Issue/524/weights in beta #532

eduardojsbarroso · 2022-07-21T15:16:44Z

Include weights in beta measurements #524

eduardojsbarroso · 2022-07-21T15:17:23Z

@marina-ricci Is this what you were thinking about? Let me know if you want something different.

coveralls · 2022-07-21T15:20:58Z

coverage: 100.0%. remained the same
when pulling b044dd7 on issue/524/weights_in_beta
into afa77f0 on main.

m-aguena · 2022-11-25T15:50:02Z

[from 25/11/2022 tag-up]: wait for PR #517

eduardojsbarroso · 2023-03-02T14:08:09Z

@m-aguena I think everything is working now. I added the option of the user to provide shape weights for the source galaxies. If you can take a look if this is what you want please.

m-aguena

Overall is ok. I made comments in specific lines, but they apply to several modifications.

m-aguena · 2023-03-17T17:39:26Z

clmm/utils.py

@@ -782,8 +782,7 @@ def integrand(z_i, z_cl=z_cl, cosmo=cosmo):
    B_mean = quad(integrand, zmin, zmax)[0] / quad(z_distrib_func, zmin, zmax)[0]
    return B_mean

-def compute_beta_s_mean(z_cl, z_inf, cosmo, zmax=10.0, delta_z_cut=0.1, zmin=None,
-                        z_distrib_func=None):
+def compute_beta_s_mean(z_cl, z_inf, cosmo, zmax=10.0, delta_z_cut=0.1, zmin=None, z_distrib_func=None, weights_option=False, z_src = None, shape_weights = None):


I think you can remove the weights_option argument and just apply weight when shape_weights is provided

m-aguena · 2023-03-17T17:41:24Z

clmm/utils.py

-        zmin = z_cl + delta_z_cut
-    Bs_mean = quad(integrand, zmin, zmax)[0] / quad(z_distrib_func, zmin, zmax)[0]
+    if weights_option == False:
+        if z_distrib_func == None:


use is None instead of == None

m-aguena · 2023-03-17T17:42:39Z

clmm/utils.py

+        else:
+            n_galaxies = len(z_src)
+            weight = 1/n_galaxies
+            Bsw_i = [shape_weights[i] * compute_beta_s(z_src[i], z_cl, z_inf, cosmo) for i in range(0,len(z_src))]


I believe a simple vector product can be used here shape_weights * compute_beta_s(z_src, z_cl, z_inf, cosmo)

Done

To change that, I had to change compute_beta to accept a list of sources instead of being called to compute each source at a time. I changed the documentation for z_source: float, array_like, but if the user passes a scalar just for one source, I'm transforming into a one argument list.

eduardojsbarroso · 2023-03-20T10:31:27Z

clmm/utils.py

-    beta = np.heaviside(z_src-z_cl, 0) * cosmo.eval_da_z1z2(z_cl, z_src) / cosmo.eval_da(z_src)
-    return beta
+    if type(z_src) is not list:
+        z_src = [z_src]


I don't know if I should keep this test or expect the user to pass a list with just one source. @m-aguena

m-aguena · 2023-03-31T15:49:27Z

@combet @marina-ricci @eduardojsbarroso I was checking this use, and to sum up we have two possible ways of computing <beta_s>, <beta_s^2>:

From N(z):
From weights:

My suggestion here is to just have different functions (ex: beta_from_distribution, beta_from_weights) as the inputs for each one are completely different.

Then we have two options for how this will be used in the theory functions (ex: eval_tangential_shear):

The user has to pre-compute <beta_s>, <beta_s^2>, and pass it to theory using z_src=(<beta_s>, <beta_s^2>), z_src_model='beta'
We add the possibility for this computation to be done inside the modeling functions. In this case we have to add an option like z_src_info='beta_from_weights' with z_src=(src_redshifts, src_weights)

Opinions?

eduardojsbarroso · 2023-04-03T09:24:00Z

@combet @marina-ricci @eduardojsbarroso I was checking this use, and to sum up we have two possible ways of computing <beta_s>, <beta_s^2>:

From N(z):

From weights:

My suggestion here is to just have different functions (ex: beta_from_distribution, beta_from_weights) as the inputs for each one are completely different.

Then we have two options for how this will be used in the theory functions (ex: eval_tangential_shear):

The user has to pre-compute <beta_s>, <beta_s^2>, and pass it to theory using z_src=(<beta_s>, <beta_s^2>), z_src_model='beta'

We add the possibility for this computation to be done inside the modeling functions. In this case we have to add an option like z_src_info='beta_from_weights' with z_src=(src_redshifts, src_weights)

Opinions?

