Discussion Forum

Hi

the message is a bit short ...

probably you have missed something like a material data (or an empty entry) or even a BC

--
Good luck
Ivar

Dear Ivar.

Thanks for replying.

There was no problem before i changed the value in parameter(<Global Definitions).

The error message "Error in multiphysics compilation" occured just after changing the parameter value.

Can you help?

Thanks

I have been struggling with similar problems for a few weeks. The error message says only "error in multiphyiscs compilation" in a pop up window. Even when you use advanced view to drill down into the actual offending subnode (It occurs under the Solver configurations>equation compilation node) it says nothing more than error in multiphysics compilation.

I have had this happen in the middle of parametric sweeps. it will perform 15 or 10 or 1 or 0 out of 50 of the parameter settings and will then just stop. I can reset the parametric sweep to pick up were it left off and sometimes it will work and sometimes it will not run at all.

It is not a memory issue: I have 21 GB of RAM free on the machine and there are not out of memory errors.

It is not a heapscape issue: I've doubled the heapspace (by modifying the steup file as suggested in other threads) and the heapspace error is not occuring.

Has anyone else had this problem?

Hi

that is strange, what I do, when I have struggled for a few days, is to ask "support" for help ;)

--
Good luck
Ivar

Dear all,

First of all, I am having a problem that seems to be very similar to the one "Paris von Lockette" experienced. Hence, I didn't want to open another thread and keep adding to this one...hope thats alright.

I am trying to carry out parametric sweeps on a relatively simple 2D axis-sym. geometry. I have a total of 7 parameter names and would like to sweep 224 parameter values. The computation, however, stops in a rather random manner and I get the following error message:

Error in sweep: The following feature has encountered a problem

The following feature has encountered a problem:
- Feature: Compile Equations: Frequency Domain (sol1/st1)
- Error: Error in multiphysics compilation.

I noticed that this thread was created 3 months ago and was hoping that you guys have found a solution to the problem.

Thank you in advance.

Martin

Eventually I got it figured out.

It is a multi-issue problem.

1. Fix the issues with the heapspace in the setup files. Note there are multiple setup files and all of them should be fixed. Different solver formulations pick up difference setup files.

2. Even so, comsol keeps previous solutions in the heap space . I have not yet determined when, why, or how many but it keeps some of them. So - you will eventually run out of heap space, period!! I was told this by folks from comsol. What will happen is that it may run fine for a full parameter set (has not hit the limit). Next time you get a partial parameter set (yu hit the limit) . Then you get 0 or one set (you are at the limit).

3. In versions before 4.2a you must perodocially delete previous solutions. I do not *think* clear solutions will work consistently. I was told to drill down into the study and delete the solvers themselves! This will allow you to use the full value of your heap space each run. I still had to to my sweeps in multiple stages, but I knew reliably where it would stop (since I had a full heap sapce for each run).

4. in version 4.2a they tell me you can tell the program to (a) save the current solution in the sweep to disk and (b) dump the heap space. This will take more time but will free up heap space. In version 4.2 and earlier you can not do this (the comsol people told me this too).

good luck!!!!!

Hi,I met the same problem,could you explain to me how to fix the issues with the heapspace in the setup files and dump the heap space?

I think you are having the same problems I did. I was using 24GB machines and first got out I memory errors (heapspace problems) then "error in multiphysics compilation" (issues with retained solutions *in the solver*). Here is what I did.


You want To increase the heaps space so that heapspace+max working memory equals about 80-90% of available memory. The working memory is that little number in the bottom right of the comsol window. I have my heapspace set to 4GB and my sims max out at 16-18GB of working mem.

this will get your sims running with as much RAM usage in computing (working mem) and storing (heapspace) as needed. I am not completely sure what the distinction is but I think the heapspace stores things needed from one run to the next.

CRITICAL STEP- see my previous post. You must still delete the solvers periodically because stuff is being stored and building up there in parameter sweeps. If you delete the solvers before each run you will get the same number of parameter combinations each time in your sweeps because you are starting with an empty memory set. You just have to pick up in our sweeps were you left off.

Let me know if this works for you!!

Are you using multiple parametric sweeps? Have you tried to sweep through all 7 parameters in single parametric sweep by defining all possible combinations? For me, going from multiple sweeps to one helps by allowing to solve for more (but not much more!) parameters.

Thanks for your reply.I had solved this problem.As you said,there's something maybe wrong with the heapspace.The program i want to run,is larger than the heaps space i set.And thanks a lot!

Hi,

I am facing an error in defining the material. Even when i have applied material to all figures visible under geometry. Some unseen boundaries are there which needs materials it seems. Can you help me in solving this error. How to remove this extra boundaries, else how to give material to this boundaries.

Regards

Ritu goel