Bård Skaflestad
c81336db05
Make buildable.
...
Pointy hat to: bska.
2010-09-22 08:07:25 +00:00
Bård Skaflestad
96c077f877
Add partial well support (cell contributions). Global assembly remaining.
2010-09-21 15:38:32 +00:00
Bård Skaflestad
883fda040f
In preparation of implementing well support, add a new field ('q')
...
to struct hybsys, and use this field to store 'src-F2*gpress' per
cell. Implement back substitution in terms of hybsys:q, rather than
computing this value in hybsys_compute_press_flux().
2010-09-21 08:28:49 +00:00
Bård Skaflestad
ebee9a2710
Add a bit of summary to functions and data structure.
2010-09-19 17:18:53 +00:00
Bård Skaflestad
8879c1fd7c
Add auto-build support for function 'mex_ifsh'. Also, add auto-build
...
support for function 'mex_compute_coarse_contrib', even if this
function is entirely for testing/development purposese.
2010-09-17 20:54:13 +00:00
Bård Skaflestad
5d0278ba1b
Add MEX gateway routine to fully assembled (mimetic/hybrid)
...
incompressible flow solver. Supports boundary conditions (addBC),
and source terms (addSource). Single phase only.
Initial testing is promising (one-phase linear pressure drop across
cartGrid([250, 250, 10]) solved in about one minute).
2010-09-17 17:51:07 +00:00
Bård Skaflestad
e61ce75f18
Use M's notion of the MAT_SIZE_T when building for M.
2010-09-17 17:43:07 +00:00
Bård Skaflestad
fa214b1264
Declare procedure for deriving cell pressures and interface fluxes
...
(i.e., projected half-contact fluxes).
2010-09-17 17:42:02 +00:00
Bård Skaflestad
9ed3196485
Modify the implementation data:
...
- Store effective Binv rather than Schur complement 'S'.
- Add integer and floating point work arrays (BC handling,
pressure/flux back-substitution &c.)
Always allocate integer work array. Consequently, no need to jump
through hoops to avoid issues surrounding malloc(0).
Add simple procedure for imposing prescribed interface pressure
values (i.e., Dirichlet conditions).
Correct a few clerical errors (use of unitialised vars &c).
Assemble global system, but only for reservoir contributions. Well
contributions remain.
Add trivial procedure for deriving cell pressures and interface
fluxes (projected half-contact fluxes).
2010-09-17 17:41:05 +00:00
Bård Skaflestad
129703add0
White-space adjustment. No functional changes.
2010-09-17 07:35:05 +00:00
Bård Skaflestad
e988e091eb
Add initial support for assembling the Schur complement system
...
directly from cell contributions. Incomplete.
2010-09-16 17:03:11 +00:00
Bård Skaflestad
790d6f4b8b
Track reverse mapping into WI/depth tables in 'cwells'. Update
...
allocation sizes accordingly.
2010-09-16 06:37:23 +00:00
Bård Skaflestad
d96229619e
Separate well connections (typically static) from well controls
...
(often dynamically changing throughout simulation).
2010-09-15 17:27:59 +00:00
Bård Skaflestad
6a1b45fda4
Spell parameter name consistently in header and body.
...
Clue-by-four provided by GCC.
2010-09-15 17:27:14 +00:00
Bård Skaflestad
aa19c72d49
Re-factor creation and destruction of cell->well mapping out to
...
separate module. Update callers.
2010-09-15 12:13:53 +00:00
Bård Skaflestad
dd0aa25511
Summarise purpose of each function.
2010-09-14 19:45:13 +00:00
Bård Skaflestad
3eb2d36c51
Summarise function purposes.
...
While here, guard against invalid DOFs in fill_grid_connections().
2010-09-14 14:12:16 +00:00
Bård Skaflestad
bf8f33be9a
We use memset(), so #include <string.h>.
2010-09-14 13:18:17 +00:00
Bård Skaflestad
237bfbc6a7
Don't state that we reverse-insert (i.e., push-forward) the bin
...
elements when we in fact use jrn's much more straight-forward
push-back mode.
2010-09-14 12:21:30 +00:00
Bård Skaflestad
bcac0e78da
Summarise purpose for remaining functions. Document a few
...
parameters and return values.
While here, convert function pair coarse_topology_build_coarsef()
and coarse_topology_build_final() to CSR push-back build mode.
2010-09-14 12:15:49 +00:00
Bård Skaflestad
54d7094311
Include <assert.h> for assert() support.
2010-09-14 11:48:48 +00:00
Bård Skaflestad
671c9ff74d
Insert "H1 lines" to summarise purpose. Document a few parameters
...
and return values as well.
While here, convert CSR build to push-back mode.
2010-09-14 11:19:08 +00:00
Bård Skaflestad
431910898c
Insert "H1 lines" to summarise purpose.
