b29a4ed68c
col area and correct divergence calculation.
2013-12-19 17:09:18 -07:00
53233e25e5
documentation updates.
2013-12-18 11:19:42 -08:00
43b81e1396
Simplifications to Regularization code.
2013-12-18 11:11:32 -08:00
32277eff94
Documentation for Regularization.
2013-12-07 00:01:25 -08:00
783cd40538
Bug fixes in Save, incorporation of data into the inversion object, taken out of problem.
2013-12-06 16:54:38 -08:00
9e26543931
fixes to the meta classes including soft linking inside the database. These are reflected in the loaded outputs which return the same object.
2013-12-06 14:19:37 -08:00
5e0fb8642d
MeatClasses to make every SimPEG object saveable to an hdf5 file.
...
cleaned up imports in a lot of places.
Made solver not copy matrices around for GS preconditioning.
2013-12-06 11:23:01 -08:00
c1767939bb
Merge branch 'develop' of https://github.com/simpeg/simpeg into hdf5
2013-12-05 11:32:24 -08:00
3221b3c750
Updates to video function (skip frames)
2013-12-05 11:31:43 -08:00
1973fde6f6
Merge branch 'develop' of https://github.com/simpeg/simpeg into hdf5
2013-12-04 17:55:31 -08:00
e25afacd1b
refactor and generalize where things are being saved
2013-12-04 17:55:09 -08:00
8fb9f3683b
Easy way to create a padded tensor mesh.
2013-12-04 17:52:40 -08:00
191d0e242c
Create a new inversion node every time inversion is ran.
2013-12-04 16:10:31 -08:00
8ce3c655b7
Merge branch 'develop' of https://github.com/simpeg/simpeg into hdf5
...
Conflicts:
SimPEG/utils/__init__.py
2013-12-04 15:32:54 -08:00
ccb4d4052d
Change callHooks structure.
2013-12-04 15:31:11 -08:00
18f9dc67d5
bug fix.
2013-12-04 14:17:56 -08:00
c9b98cc7c0
Recursive printing to view the inversion.
2013-12-04 14:12:37 -08:00
6603fe77e1
First attempt on saving to a database, have written a very light wrapper on h5py.
2013-11-26 21:35:03 -08:00
9eafec4895
Added options to the solver class that are dynamic based on your environment. If not available, they will fall back to ones that are.
2013-11-26 20:23:40 -08:00
7ef3eb578f
Merged in pyMumpsSolver (pull request #30 )
...
Mumps backend in solver class.
2013-11-22 17:39:27 -08:00
f5e3a4171d
Raise an exception if name is not provided when hooking a lambda function.
2013-11-22 16:11:43 -08:00
1439052172
Multiple RHS using mumps backend.
2013-11-22 15:39:06 -08:00
aae725ff05
TimeSteppingInversion and Estimate Initial Beta based on eigenvalues comparison.
2013-11-22 13:39:26 -08:00
c41b9f12e9
Documentation updates to optimize. Updated the hook method.
2013-11-22 11:58:31 -08:00
6216ae977e
Added hook method to utils so that we can hook new methods dynamically into the classes.
2013-11-22 10:51:01 -08:00
dbaea1fda9
callHooks generalizes some of the hook calling code in Optimize and Inversion
2013-11-21 16:56:01 -08:00
623ff674ea
Added pymumps direct solver backend.
2013-11-21 15:00:51 -08:00
fff2375acd
Moved scipy direct code to own function.
2013-11-21 14:34:32 -08:00
82edbf7a60
Merge branch 'master' of https://bitbucket.org/rcockett/simpeg into richards
...
Conflicts:
SimPEG/inverse/Optimize.py
SimPEG/regularization/Regularization.py
2013-11-21 11:56:12 -08:00
f87899465d
Merged in boundaryConditions (pull request #25 )
...
Cell Gradient and Boundary Conditions
2013-11-21 10:42:43 -08:00
86982e86da
Merged in modelBuilder (pull request #26 )
...
