));
vy(i) = (uy(ix,iy) - vy(i));
% ix = floor(x(i)/dx) + 1;
% iy = floor(y(i)/dx) + 1;
% vx(i) = (ux(ix,iy) - vx(i));
% vy(i) = (uy(ix,iy) - vy(i));
x(i) = x(i) + vx(i)*dt;
y(i) = y(i) + vy(i)*dt;
% Check for collisions with walls
if x(i) < R
x(i) = R;
vx(i) = -vx(i);
elseif x(i) + R > Lx
x(i) = Lx - R;
vx(i) = -vx(i);
end
if y(i) < R
y(i) = R;
vy(i) = -vy(i);
elseif y(i) + R > Ly
y(i) = Ly - R;
vy(i) = -vy(i);
end
% Check for collisions with other floaters
for j=1:m
if i ~= j && norm([x(i) y(i)] - [x(j) y(j)]) < 2*R % TODO überprüfen
theta = atan2(y(i)-y(j),x(i)-x(j));
v1 = [vx(i)*cos(theta) + vy(i)*sin(theta) vx(i)*-sin(theta) + vy(i)*cos(theta)];
v2 = [vx(j)*cos(theta) + vy(j)*sin(theta) vx(j)*-sin(theta) + vy(j)*cos(theta)];
v1f = (v1*(R-R) + 2*R*v2)./(R+R);
v2f = (v2*(R-R) + 2*R*v1)./(R+R);
vx(i) = v1f(1)*cos(theta) + v1f(2)*-sin(theta);
vy(i) = v1f(1)*sin(theta) + v1f(2)*cos(theta);
vx(j) = v2f(1)*cos(theta) + v2f(2)*-sin(theta);
vy(j) = v2f(1)*sin(theta) + v2f(2)*cos(theta);
x(i) = x(i) + vx(i)*dt;
y(i) = y(i) + vy(i)*dt;
x(j) = x(j) + vx(j)*dt;
y(j) = y(j) + vy(j)*dt;
end
end
end
% Apply boundary conditions
% Boundary condition for left wall
%% f(1,:,2) = f(2,:,2) - 2*w(2)*rho(1,:).*e(2,1)*bc_func(t); % Option A
f(1,:,2) = f(2,:,2) - 2*w(2)*f(1,:,2).*e(2,1)*bc_func(t); % Option B
f(1,:,2) = f(2,:,2) - 2*w(2)*f(1,:,2).*e(2,1)*bc_func(t);
f(1,:,5) = f(2,:,5) - 2*w(5)*f(1,:,5).*e(5,1)*bc_func(t);
f(1,:,6) = f(2,:,6) - 2*w(6)*f(1,:,6).*(e(6,1)*feval(bc_func,t)+ e(6,2)*bc_func(t));
f(1,:,7) = f(2,:,7) - 2*w(7)*f(1,:,7).*(-e(7,1)*feval(bc_func,t)+ e(7,2)*bc_func(t));
f(1,:,8) = f(2,:,8) - 2*w(8)*f(1,:,8).*(-e(8,1)*feval(bc_func,t)- e(8,2)*bc_func(t));
f(1,:,9) = f(2,:,9) - 2*w(9)*f(1,:,9).*(e(9,1)*feval(bc_func,t)- e(9,2)*bc_func(t));
% Boundary condition for right wall
f(Lx,:,4) = f(Lx-1,:,4) - 2*w(4)*f(Lx,:,4).*e(4,1)*bc_func(t);
f(Lx,:,8) = f(Lx-1,:,8) - 2*w(8)*f(Lx,:,8).*(-e(8,1)*feval(bc_func,t)- e(8,2)*bc_func(t));
f(Lx,:,9) = f(Lx-1,:,9) - 2*w(9)*f(Lx,:,9).*(e(9,1)*feval(bc_func,t)- e(9,2)*bc_func(t));
f(Lx,:,6) = f(Lx-1,:,6) - 2*w(6)*f(Lx,:,6).*(e(6,1)*feval(bc_func,t)+ e(6,2)*bc_func(t));
f(Lx,:,3) = f(Lx-1,:,3) - 2*w(3)*f(Lx,:,3).*(e(3,1)*feval(bc_func,t)+ e(3,2)*bc_func(t));
% Boundary condition for bottom wall
f(:,1,3) = f(:,2,3) - 2*w(3)*f(:,1,3).*e(3,2)*bc_func(t);
f(:,1,6) = f(:,2,6) - 2*w(6)*f(:,1,6).*(e(6,1)*feval(bc_func,t)+ e(6,2)*bc_func(t));
f(:,1,7) = f(:,2,7) - 2*w(7)*f(:,1,7).*(-e(7,1)*feval(bc_func,t)+ e(7,2)*bc_func(t));
f(:,1,2) = f(:,2,2) - 2*w(2)*f(:,1,2).*e(2,2)*bc_func(t);
