Lindsey Heagy
d518f1e685
fixed how mu was being included: assuming constant mu = mu_0 for now. need to write a new propmap to test inversion when mu is variable, but not part of the inversion model
2015-07-05 21:59:55 -05:00
Lindsey Heagy
4fa5e9426d
TDEM tests running and passing individually
2015-07-01 00:21:54 -07:00
Lindsey Heagy
1fbf40568d
shortened the test for travis
2015-06-24 16:57:15 -07:00
Lindsey Heagy
d9e8996336
fixed loop source and mag dipole fields for variable mu, added both to testing
2015-06-23 23:02:33 -07:00
Lindsey
ac3e7765fe
Mag dipole sources, defined using prim-sec with a zero-frequency primary actually have a non-zer S_e component if there is variable mu. This has been corrected in the vector potential definition of the sources, and an issue created for the remaining sources. Testing has also been made more robust by using a rawVec source with both S_m and S_e defined which showed a couple bugs in some of the getRHSDeriv_m (which have been corrected), along with a couple minor sizing things in fields derivs
2015-06-23 18:32:00 -07:00
Lindsey Heagy
7239d9cbe6
FDEM problems: e,b,h,j up and running with Jvec and Jtvec within the re-factored framework. Whenever a new element is created, we create its derive wrt u (the computed field) and wrt m (the model). Jvec and Jtvec then stitch these pieces together using chain rule
2015-06-16 11:28:29 -07:00
Lindsey Heagy
9a732ae5b5
Problem_b derivs and adjoint. Notation updates in Jvec and Jtvec.
2015-06-09 20:32:15 -07:00
Lindsey Heagy
d545d9e393
Working Problem_e Jvec and Jtvec
2015-06-02 22:45:48 -07:00
Lindsey Heagy
cd94fea61d
start of Jvec for Problem_e
2015-06-02 19:23:30 -07:00
Lindsey
3df2140a88
Fields now take a srcList. A lot of this could be vectorized later
2015-06-01 14:32:42 -07:00
Lindsey
35d56655f5
Fields Object functioning again. A bit of an ugly fix though... involves a Utils.mkvc on the knownFields for each
2015-05-07 16:00:50 -07:00
Lindsey
569d22bf6a
made loc a kwarg
2015-05-06 16:58:46 -07:00
Lindsey Heagy
51342e7cc8
Fairly major source refactor:
...
- removed the folder Sources and put those routines inside of SrcUtils.
- Broke apart calls for MagDipole, MagDipole_B, CircularLoop
2015-04-29 16:32:50 -07:00
Lindsey
dc7cc1c716
tx -> src
2015-04-17 16:41:54 -07:00
Lindsey
93778d13f1
ProblemFDEM_x.getSource now returns two matrices, S_m and S_e, which cleans up getRHS and the Fields
2015-04-17 10:53:25 -07:00
Lindsey
e980477031
H-J Fields objects created and consistent. Derivatives not implemented yet
2015-04-16 16:52:29 -07:00
Lindsey Heagy
50a853a3b6
CalcFields is obselete
2015-04-16 14:49:29 -07:00
Lindsey
f5a0465f1e
functioning fields object for e,b forward problems. Note that if it passes travis, that is because I am cheating and only testing things that should pass. The way getSources handles types is pretty ugly at the moment... see for example FieldsFDEM.py lines 63-69
2015-04-15 17:55:58 -07:00
Lindsey
dbf3175c39
working on deBugging
2015-04-15 09:50:28 -07:00
Lindsey
fcc0650713
start of FDEM fields refactor, definately will not pass travis at the moment
2015-04-14 19:45:24 -07:00
Lindsey
e61679bdc6
now using j_m and j_g in FDEM problem for all formulations. Note that SimpleTxFDEM has been changed toSimpleTxFDEM_g and SimpleTxFDEM_m
2015-04-14 16:40:09 -07:00
Lindsey Heagy
fa0fd7f23f
broke out calculation of source term from rhs so that you can do prb.getSource
2015-02-28 10:18:58 -08:00
Lindsey
62b4edcf6d
- changed H implementation to primary secondary for a dipole source (much faster)
...
