The final carsus merge #786

yeganer · 2017-08-25T14:35:31Z

This PR replaces #770

yeganer · 2017-08-28T07:43:15Z

docs/physics/plasma/plasma_plots/lte_ionization_balance.py

@@ -1,7 +1,8 @@
 import os
 from matplotlib import pyplot as plt
 from matplotlib import colors
-from tardis import atomic, plasma_array, util
+from tardis import plasma_array, util


I should fix that, right?

yeganer · 2017-08-28T07:43:39Z

docs/physics/plasma/plasma_plots/nebular_ionization_balance.py

@@ -1,7 +1,8 @@
 import os

 from matplotlib import colors
-from tardis import atomic, plasma_array, util
+from tardis import plasma_array, util


Same as above, there is no plasma_array

wkerzendorf · 2017-08-28T08:42:05Z

tardis/plasma/properties/partition_function.py

@@ -162,14 +162,14 @@ def _main_nlte_calculation(
                    (number_of_levels**2, len(t_electrons)))
            r_lu_matrix_reshaped[r_lu_index] = B_lus[np.newaxis].T * \
                j_blues_filtered * beta_sobolevs_filtered
-            if atomic_data.has_collision_data:
+            if atomic_data.collision_data is not None:
                if previous_electron_densities is None:
                    collision_matrix = r_ul_matrix.copy()


collision_matrix = np.zeros_like(r_ul_matrix)

wkerzendorf · 2017-08-28T08:43:15Z

tardis/plasma/tests/test_complete_plasmas.py

+        initial_t_inner + initial_t_rad + helium_treatment)
+
+
+def idfn(fixture_value):


can you quickly document this

yeganer · 2017-08-28T15:11:46Z

My verification show that the macroatom scheme does not work right now. These are three plots with current master and this PR for macroatom, downbranch and scatter.
As you can see, for macroatom there are clear differences while downbranch and scatter are the same.
Running my NLTE tests with macroatom results in an infinite loop which further points towards a problem with the macro atom. I'll have a closer look tomorrow.

yeganer · 2017-08-28T16:20:51Z

Found the bug, nevermind.

yeganer · 2017-08-29T06:44:22Z

After fixing the bug, all tests involving kurucz_cd23_chianti_h_he.h5 produce near identical results, for example here a setup from @talytha :

However my NLTE test setup (with no NLTE species) does have a slightly different spectrum:

@wkerzendorf How should we proceed?

That means we don't need to set indexes in tardis. Additionally all 'copys' were removed as there does not seem to be a special need for them.

As a result, there will be SettingWithCopy warnings but my tests showed they are not problematic.

Additionally: Internally rename 'tardis-refdata' config variable to 'tardis_refdata'.

Also we only close if we opened the HDFStore

This should show up if any property is not found during testing

This is to see the travis error message

yeganer · 2017-08-31T07:26:30Z

@wkerzendorf Tests are passing. Looks like this might be ready to merge.

yeganer mentioned this pull request Aug 25, 2017

Rebase pr692 (needed for new carsus format) #770

Closed

yeganer commented Aug 28, 2017

View reviewed changes

wkerzendorf reviewed Aug 28, 2017

View reviewed changes

yeganer mentioned this pull request Aug 29, 2017

Update tardis/docs/atomic/atomic_data_description.rst #787

Closed

yeganer force-pushed the carsus_merge branch 3 times, most recently from 45b369d to 1483138 Compare August 30, 2017 12:04

mishinma added 18 commits August 31, 2017 08:18

move atomic.py and test_atomic.py to io

d182924

fix imports

4be5f1e

fix data_path in atomic.py

4610db6

fix the imports in conftest

e45f130

fix links in atomic_data_description

53a27b4

fix paths in test_atomic

20d257a

fix import in test_w7

afc2103

implement tests_data_path in atomic

88feb16

fix import in tardis/base.py

493289f

refactor from_hdf and __init__

b461474

ditch all read functions

2458b9e

ditch symbol2atomic_number

0e806b5

write better docstrings

b57f59c

set lines attr in __init__

aa2f002

fix a typo with _v_attrs

533e51a

add attributes and methods to AtomData's docstring

b111f8a

add attributes in the else clauses

7e6021a

change _df to _data

2537725

yeganer added 25 commits August 31, 2017 08:18

The AtomData now comes with indexes set

3abc7f2

That means we don't need to set indexes in tardis. Additionally all 'copys' were removed as there does not seem to be a special need for them.