I think the second option is better than to having the users to compute themselves.

marina-ricci · 2023-04-14T13:54:22Z

@combet @marina-ricci @eduardojsbarroso I was checking this use, and to sum up we have two possible ways of computing <beta_s>, <beta_s^2>:

From N(z):

From weights:

My suggestion here is to just have different functions (ex: beta_from_distribution, beta_from_weights) as the inputs for each one are completely different.
Then we have two options for how this will be used in the theory functions (ex: eval_tangential_shear):

The user has to pre-compute <beta_s>, <beta_s^2>, and pass it to theory using z_src=(<beta_s>, <beta_s^2>), z_src_model='beta'

We add the possibility for this computation to be done inside the modeling functions. In this case we have to add an option like z_src_info='beta_from_weights' with z_src=(src_redshifts, src_weights)

Opinions?

I think the second option is better than to having the users to compute themselves.

I actually think we can live with option 1 (for which we do not have to make modifications) for the time being. I think it would be cleaner that way.

eduardojsbarroso · 2023-05-05T15:29:27Z

@m-aguena tell me if this was what you were thinking when you have time, please.

eduardojsbarroso · 2023-05-09T07:58:14Z

@m-aguena Do you know why is the test failling? The tets work in my computer.

m-aguena · 2023-05-09T09:36:17Z

@eduardojsbarroso yes, it has to do with CCL's new version. PR #593 should fix it

m-aguena · 2023-06-26T13:37:55Z

@m-aguena There are a lot of functions in parent_class that don't have the option distribution anymore, while in the notebook they were computed previously. This option should be added to all the functions or should I remove from the notebook?

@eduardojsbarroso They should be removed from the notebooks. Now, beta should be computed by the user and passed to these function, and this should be reflected on the notebooks

…e_beta_s_mean*_from_weights

…f_z_types.ipynb

m-aguena · 2023-07-13T08:37:20Z

@combet @marina-ricci can one of you guys take a final look before I merge this? I would like a second pair of eyes on it since there are some API changes.

The main implementation was the suggestion here #532 (comment)

But I also ended up adding the functions _eval_da, _eval_da_z1z2, _get_z_from_a, _get_a_from_z to the cosmology object for completeness and using them internally.

Thanks!

…Catalog.ipynb

* distinguish between compute_beta_s_*_from_distribution and compute_beta_s_*_from_weights * make compute_beta more efficient and fix shape_weights=None in compute_beta_s_mean*_from_weights * add _eval_da, _eval_da_z1z2, _get_z_from_a, _get_a_from_z to cosm * reorder funcs in cosmo parent * add fix to skip pylint for astropy.units (fails in astropy 6) * Update version to 1.12.0 --------- Co-authored-by: m-aguena <[email protected]>

eduardojsbarroso added 2 commits July 21, 2022 07:28

Merged into main

6589f93

Added weights option

b36ed4b

eduardojsbarroso self-assigned this Jul 21, 2022

eduardojsbarroso linked an issue Jul 21, 2022 that may be closed by this pull request

Include weights in beta measurements #524

Closed

m-aguena added HasDependency Depends on another issue to be done and removed HasDependency Depends on another issue to be done labels Jan 23, 2023

eduardojsbarroso added 4 commits March 1, 2023 15:28

Merge remote-tracking branch 'origin' into issue/524/weights_in_beta

6af1a3e

Fixed old code to match new features

2fd4ad8

Deleting Unwanted commited files

23a0c58

Implemented Weights option

8605658

eduardojsbarroso marked this pull request as ready for review March 2, 2023 14:08

