http://www.k-wave.org/forum/topic/finer-time-and-space-grids-make-for-more-error Support for the k-Wave MATLAB toolbox en-US Tue, 12 May 2026 23:29:14 +0000 http://bbpress.org/?v=1.0.2 q http://www.k-wave.org/forum/search.php http://www.k-wave.org/forum/topic/finer-time-and-space-grids-make-for-more-error#post-5940 Mon, 22 May 2017 21:35:48 +0000 ad000000 5940@http://www.k-wave.org/forum/ <p>Solved! I had two problems:<br /> The main one was that the sample freq for the wavelet was not what I thought.<br /> The second was that I did not define kgrid.t_array. I thought it would do it correctly by default. Maybe not. It worked well when I added:<br /> kgrid.t_array = 0:dt:T_END;</p> <p>Thanks again,<br /> Andy </p> http://www.k-wave.org/forum/topic/finer-time-and-space-grids-make-for-more-error#post-5936 Sun, 21 May 2017 16:07:03 +0000 Bradley Treeby 5936@http://www.k-wave.org/forum/ <p>HI ad000000,</p> <p>I ran a couple of quick simulations using the sine wave signal (I don't have the wavelet toolbox installed). It seems to look ok at first glance, aside from some minor variations that likely come from the length of the source changing slightly with dx to give an integer number of grid points. Do you notice the behaviour only with the wavelet source, or also with the continuous wave source?</p> <p>Brad. </p> http://www.k-wave.org/forum/topic/finer-time-and-space-grids-make-for-more-error#post-5922 Wed, 10 May 2017 21:56:19 +0000 ad000000 5922@http://www.k-wave.org/forum/ <p>Hi.<br /> The various documentation suggest that decreasing dt and dx will help accuracy. </p> <p>I have a 2D water simulation of a strip source (just a 0.3mm line) radiating a pulse of center freq=10MHz. This should produce a field pattern with a beam half-width of about 30 degrees. However as I decrease dt or dx, the beam just keeps getting more and more narrow. I start with grid points per wavelength=7, and dt=0.3*dx/Cwater.<br /> Thanks!</p> <p>code:<br /> freq=10e6;<br /> DiscWid=.31e-3;</p> <p>xlen=5e-3;<br /> ylen=7e-3; %(xdcr axis)<br /> xdcrloc=.5e-3;<br /> epw=15; %points per wavelength</p> <p>Cwater = 1500;<br /> rhoWater=1e3;</p> <p>% setup grid =========================================<br /> lambda=Cwater/freq;<br /> dx=lambda/epw;<br /> dy=dx;<br /> Nx= ceil(xlen/dx);<br /> Ny= ceil(ylen/dx);<br /> kgrid = makeGrid(Nx, dx, Ny, dy);</p> <p>medium.sound_speed=ones(kgrid.Nx, kgrid.Ny)*Cwater;</p> <p>medium.density=ones(kgrid.Nx, kgrid.Ny)*rhoWater;</p> <p>T_END=7e-3/Cwater;<br /> dt=.3*dx/(2*Cwater); %set this velocity for time steps (C should be Cmax)<br /> kgrid.t_array = 0:dt:T_END;</p> <p>% define source ==========================================<br /> DiscX1=round((xlen/2-DiscWid/2)/dx);DiscX2=round((xlen/2+DiscWid/2)/dx);<br /> DiscY=round(xdcrloc/dx);</p> <p>source.p_mask = makeLine(Nx, Ny, [DiscX1,DiscY], [DiscX2,DiscY]);</p> <p>NoPtsWave=round(4/freq/dt); %make a wavelet pulse.<br /> signal = wavelet(freq,4,0,0); %,NoPtsWave);<br /> %signal=sin(2*pi*freq*kgrid.t_array)<br /> source.p=signal;</p> <p>% define receiver ======================================<br /> sensor.mask=[1;1;Nx;Ny];<br /> sensor.record = {'p','p_max'};<br /> display_mask = source.p_mask;</p> <p>% run model<br /> input_args = {'DisplayMask', display_mask, 'PMLInside', false, 'PlotPML', false,'PlotScale',[-1 1]};<br /> sensor_data = kspaceFirstOrder2D(kgrid, medium, source, sensor, input_args{:});</p> <p>% handle data<br /> figure;imagesc(kgrid.y_vec, kgrid.x_vec, sensor_data.p_max);% sensor_data.ux_max.^2+sensor_data.uy_max.^2);<br /> axis equal;grid on</p> <p>figure;imagesc(kgrid.y_vec, kgrid.x_vec, squeeze(sensor_data.p(:,:,end)));<br /> axis equal;grid on </p>