Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Alternative fit metrics, fidual timing and more (#71)
* refactoring and adding adam * fitting epw * previous weights * angular * angular * angular * angular * num epochs * plot * sigma * no static * list dict * list dict * deepcopy * deepcopy * deepcopy * deepcopy * angular running without error bars * smooth * removing some code * dist and param handling * input deck change * forward mode working * passing tests * fixed typos in RTD and abstracted forward specral ranges and maxiter * on the fly fe calculation implemented * added series mode * updated docs and removed unused inputs * updated docs and removed unused inputs * modified 2D with sudo code * switching to rotation matrix * coded 2D formfactor for electrons only * debugging 2D edfs * forward pass "working" * vmap test * moved vres to input deck * added custom rotation * added biDLM distributions * merged forward and series modes * refactored plotters and 1D angular forward working * 2D angular forward running, need to be tested more * JAX based chi calc for 2D form factor (#55) * trying scan, vmap runs out of memory can probably scan vmapped functions but may run out of memory for more well resolved distributions * jitted scan is fast enough * tests * fixed bugs in temporal forward * refactored plotters * fixed bugs in regular plotting * added forward mode to gpu queueing * enabled 2D for IAWs * fixed double logging * added edf plots for 2d lineouts * fixed limits and contours on edf plots * fixed 90 degree rotation in edfs * fixed xy ij notation issues * DLM not follows Matte * added missing fields to configs * Squashed commit of the following: commit 521ebef Author: Avi <[email protected]> Date: Thu Jul 25 14:13:30 2024 -0400 more git ignore commit 4f200cb Author: Avi <[email protected]> Date: Thu Jul 25 14:12:50 2024 -0400 git ignore cleanup commit 6615d83 Merge: b5110c5 6a600c2 Author: Avi <[email protected]> Date: Thu Jul 25 14:05:52 2024 -0400 Merge branch 'z_time' of https://github.com/ergodicio/inverse-thomson-scattering into z_time commit b5110c5 Author: Avi <[email protected]> Date: Thu Jul 25 14:03:26 2024 -0400 minor changes for imaging behaviour commit 6a600c2 Author: Avi <[email protected]> Date: Thu Jul 25 10:53:22 2024 -0700 minor fixes for imaging data commit ad98928 Author: Avi <[email protected]> Date: Thu Jul 11 12:19:03 2024 -0400 added absolute timing based off the fiducial commit 76a2c05 Author: Avi <[email protected]> Date: Thu Jul 11 08:01:34 2024 -0700 new timing for iaw and epw commit 484c6cd Author: Avi <[email protected]> Date: Tue Jul 2 11:22:37 2024 -0700 running for shot day commit afedbc7 Author: Avi <[email protected]> Date: Thu Jun 27 08:46:40 2024 -0700 updated electron irf and plotter labeling commit 894a75f Author: Avi <[email protected]> Date: Tue May 28 09:32:06 2024 -0700 inputs updated for 111416 commit 8205e56 Author: Avi <[email protected]> Date: Tue May 28 08:52:06 2024 -0700 updated decks commit 38393d2 Author: Avi <[email protected]> Date: Wed May 22 14:18:10 2024 -0400 moved machine to default commit 6baca3d Merge: 1ff9f76 75894d2 Author: Avi <[email protected]> Date: Wed May 22 12:36:19 2024 -0400 Merge branch 'main' into bug-fix commit 1ff9f76 Author: Avi <[email protected]> Date: Wed May 22 12:31:46 2024 -0400 Fixed init losses commit 78e7987 Merge: 501b163 6916565 Author: Avi <[email protected]> Date: Wed May 22 12:05:11 2024 -0400 Merge remote-tracking branch 'origin/z_time' into bug-fix commit 501b163 Author: Avi <[email protected]> Date: Wed May 22 11:15:07 2024 -0400 Fixed parameter reporting bug and added checker for hessian commit 6916565 Author: Avi <[email protected]> Date: Wed May 22 08:08:34 2024 -0700 Reinstated table based EDFs for 1D commit 75894d2 Author: Archis Joglekar <[email protected]> Date: Thu May 16 08:18:22 2024 -0700 