[CIG-SEISMO] anti-damping in viscoelastic simulation, it crashes

[Dimitri Komatitsch]

Dear Yingzi,

PS (important): that is unrelated to SPECFEM, it would be true in any 
other explicit time scheme and related code as well.

Best regards,
Dimitri.

On 10/20/2015 04:35 PM, Dimitri Komatitsch wrote:
>
> Dear Yingzi,
>
> Unfortunately that is known for being unconditionally unstable
> (mathematically), i.e. impossible; when you amplify the signal you also
> amplify the noise and the scheme quickly blows up.
>
> Best regards,
> Dimitri.
>
> On 10/20/2015 03:23 PM, Yingzi Ying wrote:
>> Dear Dimitri and All,
>>
>> I am going to realise a function in viscoelastic simulation with
>> SPECFEM2D. I would like to build an anti-damping material, in which the
>> wave amplitude can increase rather than decrease during propagation,
>> i.e., an inverse process of visco-attenuation. Such processing is
>> physically meaningless but might be useful in signal processing.
>>
>> I added following 3 lines in "compute_forces_viscoelastic.F90"
>>
>> e1_sum  = -e1_sum
>> e11_sum = -e11_sum
>> e13_sum = -e13_sum
>>
>> but the modified code is not stable and crashes.
>>
>> I have tried to reduce the mesh size and time step, but it still
>> crashes(may be not small enough). Should I use double precision instead
>> and decrease the amplification factor of the SOURCE?
>>
>> Do you have any suggestion?
>>
>> Best regards,
>> Yingzi
>> _______________________________________________
>> CIG-SEISMO mailing list
>> CIG-SEISMO at geodynamics.org
>> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
>

-- 
Dimitri Komatitsch
CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics,
UPR 7051, Marseille, France    http://komatitsch.free.fr