2010-09-14 11:17:55 +00:00
Bård Skaflestad
a58b81c105
Remove special MATLAB compilation support. This is provided by blas_lapack.h.
2010-09-13 16:11:49 +00:00
Bård Skaflestad
61254bec37
Add traditional implementation of the finite-<something> assembly
...
process.
2010-09-13 16:11:15 +00:00
Bård Skaflestad
ad508cf6e0
Add simple type for describing boundary conditions for the flow
...
equation.
2010-09-13 15:46:06 +00:00
Bård Skaflestad
0a69e72f3c
Maintain number of columns (typically equal to number of rows) when
...
building sparsity structure.
2010-09-11 13:48:53 +00:00
Bård Skaflestad
41a3b5719b
Add utilities to assist in forming global coefficient matrix from
...
Schur complement reduction contributions.
2010-09-10 17:26:46 +00:00
Bård Skaflestad
de2b2551cc
Add initial cut at well definition.
2010-09-10 17:14:00 +00:00
Bård Skaflestad
b04ee86a03
Disable direct assembly routines. A more elaborate (and expensive)
...
scheme is, regrettably, needed for most linear solvers.
2010-09-09 16:49:16 +00:00
Bård Skaflestad
c350f18100
Add simple support routines for building CSR matrices.
2010-09-09 16:48:03 +00:00
Bård Skaflestad
c3c7000815
Complete revision 5188. Sorry about the breakage.
2010-09-09 16:46:28 +00:00
Bård Skaflestad
488b83a642
Factor hash-set support out to separate module. It will be used in
...
deriving the non-zero structure of the global sparse matrix in
models with wells. At least for now. We may wish to use another
approach at some time...
2010-09-09 16:44:58 +00:00
Bård Skaflestad
0423d13a1c
Remove particular MATLAB support. The "blas_lapack.h" include file
...
already provides pertinent definition of MAT_SIZE_T .
2010-09-09 08:37:52 +00:00
Bård Skaflestad
b836029a4a
Ignore debug output from 'coarse_sys.c'.
2010-09-08 16:22:52 +00:00
Bård Skaflestad
5366d18fc2
Correct embarrassing indexing error: Both the fine-scale
...
ip-contributions and the resulting coarse-scale ip matrix have
dimensions (number of basis functions)-by-(number of basis
functions).
2010-09-08 09:09:34 +00:00
Bård Skaflestad
5075c7a222
Only output debug info for IP/B if factoring B failed.
2010-09-08 09:05:13 +00:00
Bård Skaflestad
472d44d846
Add debugging support.
2010-09-08 08:46:47 +00:00
Bård Skaflestad
1dd34c7a4c
Add MEX gateway to constructing coarse-scale inverse inner product
...
matrix. Add test script too. Runs, but does not yet produce
expected results.
2010-09-07 17:34:16 +00:00
Bård Skaflestad
bb4dc52e70
Store explicit degrees of freedom in coarse system. Update calling
...
interfaces to reflect that only the number of coarse blocks is
needed rather than the complete coarse topology where applicable.
2010-09-07 13:46:12 +00:00
Bård Skaflestad
048da5987a
Complete moving geometry support to separate module.
2010-09-07 10:45:29 +00:00
Bård Skaflestad
8257bc51aa
Move geometry support to separate module.
2010-09-07 10:42:38 +00:00
Bård Skaflestad
97808326f5
Fix two (related) comment typos.
2010-09-06 21:04:44 +00:00
Bård Skaflestad
b6a8c75854
Fix typo in comment.
2010-09-06 19:43:25 +00:00
Bård Skaflestad
b8806b475d
Fix use-of-uninitialised bug in derivation of coarse-block
...
inner-product matrix.
Pointy hat: bska
2010-09-06 15:34:27 +00:00
Bård Skaflestad
d316d03f18
Tentative implementation of coarse-sys block contributions.
2010-09-06 15:23:46 +00:00
Bård Skaflestad
2326275d6f
Add tentative type definition for coarse-system management.
2010-09-02 16:26:37 +00:00
Bård Skaflestad
94d4adbb6b
Add local definition of MAT_SIZE_T, contingent upon preprocessor
...
symbol 'MATLAB_MEX_FILE' that is automatically defined by MATLAB's
MEX function.
Add declarations for factorisation, lin-sys solution, and matrix
inversion for (symmetric) positive definite full matrices in full
and packed formats. Will be used in the coarse-system assembly
process.
2010-09-02 16:25:29 +00:00
Jostein R. Natvig
0734f8d842
Add copy of (proposed) grid structure.
2010-08-31 14:09:05 +00:00
Bård Skaflestad
4f1939b81b
Don't use cutesy contractions in error messages. Refer to the same
...
parameter names as in the documentation.
2010-08-31 14:00:43 +00:00