Dear Konsta
I wrote the following lines
print('mfpf dofs',mfpb.nbdof(),'mfoutb dofs',mfoutb.nbdof())
mass_mat = gf.asm_mass_matrix(mimp3b, mfpb)
for step in range(steps):
nit, conv = md.solve('noisy', 'max_iter', 25, 'max_res', 1e-10, 'lsolver', 'mumps',
'lsearch', 'simplest', 'alpha max ratio', 100, 'alpha min', 0.2, 'alpha mult', 0.6,
'alpha threshold res', 1e9)
time_elapsed(timer)
it = step+1
SNRJ = gf.asm_generic(mimp3b, 1, "(Gb*Grad(ub):Grad(ub)+Gb/(1-2*nub)*Div(ub)*Div(ub))*Test_t", -1, md,
"t", True, mfpb, np.zeros(mfpb.nbdof()))
SNRJ = np.transpose(gf.linsolve_mumps(mass_mat, SNRJ))
But I get a size error as follows on the transpose line
<pre>mfpf dofs 54026 mfoutb dofs 54026
Time elapsed: 9.0 minutes and 40.03 seconds
Traceback (most recent call last):
File "/work/testr25pebrainbsftslhomnrj.py", line 271, in <module>
SNRJ = np.transpose(gf.linsolve_mumps(mass_mat, SNRJ))
File "/venv/lib/python3.8/site-packages/getfem/getfem.py", line 6447, in linsolve_mumps
return getfem('linsolve', 'mumps', M, b)
RuntimeError: (Getfem::InterfaceError) -- Argument 3 has wrong dimensions: expected 54026, found 1339902</pre>
Thank you
AC
From: Konstantinos Poulios <logari81@googlemail.com>
Sent: Wednesday, January 26, 2022 2:24 AM
To: Lesage,Anne Cecile J <AJLesage@mdanderson.org>
Cc: getfem-users@nongnu.org
Subject: Re: [EXT] Re: Computing the linear elastic strain energy
|
WARNING:
This email originated from outside of MD Anderson. Please validate the sender's email address before clicking on links or attachments as they may not be safe.
|
Ok, I see. I would not do node-averaging, like most commercial (and not only) software do, for exporting smooth scalar fields (von Mises, strain energy, etc.). Instead, my favorite approach is the one that you can find in
for exporting the plastic strain field GAMMA.
What you need for this method is:
1) Before starting your simulation you calculate and store a mass matrix for the scalar mesh_fem that you want to get the nodal values for
mass_mat = gf.asm_mass_matrix(mim, mfout)
mfout has to be a scalar fem (1 dof per node).
2) After (each step of) your simulation, you assemble the quantity that you want to export, in this case strain energy with something like
STRAIN_ENERGY = gf.asm_generic(mim, 1, "(Gb*Grad(ub):Grad(ub)+Gb/(1-2*nub)*Div(ub)*Div(ub))*Test_t", -1, md,
"t", True, mfout, np.zeros(mfout.nbdof()))
(check the correctness of the strain energy _expression_ that I took from some code you sent earlier)
3) Recover nodal values with
STRAIN_ENERGY = np.transpose(gf.linsolve_mumps(mass_mat, STRAIN_ENERGY))
Then the vector STRAIN_ENERGY has the nodal values of the strain energy on all dofs of mfout. If you calculate this vector for several simulations or simulation steps, you can copy the
result into different columns of a big matrix as in the paper you are referring to.
Dear Konstantinos
The problem with patient specific surgery modeling is that we ignore a lot of the surgery scenario
One way to model it with FEM is to precompute an atlas of organ deformation with a variety of surgery scenario parameter (here direction of gravity, loss of brain buoyancy by csf
drainage, shrinkage due to drug,
swelling due to edema, tumor resection, etc …) That is about 1000 simulations
Then the idea is to match intraoperative data, the surgeons measure a ground truth displacement on a few brain surface points
The idea is to minimize function equation 7 in attached pdf:
the first term is the least square between the ground truth displacement and the simulated one see equation
the second term use the elastic strain energy on fem points pondered by the distance to the ground truth measurement points
Finally we want to get the optimal alpha pondering vector to match the surgery measurement
Ideally I would like my python script to output after poroelastic simulation converge
The simulated displacements on the few measurement points (15)
The pondered energy term
So that I can assemble the function to minimize for the whole atlas
Thank you
Regards
Anne-Cecile
From: Konstantinos Poulios <logari81@googlemail.com>
Sent: Tuesday, January 25, 2022 2:40 AM
To: Lesage,Anne Cecile J <AJLesage@mdanderson.org>
Cc: getfem-users@nongnu.org
Subject: [EXT] Re: Computing the linear elastic strain energy
|
WARNING: This email originated from outside of MD Anderson. Please validate the sender's email address before clicking on links or attachments as they may not
be safe.
|
The elastic strain energy density (is a scalar) at each node is easy to calculate. However, I have to ask you what you need this quantity for. Your reference suggests calculating
this quantity on nodes which is probably not the best solution. If you calculate the strain energy density at a node between different elements, you get different results depending on which element you are evaluating the common node from. Then some averaging
is required to get a value somewhere between all neighboring elements. Normally we evaluate strain energy densities at integration points. If you tell us what you need this result for I can maybe help with a better solution.
Dear all
I need to output tat the end of my poroelastic simulation the linear elastic strain energy matrix at each node of my mesh
See attached formula
What is an optimal way to compute it from the displacement u?
Epsilon i = 0.5 (grad u + grad ut) right?
And S here is the stress-strain law for isotropic materials?
Thank you
Anne-Cecile
The information contained in this e-mail message may be privileged, confidential, and/or protected from disclosure. This e-mail message may contain protected health information
(PHI); dissemination of PHI should comply with applicable federal and state laws. If you are not the intended recipient, or an authorized representative of the intended recipient, any further review, disclosure, use, dissemination, distribution, or copying
of this message or any attachment (or the information contained therein) is strictly prohibited. If you think that you have received this e-mail message in error, please notify the sender by return e-mail and delete all references to it and its contents from
your systems.
The information contained in this e-mail message may be privileged, confidential, and/or protected from disclosure. This e-mail message may contain protected health information (PHI); dissemination of PHI should comply with applicable federal
and state laws. If you are not the intended recipient, or an authorized representative of the intended recipient, any further review, disclosure, use, dissemination, distribution, or copying of this message or any attachment (or the information contained therein)
is strictly prohibited. If you think that you have received this e-mail message in error, please notify the sender by return e-mail and delete all references to it and its contents from your systems.
The information contained in this e-mail message may be privileged, confidential, and/or protected from disclosure. This e-mail message may contain protected health information (PHI); dissemination of PHI should comply with applicable federal and state laws. If you are not the intended recipient, or an authorized representative of the intended recipient, any further review, disclosure, use, dissemination, distribution, or copying of this message or any attachment (or the information contained therein) is strictly prohibited. If you think that you have received this e-mail message in error, please notify the sender by return e-mail and delete all references to it and its contents from your systems.
|