k-Wave User Forum » Topic: Reflection From Boundary

k-Wave User Forum » Topic: Reflection From Boundary http://www.k-wave.org/forum/topic/reflection-from-boundary Support for the k-Wave MATLAB toolbox en-US Wed, 13 May 2026 14:55:36 +0000 http://bbpress.org/?v=1.0.2 <![CDATA[Search]]> q http://www.k-wave.org/forum/search.php sedy_shes on "Reflection From Boundary" http://www.k-wave.org/forum/topic/reflection-from-boundary#post-6791 Mon, 18 Mar 2019 20:41:54 +0000 sedy_shes 6791@http://www.k-wave.org/forum/ Thank you Brad. That makes a lot of sense for me now. Best, Sedy. Bradley Treeby on "Reflection From Boundary" http://www.k-wave.org/forum/topic/reflection-from-boundary#post-6780 Sun, 17 Mar 2019 22:30:49 +0000 Bradley Treeby 6780@http://www.k-wave.org/forum/ Hi Sedy, The performance of the PML depends on the frequency of incoming waves. So if you reduce the frequency of the source (or decrease the grid spacing), things should improve. I noticed that your source is essentially a step with a value of -2 lasting for 150 microseconds: <code>figure; plot(1e6 * kgrid.t_array(1:length(source.ux)), source.ux);</code> Is that what you intended? The sharp jumps at the beginning and end of the time trace will generate high frequencies and these won't be properly absorbed by the PML as mentioned above. If you use a smoother signal, things will improve. A note to be careful when using the automatic plot scale, as when the waves have left from the domain, you will always see reflections and wave wrapping from the PML, even if the values are very small. Finally, if all else fails and you want to improve PML performance, you can try increasing the thickness. Hope that helps, Brad. sedy_shes on "Reflection From Boundary" http://www.k-wave.org/forum/topic/reflection-from-boundary#post-6755 Mon, 11 Feb 2019 22:57:51 +0000 sedy_shes 6755@http://www.k-wave.org/forum/ Hi Brad, Thanks for the useful toolbox. I guess it is a really basic question, but I've always had the reflection problems since I started to learn K-wave, although I have read some examples provided in the website and tried to follow what is usually done for applying PML. Here is a sample code. We see all the reflection in the normal stress,for example, after running the code. I really appreciate your guidance in advance. Best, Sedy %--------------------------------------- clearvars; PML_size = 20; % [grid points] Nx = 168 - 2*PML_size; % [grid points] Ny = 168 - 2*PML_size; % [grid points] dx = 0.1e-4; % grid point spacing in the x direction [m] dy = 0.1e-4; % grid point spacing in the y direction [m] kgrid = kWaveGrid(Nx, dx, Ny, dy); % define the properties of the propagation medium medium.sound_speed_compression = 10; % [m/s] medium.sound_speed_shear = 3.5; % [m/s] medium.density = 1000; % [kg/m^3] % define the absorption properties medium.alpha_coeff_compression = 0.03; % [dB/(MHz^2 cm)] medium.alpha_coeff_shear = 0.999; % [dB/(MHz^2 cm)] % define a single source point source.u_mask = zeros(Nx, Ny); % source.u_mask(5, Ny/2) = 1; source.u_mask(1:Nx/2, Ny/2-10:Ny/2+11) = 1; % Defining Guassian source sig= 5e-4; xx= kgrid.y_vec(Ny/2-10:Ny/2+11)'; FGaus =zeros(Nx/2,Ny/2+11-(Ny/2-10)+1); FGaus=repmat((-1/2/sig^2)*exp(-(xx./(2*sig)).^2),Nx/2,1); % FGaus(:,Ny/2-10:Ny/2+10)=repmat((-1/2/sig^2)*exp(-(xx./(2*sig)).^2),Nx,1) ux_F = FGaus.*kgrid.dx./medium.sound_speed_compression; ux=reshape(ux_F,[],1); source.ux=ux; sensor.mask = zeros(Nx, Ny); % sensor.mask(Nx/2-10:Nx/2+10, Ny/2) = 1; sensor.mask(Nx/2, Ny/2-40:Ny/2+40) = 1; % sensor.mask(pos2, Ny/2) = 1; sensor.record = {'u','p',}; % set the CFL cfl = 0.1; % define the properties of the PML to allow plane wave propagation pml_alpha = 2; % pml_size = [10, 2]; % set the input arguments % input_args = {'PlotScale', 'auto', 'PMLSize', pml_size,... % 'PMLAlpha', pml_alpha, 'PlotPML', false,'PMLInside',false}; input_args = {'PlotScale', 'auto','PlotPML', false,'Smooth',true... 'PMLInside',false,'PMLAlpha', pml_alpha, 'RecordMovie',true,'DataCast', 'single'}; % set end time t_end = 0.3e-3; % create the time array kgrid.makeTime(max(medium.sound_speed_compression, medium.sound_speed_shear), cfl, t_end); % run the simulation sensor_data = pstdElastic2D(kgrid, medium, source, sensor, input_args{:}); sensor_data_reordered.p = reorderSensorData(kgrid, sensor, sensor_data.p); sensor_data_reordered.ux = reorderSensorData(kgrid, sensor, sensor_data.ux); sensor_data_reordered.uy = reorderSensorData(kgrid, sensor, sensor_data.uy); % figure; imagesc(sensor_data_reordered.p,[-2000,2000]); colormap(getColorMap); ylabel('Sensor Position'); xlabel('Time Step'); colorbar;