Hi guys,

Attached is a very simple program in C which uses SNESSetUpdate to switch 
variables while solving the following nonlinear equation:

In region 1 (x <= 5):
    x^3 - 50 = 0

In region 2 (x > 5):
    y = x + 3
    so solve:
    (y - 3)^3 - 50 = 0

The variable switching stops the solution a small distance across the region 
boundary so:

if currently in region 1 and x > 5:
    y -> 8 + epsilon,     region -> 2
if currently in region 2 and x <= 8:
   x -> 5 - epsilon,     region -> 1

This is very simplified but analogous to what we want to do. It seems to work 
ok and has similar convergence to the non-swtiching case (also included).

Obviously this example doesn't change gradients when the variable is switched 
though.

Could you guys add the fortran interfaces for SNESSetUpdate (missing), and 
SNESSetApplicationContext and SNESGetApplicationContext (if they're missing) 
and I will try it out on some more complicated examples.

Thanks very much
John


--
Dr John O'Sullivan
Lecturer
Department of Engineering Science
University of Auckland, New Zealand
email: jp.osullivan at 
auckland.ac.nz<https://lists.mcs.anl.gov/mailman/listinfo/petsc-dev>
tel: +64 (0)9 923 85353

________________________________
From: Peter Lichtner [[email protected]]
Sent: Wednesday, 17 June 2015 8:42 a.m.
To: Barry Smith
Cc: John O'Sullivan; [email protected]; Richard Mills
Subject: Re: [petsc-dev] Missing Fortran interfaces

See comments below …Peter

On Jun 16, 2015, at 13:43, Barry Smith 
<[email protected]<mailto:[email protected]>> wrote:


On Jun 12, 2015, at 10:03 PM, Peter Lichtner 
<[email protected]<mailto:[email protected]>> wrote:

Barry: here is an attempt to explain how variable switching works in pflotran 
(we may need to iterate on this if it's not clear). I will consider as an 
example the 2-phase, 3-component water-supercritical CO2 system.

Any grid cell (we use finite volume) can exist in one of three possible states: 
single-phase pure liquid water (l); single-phase pure supercritical CO2 (g); or 
two-phase coexistence of water and supercritical CO2 (2ph).

Each of these states is described by three independent (nonunique) state 
variables. One possibility is:

l: (p_l, T, X_CO2^l)
g: (p_g, T, X_CO2^g)
2ph: (p_g, T, s_g)

where p_l,g = liquid or gas pressure, T = temperature, and s_g = volume 
fraction of pore space occupied by supercritical CO2.

Where variable switching comes into play is when a grid cell undergoes a phase 
change with the following possibilities:

l -> 2ph
g -> 2ph
2ph -> l
2ph -> g

Note that we do not include directly l <-> g.

As a consequence different regions of space have different primary variables 
the boundaries of which are changing with time along with the state variables.

Line search is not feasible with the current implementation in petsc.

After a completed time step or Newton iteration, the stability of the solution 
in each grid cell must be checked for a change of phase. For example, if the 
cell is in a 2 phase state, but the solution gives s_g > 1 or s_g < 1,

  Do you mean s_q < 0 or can s_g only be 1?

This was a typo—sorry about that: I meant s_g < 0 or s_g > 1 triggers a phase 
change. For a two-phase system the physical values of s_g are: 0 < s_g < 1.

then a phase change occurs from 2ph -> g or 2ph -> l, respectively, and a 
change of variables is necessary:

(p_g, T, s_g) -> (p_l, T, X_CO2^l)

or

(p_g, T, s_g) -> (p_g, T, X_CO2^g)

Agree with John here.

what is the formula for these change of variables?

how/when do g and l go to 2p_h ?

 Thanks

 Barry


…Peter

On Jun 12, 2015, at 18:14, Barry Smith 
<[email protected]<mailto:[email protected]>> wrote:


John,

We can provide any Fortran stubs you need but before we do could you please 
outline to us (in algebraic notation; not in reference to PETSc code) how you 
plan to handle your "variable switching". That is, what is the problem being 
solved, what are the constraints, when do you need to switch variables etc?

I would say everyone who has tried to do variable switching in PETSc before has 
failed because we (the PETSc developers) did not help them enough and  
understand what is needed in the Newton solver. I suspect we need to provide a 
bit more functionality in PETSc to make variable switching a success and I'd 
like to understand what that is and add it to the PETSc algorithms. Also if you 
know of any references that actually explain variable switching in detail, that 
would be great, all I've found are vague hand-wavy discussions that leave too 
many important details unexplained.

Thanks

Barry

On Jun 12, 2015, at 5:54 PM, John O'Sullivan 
<[email protected]<mailto:[email protected]>> wrote:

Hi,

We're working on changing variables during an SNES solve so that we can handle 
phase changes.

I'm looking into using SNESSetUpdate but there doesn't seem to be a fortran 
interface. I get an error:

Undefined symbols for architecture x86_64:
"_snessetupdate_", referenced from:
   _MAIN__ in phaseChange.o

I've looked in the interface files and can't find it either. I tried writing 
one myself but things get a bit too complicated to guess...

Can an interface be added please?

Is this the best to do variable switching? If not where would be the right 
place and are there Fortran interfaces? We will be changing from Pressure, 
Temperature to Pressure, Saturation etc. But for now I'm just working on a 
simple example code.

Cheers
John
--
Dr John O'Sullivan
Lecturer
Department of Engineering Science
University of Auckland, New Zealand
email:
jp.osullivan at auckland.ac.nz<http://auckland.ac.nz>

tel: +64 (0)9 923 85353

Attachment: ex1ajpo.c
Description: ex1ajpo.c

Reply via email to