Fix several issues with indices

6bfbdeb

[PEP8] In file tardis/io/atomic.py

1e0a182

Remove some pd.DataFrame.copy()

9ff3bf0

As a result, there will be SettingWithCopy warnings but my tests showed they are not problematic.

Remove unused parameter 'max_ion_number'

61f4484

Change default AtomData UUID

5caba15

Fix a bug in _main_nlte_calculation

6738a58

Small changes to the plasma tests

2853fc1

fixup! Change default AtomData UUID

e530118

Fix tests for io/atomic.py

35312e8

Add 'tardis_ref_data' and 'generate_reference' fixtures

a49c123

Additionally: Internally rename 'tardis-refdata' config variable to 'tardis_refdata'.

Rewrite tests_plasmas_full to use 'tardis-refdata'

81ccd4e

[IO] Open HDFStore only once during to_hdf

46155ee

Also we only close if we opened the HDFStore

Update tardis-refdata version to pull

31186b2

Rewrite test_tardis_full

9b91d79

Rewrite test_simulation

6dce3df

Rewrite test_plasmas_full

0fb77ef

Add warning to test_complete_plasmas

c0e034d

This should show up if any property is not found during testing

Remove old reference files

acd4385

Fix a bug when calculating the 'block_references'

aa0b054

Remove all occurences of 'plasma_array' in code

8ea67b6

Use np.zeros_like to initialize collision_matrix

73f6ec2

Add docstring to idfn in test_complete_plasmas.py

d6da4b1

Delete old test

b04add2

Always overwrite Reference data when regenerating

1a200c7

yeganer force-pushed the carsus_merge branch from 1483138 to e34218c Compare August 31, 2017 06:39

Download chianti_He.h5 during travis

7904fd6

This is to see the travis error message

yeganer force-pushed the carsus_merge branch from e34218c to 7904fd6 Compare August 31, 2017 06:55

wkerzendorf merged commit 4dfe08b into tardis-sn:master Aug 31, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

The final carsus merge #786

The final carsus merge #786

yeganer commented Aug 25, 2017 •

edited

Loading

yeganer Aug 28, 2017

wkerzendorf Aug 28, 2017

yeganer Aug 28, 2017

wkerzendorf Aug 28, 2017

wkerzendorf Aug 28, 2017

wkerzendorf Aug 28, 2017

yeganer commented Aug 28, 2017

yeganer commented Aug 28, 2017

yeganer commented Aug 29, 2017

yeganer commented Aug 31, 2017

		initial_t_inner + initial_t_rad + helium_treatment)


		def idfn(fixture_value):

The final carsus merge #786

The final carsus merge #786

Conversation

yeganer commented Aug 25, 2017 • edited Loading

yeganer Aug 28, 2017

Choose a reason for hiding this comment

wkerzendorf Aug 28, 2017

Choose a reason for hiding this comment

yeganer Aug 28, 2017

Choose a reason for hiding this comment

wkerzendorf Aug 28, 2017

Choose a reason for hiding this comment

wkerzendorf Aug 28, 2017

Choose a reason for hiding this comment

wkerzendorf Aug 28, 2017

Choose a reason for hiding this comment

yeganer commented Aug 28, 2017

yeganer commented Aug 28, 2017

yeganer commented Aug 29, 2017

yeganer commented Aug 31, 2017

yeganer commented Aug 25, 2017 •

edited

Loading