@pyf
Hi,
This is what I found in the tutorial scripts for thermal simulations.

Check the RunTutorial function. It may help. Best.

def CreateModel():
...

def CreateEmFdtdSimulation():
...
return em_sim

def CreateThermalSimulation(em_sim):
...
return thermal_sim

def AnalyzeEmAndThermalSimulation(em_sim,thermal_sim):
# Create extractor for the two simulation output files

em_results = em_sim.Results() thermal_results = thermal_sim.Results() # overall field sensors em_overall_field_sensor = em_results[ 'Overall Field' ] em_overall_field_sensor.Normalization.Normalize = True thermal_overall_field_sensor = thermal_results[ 'Overall Field' ] # Create a slice viewer for the SAR slice_field_viewer_sar = analysis.viewers.SliceFieldViewer() slice_field_viewer_sar.Visualization.Smooth = True slice_field_viewer_sar.Inputs[0].Connect( em_overall_field_sensor['SAR(x,y,z,f0)'] ) slice_field_viewer_sar.Update(0) slice_field_viewer_sar.GotoMaxSlice() document.AllAlgorithms.Add( slice_field_viewer_sar ) #THERMAL slice_field_viewer_t = analysis.viewers.SliceFieldViewer() slice_field_viewer_t.Visualization.Smooth = True slice_field_viewer_t.Inputs[0].Connect( thermal_overall_field_sensor['T(x,y,z,t)'] ) slice_field_viewer_t.Update(0) document.AllAlgorithms.Add( slice_field_viewer_t )

def RunTutorial( path ):

document.New() CreateModel() em_sim = CreateEmFdtdSimulation() document.AllSimulations.Add(em_sim) em_sim.UpdateGrid() em_sim.CreateVoxels(path) thermal_sim = CreateThermalSimulation(em_sim) document.AllSimulations.Add(thermal_sim) thermal_sim.UpdateGrid() thermal_sim.CreateVoxels(path) em_sim.RunSimulation(wait=True) thermal_sim.RunSimulation(wait=True) AnalyzeEmAndThermalSimulation(em_sim,thermal_sim)

if name == 'main':
RunTutorial( project_path )

@mingjuan
Hi,
I recommend updating your driver (https://developer.nvidia.com/opengl-driver).

There's no subgridding just a rectilinear / adaptive grid (you can create new grid settings and drag and drop solids or wireframes to these)

What you can do (at your own 'risk') is manually set the grid to the resolution you want and not use the automatic grid settings. Typically you want at least 10 points per wavelength (more is better) and the wavelength in air (343 / freq) and water (1500 / freq) are quite different.

Don't worry about the time step, just make sure it's fine enough, Sim4Life will then interpolate your function into it's own timestep as needed (so, you just need to be careful that your timestep is not too coarse such that Sim4Life won't be able to linearly (I think) interpolate it

Essentially, the answer is yes, its sufficient

Typically 1 MHz is way too large

wavelength = speed of sound / frequency
resolution = wavelength / 10

so, if you speed of sound in water is 1500 m/s and you want to run a simulation at 100 kHz, then the wavelength is around 15 mm, and the resolution should be at least 1.5mm

From the resolution and your domain size you can calculate the grid size

Grid Size: max. 20 Mio cells

@montanaro That would be of great help.

Thank you so much
Annalisa

I don't remember so well, but I'm pretty sure it's 1/timestep used to discretize / sample the signal. (Note that this is used to create a list of time points and pressure signals, but the acoustic solver then again reinterpolates this signal to it's own time step that is calculated from the Setup Frequency) Setup Frequency is mostly ignored now that you have user defined source, it was mostly used as a frequency of the sources and for the final FFT analysis. However, if I remember correctly it is used to calculate the time step used in the solver, so make sure it is set to the highest frequency available in your sources That expression was just a random signal to give the user an idea of how to input expressions. It seems in your case you want Element 5 to be sin(2 * pi * 0.5E9 *_t) and Element 6 to be sin(2 * pi * 0.75E9 *_t) if you want to excite one transducer at 500 KHz and the other at 750 KHz User defined sources sets an equation which is then sampled / interpolated at that sampling frequency (fs = 1/dt)

Dear Aiping, please send the simulation file to the support team. Thanks.

@pastis Hello, Have you solved this question? can we compare the amount of electric field that reach the targeted region to evaluate the TI simulation ?

Hello,

It seems that you now have the Neuron installer and it is working, but something is missing in your setup.
Could you please check for error messages in the Console?

One possible reason could be that you have installed NEURON before Sim4Life completely finished its installation.
Please uninstall Sim4Life, install it again, and only after finishing install NEURON.

Hello,
Unfortunately there is no direct way from Sim4Life to limit the amount of RAM to allocate for the program. You may be able to do so using external tools that tap into the operating system's settings and manage the resources. But we are not aware of any and we did not test any that we would recommend.

The reason your PC is slowing down is because the OS switches to paging on disk instead of using fast RAM. Paging is slow and uses CPU resources. My guess is that even if you limit the RAM for Sim4Life, the solver will still need a lot more memory and it will either fail or use the disk for paging resulting in the similar slowing effect.

@pastis the anisotropic conductivity is subject specific, and could be evaluated from diffusion weighted images (based on Tuch et al.). Sim4life has an implementation to compute this, but we recently noticed a bug in 6.2 which has been fixed for the 7.0 branch. We will include a python tutorial for this workflow in the 7.0 release.

Alternatively, Sim4Life allows to generate anisotropic conductivity tensor fields given a field that encodes the main directionality (i.e. direction of largest eigenvector). This can be useful for tissues where there is a known directionality, e.g. along some elongated structure.

If the cable bounding box is smaller than the body, maybe you could use sub-gridding.

Hi @Pedro

The FDTD solver does not make use of any CPU parallelization and is actually not very suitable for CPU computation except for very small simulations. It is strongly recommended to run FDTD on GPU to benefit from much higher simulation speeds.

Hi, most likely you have some other boundary conditions that are affecting the solution. Please share a screenshot of your boundary condition settings, maybe there is a clue there.

How can I get the download code of neuro solver?