Feat time domain solar wind covariance by jeremy-baier · Pull Request #1966 · nanograv/PINT

jeremy-baier · 2026-02-23T06:52:53Z

Needed for NG20.
Major changes

Making phi either 1d or 2d. Necessary for 2-dimensional kernels.
Addition of said 2d kernels for the solar wind only for now.
Time domain interpolation bases to go with the above. Both custom ( for Bayesian Blocking ) and uniform bins.
Since above, need to switch trivial phi inversion to use a cholesky decomposition with a np.linalg.inverse fallback.

More work should really explore additional caching of various parts of the GLS fit when the noise model is being held constant. This might be sub blocks of the design matrix. Or a refactorization of the GLS implementation.

TODO

unit tests
should check that our basis and phi matrix match those of discovery. (might consider adding tests for this).
additional model validations ?
Try refactoring so there is one TimeDomainSolarWindNoise class with a kernel parameter

Disclosure that I used co-pilot for generating the unit tests as well as some docstrings and type hinting.

…uble; add tests for time domain solar wind; add tests for noise design matrix precision

…are not changing; add tests

scottransom

There's a lot in here and it is beyond me to go through carefully line-by-line! We definitely need to make sure that there are some test cases and/or examples available that provide results that you believe are correct.

jeremy-baier · 2026-04-10T18:46:25Z

thanks for the comments, Scott ! I will take a look and see about implementing them.

…INT into feat/jgb-td-solar-wind-gp

dlakaplan · 2026-06-02T14:04:42Z

Maybe @abhisrkckl can look at this, since it involves changes to the noise model and some of the whitened residual pieces he put in recently?

abhisrkckl · 2026-06-02T15:35:29Z

Is this ready to be looked at?

jeremy-baier · 2026-06-02T19:10:24Z

Is this ready to be looked at?

Yes ! I just need to get the black to pass. I think I have been using the wrong line length settings.

jeremy-baier · 2026-07-14T15:58:48Z

@abhisrkckl some of the tests are now failing but I think that is from the rebases above. this is still ready to be looked at, but in the mean time i will get the tests passing.

scottransom · 2026-07-20T18:52:48Z

For the black issue, what version are you using? And are you overriding any of the default black settings? The requirements_dev.txt is saying black~=24.0, but black is over v26 now.

jeremy-baier · 2026-07-20T21:01:53Z

For the black issue, what version are you using? And are you overriding any of the default black settings? The requirements_dev.txt is saying black~=24.0, but black is over v26 now.

Okay, yea. Somehow I had a bad version of black installed which was part of the issue. Seems to be fixed now as the black is passing.

abhisrkckl · 2026-08-11T01:56:41Z

+def get_tdb_seconds(tbl) -> np.ndarray:
+    """Return TOA TDB times in seconds as float64."""
+    return np.asarray((tbl["tdbld"].quantity * u.day).to(u.s).value, dtype=np.float64)


I think this makes more sense as a method in the TOAs class.

yep. I have moved this to a method there !

abhisrkckl · 2026-08-11T01:57:56Z

+def project_basis_covariance(U: np.ndarray, Phi: np.ndarray) -> np.ndarray:
+    """Project basis-space covariance to data-space covariance."""
+    if np.ndim(Phi) == 1:
+        return np.dot(U * Phi[None, :], U.T)
+    return np.dot(U, np.dot(Phi, U.T))


Please add an assert to make sure that Phi is 1D or 2D.

Done. I have elaborated in that docstring on the cases in which Phi should be 1D or 2D as well as specifically checking that the dimensions match between phi and U appropriately.

abhisrkckl · 2026-08-11T01:59:18Z

+    return np.asarray((tbl["tdbld"].quantity * u.day).to(u.s).value, dtype=np.float64)
+
+
+def _add_tdsw_node_component(model, node, index=None):


Our usual convention for utility functions like this is to put them as methods in the component class. For example, see add_dmx_range method in DispersionDMX. The code will be better organized that way.