m-aguena requested changes Mar 17, 2023

View reviewed changes

Implemented resquested changes

5766cfa

eduardojsbarroso commented Mar 20, 2023

View reviewed changes

eduardojsbarroso and others added 2 commits May 5, 2023 17:25

Finished implementation of functions

37a3e76

Merge branch 'main' into issue/524/weights_in_beta

724faec

Fixed failling tests

b990d0e

m-aguena added 3 commits May 9, 2023 14:15

rm compute_beta from theory.parent

6335ed9

revert beta functions

744a8ae

improve compute_beta

88765f3

Finished updating notebook

34dd89f

Added description in notebook

01a0e90

eduardojsbarroso requested a review from m-aguena June 27, 2023 13:53

m-aguena added 10 commits July 12, 2023 16:50

make compute_beta more efficient and fix shape_weights=None in comput…

835fdcc

…e_beta_s_mean*_from_weights

improve compute_beta_s*_mean_from_weights

ef11681

add _eval_da, _eval_da_z1z2, _get_z_from_a, _get_a_from_z to cosm

d04f6e3

reorder funcs in cosmo parent

00d6e02

update tests

4fc845e

add compute_beta_s_mean_from_weights to demo_theory_functionality_dif…

a500d61

…f_z_types.ipynb

reorder funcs in cosmo parent

6faa95e

add fix for pylint duplicate-code error bug

2416457

add pylint skip

82ca497

fix linebreaks

2dd7b83

m-aguena approved these changes Jul 13, 2023

View reviewed changes

m-aguena requested review from combet and marina-ricci July 13, 2023 08:38

m-aguena added 4 commits October 18, 2023 10:04

fix merge conflicts

5e14ff2

fix new use in examples/mass_fitting/Example3_Fit_Halo_Mass_to_Shear_…

1cc32c2

…Catalog.ipynb

fix new use in examples/demo_theory_functionality.ipynb

1b29e0b

fix new use in examples/demo_theory_functionality_oo.ipynb

5e8af97

m-aguena force-pushed the issue/524/weights_in_beta branch from f13e36e to 5e8af97 Compare February 20, 2024 16:01

m-aguena added 4 commits February 21, 2024 03:25

Merge branch 'main' into issue/524/weights_in_beta

0b41b60

Merge branch 'main' into issue/524/weights_in_beta

9169407

Update version to 1.12.0

024139c

add fix to skip pylint for astropy.units (fails in astropy 6)

b044dd7

m-aguena merged commit 50876a8 into main Mar 28, 2024
2 checks passed

m-aguena mentioned this pull request Mar 28, 2024

Evaluate if there is possible solution for pylint on astropy #621

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Issue/524/weights in beta #532

Issue/524/weights in beta #532

eduardojsbarroso commented Jul 21, 2022

eduardojsbarroso commented Jul 21, 2022

coveralls commented Jul 21, 2022 •

edited

Loading

m-aguena commented Nov 25, 2022

eduardojsbarroso commented Mar 2, 2023

m-aguena left a comment

m-aguena Mar 17, 2023

eduardojsbarroso Mar 20, 2023

m-aguena Mar 17, 2023

eduardojsbarroso Mar 20, 2023

m-aguena Mar 17, 2023

eduardojsbarroso Mar 20, 2023

eduardojsbarroso Mar 20, 2023

m-aguena commented Mar 31, 2023

eduardojsbarroso commented Apr 3, 2023

marina-ricci commented Apr 14, 2023

eduardojsbarroso commented May 5, 2023

eduardojsbarroso commented May 9, 2023

m-aguena commented May 9, 2023

m-aguena commented Jun 26, 2023

m-aguena commented Jul 13, 2023 •

edited

Loading

Issue/524/weights in beta #532

Issue/524/weights in beta #532

Conversation

eduardojsbarroso commented Jul 21, 2022

eduardojsbarroso commented Jul 21, 2022

coveralls commented Jul 21, 2022 • edited Loading

m-aguena commented Nov 25, 2022

eduardojsbarroso commented Mar 2, 2023

m-aguena left a comment

Choose a reason for hiding this comment

m-aguena Mar 17, 2023

Choose a reason for hiding this comment

eduardojsbarroso Mar 20, 2023

Choose a reason for hiding this comment

m-aguena Mar 17, 2023

Choose a reason for hiding this comment

eduardojsbarroso Mar 20, 2023

Choose a reason for hiding this comment

m-aguena Mar 17, 2023

Choose a reason for hiding this comment

eduardojsbarroso Mar 20, 2023

Choose a reason for hiding this comment

eduardojsbarroso Mar 20, 2023

Choose a reason for hiding this comment

m-aguena commented Mar 31, 2023

eduardojsbarroso commented Apr 3, 2023

marina-ricci commented Apr 14, 2023

eduardojsbarroso commented May 5, 2023

eduardojsbarroso commented May 9, 2023

m-aguena commented May 9, 2023

m-aguena commented Jun 26, 2023

m-aguena commented Jul 13, 2023 • edited Loading

coveralls commented Jul 21, 2022 •

edited

Loading

m-aguena commented Jul 13, 2023 •

edited

Loading