[Getfem-users] waring messsage : pivot too small. try with ilutp

[Zhenghuai Guo]

Hi Yves, Andriy and Konstantinos,

Thank you for the advice. As you suggested, the mesh had strange elements.

I review my mesh method. The old method gives a bad looking mesh (attached). 
The improved one is also attached. The new good one does not give the error 
message "non invertible matrix" anymore.

However, I now have a warning message:

" Level 2 Warning in ./gmm/gmm_precond_ilut.h, line 172: pivot 303353 too 
small. try with ilutp ?"

Could you please advise what it suggests and please let me know where I can 
find example of using the "ilutp" to help?

Thank you very much
Regards
Zhenghuai Guo



-----Original Message-----
From: Yves Renard <address@hidden> 
Sent: Wednesday, May 1, 2019 10:43 PM
To: Zhenghuai Guo <address@hidden>
Cc: andriy andreykiv <address@hidden>; getfem-users <address@hidden>; 
Konstantinos Poulios <address@hidden>
Subject: Re: How to apply hydraulic pressure on the boundary


Dear Zhenghuai,

It indicates that an element has a non-invertible transformation (a completely 
flat element).

It should be removed before doing any computation.

Best regards,

Yves




----- Original Message -----
From: "Zhenghuai Guo" <address@hidden>
To: "yves renard" <address@hidden>, "andriy andreykiv" <address@hidden>
Cc: "getfem-users" <address@hidden>, "Konstantinos Poulios" <address@hidden>
Sent: Mercredi 1 Mai 2019 11:49:35
Subject: RE: How to apply hydraulic pressure on the boundary

Hi, Andriy and Yves and Konstantinos,



I use gmsh to build a mesh and import in getfem. In the code, I only add the

getfem::add_isotropic_linearized_elasticity_brick(model, mim_1, "u_1", 
"lambda_1", "mu_1");

and

getfem::add_Dirichlet_condition_with_multipliers(model, mim_1, "u_1", mf_u_1, 
total_outer_bounday_1,"Dirichlet_value" );



When I try to solve the equations, I got

[cid:image001.png@01D50055.BCCBF140]



The mesh I loaded has an irregular outer boundary as below. When I change the 
mesh to a simple rectangle outer boundary mesh (also from gsmh), the test 
program runs ok. Can you please advise,  what normally does this error message 
indicates? I suppose something is wrong with the irregular boundary mesh. Thank 
you very much

[cid:image002.png@01D50056.D51FDBC0]

bad.PNG
Description: bad.PNG

good.PNG
Description: good.PNG