This is a good suggestion. The helper function is unnecessary and could be useful as a method.

abhisrkckl · 2026-08-11T02:01:58Z

+
+def _add_tdsw_node_component(model, node, index=None):
+    """Add one TDSWNODE_ prefix parameter to a time-domain SW noise component."""
+    dct = model.get_prefix_mapping_component("TDSWNODE_")


what does dct stand for? maybe rename this variable?

just dictionary. updated !

abhisrkckl · 2026-08-11T02:02:36Z

+    """Add one TDSWNODE_ prefix parameter to a time-domain SW noise component."""
+    dct = model.get_prefix_mapping_component("TDSWNODE_")
+    if index is None:
+        available = [


available --> available_indices

abhisrkckl · 2026-08-11T02:26:32Z

+        "ridge": {"required": ["TDSWLOGSIG"], "optional": []},
+        "sqexp": {"required": ["TDSWLOGSIG", "TDSWLOGELL"], "optional": []},
+        "matern": {
+            "required": ["TDSWLOGSIG", "TDSWLOGELL"],
+            "optional": ["TDSWNU"],
+        },
+        "quasi_periodic": {
+            "required": ["TDSWLOGSIG", "TDSWLOGELL", "TDSWLOGGAMP", "TDSWLOGP"],
+            "optional": [],
+        },


par files usually use all caps by default for parameter names and values.

abhisrkckl · 2026-08-11T02:26:35Z

+        self.add_param(
+            strParameter(
+                name="TDSWKERNEL",
+                value="ridge",


par files usually use all caps by default.

Sorry I don’t understand this comment. Do you want ‘ridge’ to be in all caps ?

ah okay. the above comment seems to clarify my confusion. switching the kernel names and interpolation kinds to all caps.

abhisrkckl · 2026-08-11T02:29:03Z

+        index : int or None
+            Integer index label for the node. If None, increments max index by 1.
+        """
+        return _add_tdsw_node_component(self, node=node, index=index)


This code can be brought inside the class.

abhisrkckl · 2026-08-11T02:30:30Z

+    toas,
+    nodes=None,
+    dt=None,
+    kind="linear",


Add type hints

abhisrkckl · 2026-08-11T02:31:29Z

+    return K
+
+
+def se_kernel(


se_kernel ---> squared_exp_kernel (?)

updated as such. ( I copied se_kernel from enterprise extensions, but I agree that square_exponential_kernel is clearer )

abhisrkckl · 2026-08-11T02:34:07Z

I haven't gone through the entire code yet.

abhisrkckl · 2026-08-11T02:35:03Z

The black issue should be taken care if you pre-commit install.

abhisrkckl · 2026-08-11T02:40:14Z

The new component does not appear in this documentation page:
https://nanograv-pint--1966.org.readthedocs.build/en/1966/timingmodels.html

jeremy-baier · 2026-08-19T18:38:10Z

The new component does not appear in this documentation page: https://nanograv-pint--1966.org.readthedocs.build/en/1966/timingmodels.html

Okay I just added a docs page which describes the solar wind as well as the differences between the Fourier basis and the time domain bases.

jeremy-baier · 2026-08-19T21:25:12Z

@abhisrkckl
Big thanks for your first round of comments. At this time, I think I have addressed them all and it looks like the tests are passing ( 🎉 ). Please let me know what you think. I am mostly happy with how the docs page ended up for the time domain models.
Please let me know if you have anymore concerns on this PR. Otherwise, it needs to be merged in the very near future for ng20 efforts.

codecov · 2026-08-21T20:29:04Z

Codecov Report

❌ Patch coverage is 83.53659% with 54 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.60%. Comparing base (c12e9be) to head (7d55691).
⚠️ Report is 18 commits behind head on master.

Files with missing lines	Patch %	Lines
src/pint/models/noise_model.py	85.02%	16 Missing and 15 partials ⚠️
src/pint/fitter.py	65.38%	5 Missing and 4 partials ⚠️
src/pint/utils.py	70.00%	4 Missing and 2 partials ⚠️
src/pint/residuals.py	75.00%	3 Missing and 1 partial ⚠️
src/pint/models/timing_model.py	96.07%	1 Missing and 1 partial ⚠️
src/pint/simulation.py	66.66%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1966      +/-   ##
==========================================
+ Coverage   70.45%   70.60%   +0.15%     
==========================================
  Files         109      109              
  Lines       25765    26051     +286     
  Branches     4114     4170      +56     
==========================================
+ Hits        18152    18393     +241     
- Misses       6441     6464      +23     
- Partials     1172     1194      +22

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Jeremy Baier added 7 commits February 22, 2026 22:52

add time domain solar wind covariance

d9fa175

make phi 2d

fcc9bbd

add td sw noise models to init.py

a65be32

fix fitter phi inversion to be more stable (hopefully)