- removed extra MeMui matrix floating around in the B implementation
- changed tests so that we do not cross-check the z-components, as this doesn't make sense for the primary secondary approach if we don't add back the primary
2015-02-27 18:43:47 -08:00
Lindsey
2780a4cea7
HJ formulation implemented. EB formlation will fail the cross-check test... maybe a problem in the source definition?
2015-02-27 13:55:18 -08:00
Lindsey
564fa14826
Added CrossCheck test to make sure that both formulations give the same results. HJ is going to make travis fail...
2015-02-26 16:30:11 -08:00
Lindsey
94d12b257a
Test everything
2015-02-26 15:40:51 -08:00
Lindsey
8d4e001301
HJ formulation solving for h, with h data
2015-02-26 13:51:42 -08:00
Lindsey
5bb5e8f6b3
Start of JH implementation, solving for h. Derivatives for ProblemFDEM_h NOT working yet
2015-02-25 20:21:44 -08:00
Lindsey
57380ea2d8
Tested and passing HJ formulation, solving for J. Only works with diagonal anisotropy
2015-02-25 16:28:54 -08:00
Lindsey
2a6f0ab440
Start of HJ formulation: FAILING TESTS right now
2015-02-24 18:00:34 -08:00
Lindsey
0a98c5f1c3
Fixed Travis?
2015-02-24 12:54:45 -08:00
Lindsey
71881c7ce8
- added capability for adding a mu model
...
- added variable mu in the FDEM testing
- also added capability to add a small random vector to sigma and mu for testing
2015-02-24 11:43:59 -08:00
rowanc1
d58cb21123
Minor Memory improvements and speedups.
...
Speedups are from brackets mat-vec prods vs mat-mat
Memory is from not setting all 'b' fields to zero in Gvec
2014-05-15 08:52:30 -07:00
rowanc1
06de6a346f
Changes to MeshTensor.
2014-04-26 12:20:29 -07:00
rowanc1
eab87c892b
Updates to for mapping changes in SimPEG
2014-04-15 12:07:58 -07:00
rowanc1
b1613dba65
All derivatives/adjoints working for FDEM.
2014-03-30 14:06:54 -07:00
rowanc1
9acaae66db
Projections and derivatives working.
2014-03-22 13:22:44 -07:00
rowanc1
aacbc03cf6
Change rxList --> [rx,rx]
...
Projections are not shared at the moment, but this can be changed later.
2014-03-22 12:24:54 -07:00
rowanc1
a8f09ee869
forward problem changes in data projection. have broken the derivative test.
2014-03-21 17:09:39 -07:00
rowanc1
0a5a7aa5e3
Allows calculation of other fields, and derivatives. e.g. b from when solving e, and visa versa
2014-03-19 23:58:44 -07:00
rowanc1
3197689fb7
bug fix in test.
2014-03-19 17:46:14 -07:00
rowanc1
67aa72f494
Tested J and Jt for b and e formulations. Generalized code so it is easy to reuse. New receiver types.
2014-03-19 17:33:41 -07:00
rowanc1
49eddb4b83
Analytics, b and e formulations, and better solver access
2014-03-18 10:53:20 -07:00
rowanc1
d177960abb
fake the sources for FDEM
2014-03-17 11:26:06 -07:00
rowanc1
95e5a09523
Adjoint test and Jtvec for multiple transmitters
2014-03-17 10:28:43 -07:00
rowanc1
5c4ec4a1cd
Updates to data objects.
2014-03-17 09:55:34 -07:00
rowanc1
e1010f9eca
Initial fields and data implementation.
2014-03-10 22:47:43 -07:00
rowanc1
f10c029161
Updates to fields storage
2014-03-10 22:17:11 -07:00
rowanc1
a14d771515
Change Data to Survey
2014-03-07 14:01:46 -08:00
rowanc1
994bb2159e
Tested Jvec for FDEM
2014-03-05 16:33:30 -08:00
Lindsey Heagy