[COMSOL_Users] how to specify an initial value for avoiding singular matrix

[Lei Wang]

Dear All

I am trying to introduce an expression into my mocel, which gives a strong dependency in terms of coordinate derivatives of the solution variable shown below:

e_sigma=
18800*exp(-12.3671497584541)*exp(2e-008*sqrt(Vx^2+Vy^2+Vz^2+eps))


Once I tried to run this model , I have always got the same error message: 'singular matrix'. It was known from comsol support knowledge that I should specify a non-zero derivative to solve this problem. But I am not really sure what it means and how I can apply this method to my case to solve my problem. I just wonder if anyone could give me some ideas and I am very looking forward to hearing from you.

best regards
Lei




---------------------------------
For ideas on reducing your carbon footprint visit Yahoo! For Good this month.

[Jens Krause]

Hello Lei,

what is your initial value for V? Maybe you should start with something
that is not constant in space to avoid this. For example you can
generate a solution with constant e_sigma and use this solution as
initial guess for the real solve.

Maybe the 'nojac' operator can help you, but I have no experience with
it.

Regards,

Jens

Post by Lei Wang
Dear All
I am trying to introduce an expression into my mocel, which
gives a strong dependency in terms of coordinate derivatives of the
e_sigma=
18800*exp(-12.3671497584541)*exp(2e-008*sqrt(Vx^2+Vy^2+Vz^2+eps))
Once I tried to run this model , I have always got the same error
message: 'singular matrix'. It was known from comsol support knowledge
that I should specify a non-zero derivative to solve this problem. But
I am not really sure what it means and how I can apply this method to
my case to solve my problem. I just wonder if anyone could give me
some ideas and I am very looking forward to hearing from you.
For ideas on reducing your carbon footprint visit Yahoo! For Good this month.

[Lei Wang]

Dear Jens

The initial value for V is zero in my model. Do you mean that firstly I give an constant value to e_sigma, say 0.1 and solve the model and then restart the model by replacing 0.1 with space dependent expression? Many thanks for your help.

best regards
Lei

Jens Krause <***@fh-kiel.de> wrote:
Hello Lei,

what is your initial value for V? Maybe you should start with something
that is not constant in space to avoid this. For example you can
generate a solution with constant e_sigma and use this solution as
initial guess for the real solve.

Maybe the 'nojac' operator can help you, but I have no experience with
it.

Regards,

Jens

Post by Lei Wang
Dear All
I am trying to introduce an expression into my mocel, which
gives a strong dependency in terms of coordinate derivatives of the
e_sigma=
18800*exp(-12.3671497584541)*exp(2e-008*sqrt(Vx^2+Vy^2+Vz^2+eps))
Once I tried to run this model , I have always got the same error
message: 'singular matrix'. It was known from comsol support knowledge
that I should specify a non-zero derivative to solve this problem. But
I am not really sure what it means and how I can apply this method to
my case to solve my problem. I just wonder if anyone could give me
some ideas and I am very looking forward to hearing from you.
For ideas on reducing your carbon footprint visit Yahoo! For Good this month.

---------------------------------
Yahoo! Answers - Get better answers from someone who knows. Tryit now.

[Steven Conrad]

Lei, sorry to contact you like this but my replies to your email keep
bouncing.

I will be out of the office this week and don't have COMSOL on my
laptop. If you could send the model file to me, I may have time to look
at it early next week.



If your model file is over 2 to 3 MB in size, please use Reset Model
from the file menu to remove solutions and mesh, then zip it up. Let me
know what type of mesh you generated (so I can get the same mesh), and
I'll look at the model if I can find some time.


Steve Conrad

Post by Lei Wang
Dear All
I am trying to introduce an expression into my mocel, which gives a

strong dependency in terms of coordinate derivatives of the solution

Post by Lei Wang
e_sigma=
18800*exp(-12.3671497584541)*exp(2e-008*sqrt(Vx^2+Vy^2+Vz^2+eps))
Once I tried to run this model , I have always got the same error

message: 'singular matrix'. It was known from comsol support knowledge
that I should specify a non-zero derivative to solve this problem. But I
am not really sure what it means and how I can apply this method to my
case to solve my problem. I just wonder if anyone could give me some
ideas and I am very looking forward to hearing from you.

Post by Lei Wang
best regards
Lei
---------------------------------
For ideas on reducing your carbon footprint visit Yahoo! For Good this month.