2dedf oom (#57) * refactoring and adding adam * fitting epw * previous weights * angular * angular * angular * angular * num epochs * plot * sigma * no static * list dict * list dict * deepcopy * deepcopy * deepcopy * deepcopy * angular running without error bars * smooth * removing some code * dist and param handling * input deck change * forward mode working * passing tests * fixed typos in RTD and abstracted forward specral ranges and maxiter * on the fly fe calculation implemented * added series mode * updated docs and removed unused inputs * updated docs and removed unused inputs * modified 2D with sudo code * switching to rotation matrix * coded 2D formfactor for electrons only * debugging 2D edfs * forward pass "working" * vmap test * moved vres to input deck * added custom rotation * added biDLM distributions * merged forward and series modes * refactored plotters and 1D angular forward working * 2D angular forward running, need to be tested more * JAX based chi calc for 2D form factor (#55) * trying scan, vmap runs out of memory can probably scan vmapped functions but may run out of memory for more well resolved distributions * jitted scan is fast enough * tests * fixed bugs in temporal forward * refactored plotters * fixed bugs in regular plotting * added forward mode to gpu queueing * enabled 2D for IAWs * fixed double logging * added edf plots for 2d lineouts * fixed limits and contours on edf plots * fixed 90 degree rotation in edfs * fixed xy ij notation issues * DLM not follows Matte * added missing fields to configs * 1D numericaly generated EDFs working in fitting * 1d fully running, 2d OOM * Working on new multispecies * Working until postprocess * Potentialy working * updated calibration for 3-27-24 shot day * Fixed plotting errors * Added fit windowing to normalization * Added fit windowing to normalization * added held Ti * set up for shot 98612 * analyzing ionization * small bug fixes * fixed issues with forward mode * set up for oom error * refit working and additional documentation added * fixed multi-species input bug for 2D edfs * error analysis runs into memory issues * working on tests * working on tests, added some documentation * refitting and hessian errors working * 320x320 works with checkpointing and scan but is slow * new inps * updates * cleanup * more cleanup * dupes --------- Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avi <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> commit 2d81f89 Author: Avi <[email protected]> Date: Mon May 13 08:55:41 2024 -0700 cleanup commit b9c36ed Author: almilder <[email protected]> Date: Tue May 7 12:45:28 2024 -0400 Multi species rework (#56) * refactoring and adding adam * fitting epw * previous weights * angular * angular * angular * angular * num epochs * plot * sigma * no static * list dict * list dict * deepcopy * deepcopy * deepcopy * deepcopy * angular running without error bars * smooth * removing some code * dist and param handling * input deck change * forward mode working * passing tests * fixed typos in RTD and abstracted forward specral ranges and maxiter * on the fly fe calculation implemented * added series mode * updated docs and removed unused inputs * updated docs and removed unused inputs * modified 2D with sudo code * switching to rotation matrix * coded 2D formfactor for electrons only * debugging 2D edfs * forward pass "working" * vmap test * moved vres to input deck * added custom rotation * added biDLM distributions * merged forward and series modes * refactored plotters and 1D angular forward working * 2D angular forward running, need to be tested more * JAX based chi calc for 2D form factor (#55) * trying scan, vmap runs out of memory can probably scan vmapped functions but may run out of memory for more well resolved distributions * jitted scan is fast enough * tests * fixed bugs in temporal