37e6386

bug fix noise model

ef6d8db

refactor get_phiinv so that normalization is outside

f3ab176

bugfix circular imports

d53cb23

jeremy-baier mentioned this pull request Feb 23, 2026

Add time domain GP models into the noise module #1906

Open

16 tasks

Jeremy Baier added 2 commits February 23, 2026 14:40

add prefix param for nodes

75a1af8

small commit

18a0a12

jeremy-baier mentioned this pull request Feb 24, 2026

Allow the noise prior weight matrix to be non-diagonal #1945

Open

Jeremy Baier and others added 9 commits February 23, 2026 20:58

rename *SWNoise to TD*SWNoise

2f999e5

add prefix parameter for custom basis nodes; add interpolation kind

4d9e8e5

run black

693587e

add matern kernel

b7494a2

Merge branch 'nanograv:master' into feat/jgb-td-solar-wind-gp

74abc38

change noise design matrix to be double precision rather than long do…

c99a59c

…uble; add tests for time domain solar wind; add tests for noise design matrix precision

move change to before the fourier design matrix is created

9009482

cache noise designmatrix U with checks to make sure noise parameters …

63575a6

…are not changing; add tests

black

b5e7183

jeremy-baier mentioned this pull request Mar 2, 2026

investigate data types used in noise design matrix #1973

Open

jeremy-baier added 2 commits March 4, 2026 11:04

add PINT caching for phi matrix

b8ec690

resolve merge conflict w/ PULL in changelog

19e9f60

scottransom reviewed Apr 9, 2026

View reviewed changes

Comment thread src/pint/models/noise_model.py Outdated

Comment thread src/pint/models/noise_model.py Outdated

Comment thread src/pint/models/noise_model.py Outdated

refactor TDSWNOISE into one class; edit tests

f2d9c1d

jeremy-baier mentioned this pull request May 21, 2026

WIP: cache woodbury intermediate matrix and GLS noise blocks #1994

Open

Jeremy Baier and others added 3 commits May 26, 2026 13:12

fix merge conflicts in changlog, wideband state

ff39084

Merge branch 'nanograv:master' into feat/jgb-td-solar-wind-gp

edb5648

Merge branch 'feat/jgb-td-solar-wind-gp' of github.com:jeremy-baier/P…

52104b7

…INT into feat/jgb-td-solar-wind-gp

jeremy-baier added 2 commits July 8, 2026 15:23

Merge branch 'nanograv:master' into feat/jgb-td-solar-wind-gp

616ca22

Merge branch 'nanograv:master' into feat/jgb-td-solar-wind-gp

954ef89

jeremy-baier mentioned this pull request Jul 20, 2026

Matrix operations with float128 are deprecated #2016

Closed

scottransom mentioned this pull request Jul 20, 2026

Make residuals, scaled uncertainties, and design matrices return float arrays #2020

Merged

jeremy-baier changed the title ~~WIP: feat time domain solar wind covariance~~ Feat time domain solar wind covariance Jul 20, 2026

add docstring; run black

3843888

jeremy-baier and others added 5 commits July 20, 2026 22:54

merge main; resolve conflict

9f02b37

fix merge conflict

e1746ab

run black

fb7f9f5

fix merge conflic; casting to float at wrong point breaks caching

5a6490f

Merge branch 'master' into feat/jgb-td-solar-wind-gp

9449b28

abhisrkckl reviewed Aug 11, 2026

View reviewed changes

jeremy-baier and others added 2 commits August 18, 2026 13:47

Merge branch 'nanograv:master' into feat/jgb-td-solar-wind-gp

341ceaa

add td sw docs; run black

63438fc

jeremy-baier added 2 commits August 19, 2026 13:33

update solar wind docs; address PR comments

9089910

update docs

3657d72

jeremy-baier added 2 commits August 20, 2026 09:28

change register=True so TDSW appears in docs

f77dbf4

update docs

7d55691

		return np.asarray((tbl["tdbld"].quantity * u.day).to(u.s).value, dtype=np.float64)


		def _add_tdsw_node_component(model, node, index=None):

Uh oh!

Conversation

jeremy-baier commented Feb 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Needed for NG20. Major changes

TODO

Uh oh!

scottransom left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jeremy-baier commented Apr 10, 2026

Uh oh!

dlakaplan commented Jun 2, 2026

Uh oh!

abhisrkckl commented Jun 2, 2026

Uh oh!

jeremy-baier commented Jun 2, 2026

Uh oh!

jeremy-baier commented Jul 14, 2026

Uh oh!

scottransom commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jeremy-baier commented Jul 20, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

abhisrkckl commented Aug 11, 2026

Uh oh!

abhisrkckl commented Aug 11, 2026

Uh oh!

abhisrkckl commented Aug 11, 2026

Uh oh!

jeremy-baier commented Aug 19, 2026

Uh oh!

jeremy-baier commented Aug 19, 2026

Uh oh!

jeremy-baier commented Feb 23, 2026 •

edited

Loading

Needed for NG20.
Major changes

scottransom commented Jul 20, 2026 •

edited

Loading