f(:,1,9) = f(:,2,9) - 2*w(9)*f(:,1,9).*(e(9,1)*feval(bc_func,t)- e(9,2)*bc_func(t));
% Boundary condition for top wall
f(:,Ly,5) = f(:,Ly-1,5) - 2*w(5)*f(:,Ly,5).*e(5,2)*bc_func(t);
f(:,Ly,8) = f(:,Ly-1,8) - 2*w(8)*f(:,Ly,8).*(-e(8,1)*feval(bc_func,t)- e(8,2)*bc_func(t));
f(:,Ly,6) = f(:,Ly-1,6) - 2*w(6)*f(:,Ly,6).*(e(6,1)*feval(bc_func,t)+ e(6,2)*bc_func(t));
f(:,Ly,7) = f(:,Ly-1,7) - 2*w(7)*f(:,Ly,7).*(-e(7,1)*feval(bc_func,t)+ e(7,2)*bc_func(t));
f(:,Ly,1) = f(:,Ly-1,1) - 2*w(1)*f(:,Ly,1).*e(1,2)*bc_func(t);
f(:,Ly,9) = f(:,Ly-1,9) - 2*w(9)*f(:,Ly,9).*(e(9,1)*feval(bc_func,t)- e(9,2)*bc_func(t));
% Plot fluid and floaters
if mod(t,10) == 0
figure(1)
clf
subplot(2,2,1)
imagesc(ux')
colorbar
title('Fluid x-velocity')
subplot(2,2,2)
imagesc(uy')
colorbar
title('Fluid y-velocity')
subplot(2,2,3)
scatter(x,y,10,'filled')
axis equal
xlim([0 Lx])
ylim([0 Ly])
title('Floaters')
subplot(2,2,4)
quiver(ux',uy')
axis equal
title('Fluid velocity field')
drawnow
end
end
  1 个评论
Walter Roberson
Walter Roberson 2023-4-2
paralell workers and background workers can never (directly) display graphics. They are not connected to the user's display. They can construct graphs, but unless the graphs are captured or the graphics objects are saved to file or copied to the managing thread, the graphics cannot be visible to the user.
... And if they could be displayed, you would have trouble with the fact that multiple workers might be trying to write to the same figures.

请先登录，再进行评论。

更多回答（1 个）

Walter Roberson
Walter Roberson 2023-4-2
initializing feq before the parfor loop implies that you expect feq to be an output, that you expect to be able to access feq after the parfor. However you do not index feq by the loop variable t
The implication is that you want feq to be shared between all of the workers, so that if one worker happens to write to feq(:, :, 3) and a different worker with a different t happens to write to there as well before the first worker happens to read back from feq(:, :, 3) that you want the revised values to be read back. The implication is that you want the results to depend upon continual races between the workers to be the most recent worker to write to the array.
parfor does not support that kind of deliberate race between workers.
If you did not intend that, if you intend that feq should be local to each worker, then initialize feq on the worker rather than before the parfor.

类别

在 Help Center 和 File Exchange 中查找有关 Graphics Objects 的更多信息

标签

产品


版本

R2022b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by