forward * refactored plotters * fixed bugs in regular plotting * added forward mode to gpu queueing * enabled 2D for IAWs * fixed double logging * added edf plots for 2d lineouts * fixed limits and contours on edf plots * fixed 90 degree rotation in edfs * fixed xy ij notation issues * DLM not follows Matte * added missing fields to configs * 1D numericaly generated EDFs working in fitting * 1d fully running, 2d OOM * Working on new multispecies * Working until postprocess * Potentialy working * updated calibration for 3-27-24 shot day * Fixed plotting errors * Added fit windowing to normalization * Added fit windowing to normalization * added held Ti * ran some new spectra * set up for shot 98612 * analyzing ionization * small bug fixes * fixed issues with forward mode * best fiting off shot 98612 * set up for oom error * refit working and additional documentation added * fixed multi-species input bug for 2D edfs * error analysis runs into memory issues * working on tests * working on tests, added some documentation * refitting and hessian errors working * test bugfix * test bugfix * added matte option for m * passing all tests * updated iaw dispersion to negative * iaws now flipped to use positive dispersion. Added ylimits to lineout plots * documentation update to reflect multi-species --------- Co-authored-by: archis <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avi <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> commit 2a519bf Merge: f1525e0 154043f Author: Avi <[email protected]> Date: Tue May 7 12:17:19 2024 -0400 documentation update to reflect multi-species commit f1525e0 Author: Avi <[email protected]> Date: Tue May 7 10:57:43 2024 -0400 documentation update to reflect multi-species commit e5a37d7 Author: Avi <[email protected]> Date: Tue May 7 10:05:50 2024 -0400 iaws now flipped to use positive dispersion. Added ylimits to lineout plots commit 739c21d Author: Avi <[email protected]> Date: Mon May 6 13:21:18 2024 -0400 updated iaw dispersion to negative commit d177239 Merge: 43de3cb 08049a5 Author: Avi <[email protected]> Date: Mon May 6 13:10:08 2024 -0400 Merge branch 'multi-species-rework' into shot98612 commit 08049a5 Author: Avi <[email protected]> Date: Mon May 6 12:54:39 2024 -0400 passing all tests commit b62f21d Author: Avi <[email protected]> Date: Mon May 6 12:12:13 2024 -0400 added matte option for m commit 257b65e Author: Avi <[email protected]> Date: Mon May 6 10:56:17 2024 -0400 test bugfix commit 4e746bf Merge: 374de2a 238f733 Author: Avi <[email protected]> Date: Mon May 6 10:02:30 2024 -0400 Merge branch '2dedf_oom' into multi-species-rework commit 374de2a Merge: 8814423 85aec07 Author: Avi <[email protected]> Date: Mon May 6 09:38:38 2024 -0400 Merge branch '2dedf_1Dfit' into multi-species-rework commit 8814423 Author: Avi <[email protected]> Date: Mon May 6 09:19:37 2024 -0400 test bugfix commit f2cdcaa Author: Avi <[email protected]> Date: Fri May 3 14:21:47 2024 -0400 refitting and hessian errors working commit c7181c3 Author: Avi <[email protected]> Date: Thu May 2 11:52:24 2024 -0400 working on tests, added some documentation commit a74236c Author: Avi <[email protected]> Date: Wed May 1 15:57:45 2024 -0400 working on tests commit eca3313 Author: Avi <[email protected]> Date: Fri Apr 26 16:29:20 2024 -0400 error analysis runs into memory issues commit 238f733 Author: Avi <[email protected]> Date: Wed Apr 24 09:37:12 2024 -0400 fixed multi-species input bug for 2D edfs commit 519595c Author: Avi <[email protected]> Date: Tue Apr 23 16:23:57 2024 -0400 refit working and additional documentation added commit c42ce8a Author: Avi <[email protected]> Date: Tue Apr 23 07:17:39 2024 -0700 set up for oom error commit 43de3cb Author: Avi <[email protected]> Date: Tue Apr 23 06:48:05 2024 -0700 best fiting off shot 98612 commit fa42b16 Author: Avi <[email