Create a random model
2013-11-21 10:42:00 -08:00
2782a6fcfb
Simple profiling through decorator functions. (counting and timing of functions decorated with @count and @timeIt that have a SimPEG.utils.Counter)
...
e.g. output:
‘’’
Counters:
InexactGaussNewton.doEndIteration : 6
InexactGaussNewton.printIter : 7
InexactGaussNewton.scaleSearchDirection : 6
Times: mean sum
InexactGaussNewton.findSearchDirection : 1.55e-02, 9.29e-02, 6x
InexactGaussNewton.minimize : 1.10e-01, 1.10e-01, 1x
InexactGaussNewton.modifySearchDirection: 2.89e-04, 1.73e-03, 6x
InexactGaussNewton.projection : 3.69e-06, 1.11e-04, 30x
InexactGaussNewton.stoppingCriteria : 1.16e-04, 1.51e-03, 13x
Inversion.dataObj : 6.60e-05, 8.58e-04, 13x
Inversion.dataObj2Deriv : 1.03e-04, 6.20e-03, 60x
Inversion.dataObjDeriv : 5.06e-05, 3.54e-04, 7x
Inversion.evalFunction : 7.75e-04, 1.01e-02, 13x
Inversion.run : 1.10e-01, 1.10e-01, 1x
Regularization.modelObj : 3.56e-04, 4.63e-03, 13x
Regularization.modelObj2Deriv : 1.29e-03, 7.76e-02, 60x
Regularization.modelObjDeriv : 5.01e-04, 3.51e-03, 7x
‘’’
2013-11-20 21:59:53 -08:00
3015a78c32
Create a random model based on a convolution of a kernel and a random field. works in 1D 2D and 3D and has beautiful documentation.
2013-11-20 20:05:07 -08:00
b246b629a5
Merge branch 'BFGS' of https://bitbucket.org/rcockett/simpeg into richards
2013-11-19 17:46:48 -08:00
f71a6f8f06
Merge branch 'master' of https://bitbucket.org/rcockett/simpeg into richards
...
Conflicts:
SimPEG/__init__.py
2013-11-19 17:46:02 -08:00
500844ccb5
Brought BFGS into Optimize. updated InexactGaussNewton to use BFGS as a preconditioner.
2013-11-19 17:41:51 -08:00
4472b39d6e
Iterative Solvers in SimPEG.Solver
2013-11-19 16:00:36 -08:00
b6ec17a4d3
Merge branch 'master' of https://bitbucket.org/rcockett/simpeg into boundaryConditions
...
Conflicts:
SimPEG/__init__.py
2013-11-18 12:33:09 -08:00
a950ba8dd1
Bug fix in emSources. Beginning to change Cyl1DMesh to be more comparable with TensorMesh/BaseMesh.
2013-11-18 10:54:54 -08:00
cdf1dcf79f
Reorganized.
2013-11-14 15:48:32 -08:00
72e844c87d
Added a sub-module to utils to put analytics and source functions.
2013-11-14 11:53:14 -08:00
358411792d
Cleaned up averaging. Put it in an AveExtrap function.
2013-11-13 19:37:14 -08:00
5248ac81ce
Merge branch 'master' of https://bitbucket.org/rcockett/simpeg into seismicExample
...
Conflicts:
docs/api_TestResults.rst
2013-11-12 13:43:06 -08:00
cea1141dcd
Bug fix for documentation.
2013-11-12 12:14:58 -08:00
130592a9e2
Bug fix to solver. and Documentation.
2013-11-12 11:34:49 -08:00
1f13a74bb9
Merge branch 'master' of https://bitbucket.org/rcockett/simpeg into BoundConstraint
...
Conflicts:
.gitignore
SimPEG/utils/__init__.py
2013-11-12 11:24:40 -08:00
d3f38047e4
Multiple RHSs on solvers in Fortran. ~2x speed up on matlab implementation for a single RHS. for multiple RHS there are still some problems.
...
Someone with some knowledge of how fortran works should look at this code.
Added a setup.py script that complies things. f2py should work on most computers, because it is included in the numpy distribution.
2013-11-12 10:36:20 -08:00
ea5dc21517
Fortran Solvers. Faster than Matlab!!
...
Need to implement multiple RHSs.
2013-11-10 17:38:23 -08:00
19f79b3275
updates to Inversion. Bug Fixes.
2013-11-07 17:42:24 -08:00
0dc3b02c45
Combined stopping criteria and printers into a class, for easy reuse.
2013-11-07 16:02:22 -08:00
rowanc1