protected]> Date: Mon Apr 22 14:51:19 2024 -0400 fixed issues with forward mode commit 88bfa68 Merge: 1c86245 5b35cf7 Author: Avi <[email protected]> Date: Mon Apr 22 09:55:33 2024 -0400 Merge remote-tracking branch 'origin/multi-species-rework' into multi-species-rework commit 1c86245 Author: Avi <[email protected]> Date: Mon Apr 22 09:52:23 2024 -0400 small bug fixes commit 5b35cf7 Author: Avi <[email protected]> Date: Thu Apr 18 07:19:53 2024 -0700 analyzing ionization commit b0cd124 Author: Avi <[email protected]> Date: Mon Apr 15 07:04:40 2024 -0700 set up for shot 98612 commit 85aec07 Author: Avi <[email protected]> Date: Tue Apr 9 07:09:08 2024 -0700 ran some new spectra commit 2fe872d Author: Avi <[email protected]> Date: Sun Apr 7 11:58:43 2024 -0700 added held Ti commit dd9dd35 Author: Avi <[email protected]> Date: Wed Apr 3 15:15:17 2024 -0400 Added fit windowing to normalization commit a61be37 Author: Avi <[email protected]> Date: Wed Apr 3 14:44:46 2024 -0400 Added fit windowing to normalization commit 4f3648d Author: Avi <[email protected]> Date: Wed Apr 3 14:32:09 2024 -0400 Fixed plotting errors commit c59700c Merge: e0e6295 6772e20 Author: Avi <[email protected]> Date: Wed Apr 3 11:16:28 2024 -0400 Merge branch '2dedf_1Dfit' into multi-species-rework commit 6772e20 Author: Avi <[email protected]> Date: Wed Apr 3 08:13:39 2024 -0700 updated calibration for 3-27-24 shot day commit e0e6295 Author: Avi <[email protected]> Date: Wed Apr 3 11:10:31 2024 -0400 Potentialy working commit 849d48e Author: Avi <[email protected]> Date: Tue Apr 2 16:16:14 2024 -0400 Working until postprocess commit 00743f1 Author: Avi <[email protected]> Date: Tue Apr 2 11:50:48 2024 -0400 Working on new multispecies commit 7bb5860 Author: Avi <[email protected]> Date: Wed Mar 27 06:45:18 2024 -0700 1d fully running, 2d OOM commit f6a7925 Author: Avi <[email protected]> Date: Mon Mar 25 12:47:28 2024 -0400 1D numericaly generated EDFs working in fitting commit 154043f Author: almilder <[email protected]> Date: Wed Jan 31 10:51:20 2024 -0500 switched amplitude to be found within fit range (#53) Co-authored-by: Avi <[email protected]> commit cee3057 Author: Archis Joglekar <[email protected]> Date: Mon Jan 29 09:25:59 2024 -0800 Angular sigmas via ensemble averaging (#50) * ensemble averagimg * working sigmas for 1d * small update * black formatted yaml, and plotting fe sigmas * test for sigmas 1d * Updated input deck for arts to refelct multispecies formating * angular 2d fp running * fixed prefit errors for 2D edf * modified inputs * deleted duplicate code * attempt to address cuda bug * denormalized arts outputs and added statesaving, switched from to adamw * fixed plotting bugs, and finalized saving of states * Fixed buffer background amplitudes to prevent divide by zero * implemented batch_vmap as well as a switching scheme (#59) Co-authored-by: almilder <[email protected]> * added gradient scaling for optax * switched from jaxopt to optax * bugfixes * optax fully working * Fixed 2D Te normalization * fixed edf saving for non angular * implemented hard constraints on ne te fe * Encountering NANs in edf * added learing rate schedule and exit condition * Removed EDF normalization for testing * Created an edf movie * implemented multiplexed ARTS but no postprocessing * bug potentialy in data handleing * prebugfixes * fixed duplicate lineout resizing * Added 2d smoothing and cubic interpolation * Added mlruns to git ignore * fitmetric is now changeable by keyword in default deck * updated gitignore * changed forlder to tsadar * more file moves * fixed merge bugs * fixed param penalty bug --------- Co-authored-by: archis <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avi <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avram Milder <[email protected]> Co-authored-by: Avi <amild@Auros>
- Loading branch information