Codestin Search App

caioolivv · 2026-04-28T09:08:05Z

Update GalaxySDShape naming scheme and create support for sigma fitting with shapeHSM products

…M calibration products

…SDShapeHSMGauss in configuration files

…_shape_hsm_gauss_global_gen function

…ncated)

…se handling

…irectly

…rectory structure, remove unused sigma conversion test and fix coordinate conversion tests Co-authored-by: Copilot <[email protected]>

… SIGMA directly" This reverts commit 2d4a877.

…(not truncated)" This reverts commit 43d96e3.

… when there's no shear

… model

…ention and remove R factor for cross components

…_stats for performance optimization

…parameters in gen_galaxies function

caioolivv · 2026-05-24T19:32:42Z

@vitenti this is ready for review!

Copilot

Pull request overview

Copilot reviewed 17 out of 17 changed files in this pull request and generated 5 comments.

Comments suppressed due to low confidence (5)

numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss.c:179

The model name/nick still refers to "HSC data", but this class was renamed to NcGalaxySDShapeHSMGauss and the header comment says it uses shapeHSM products (not HSC-specific). Update the user-visible name/nick strings to match the new semantics.
numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss.c:621
In the TRACE_DET branch, gt/sigma_t are scaled by the responsivity R but gx/sigma_x are not. Since R is scalar and the estimator is applied to both components, this asymmetry will bias gx and mis-scale its uncertainty.
numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss.c:628
In the TRACE branch, gt/sigma_t include the expected 0.5 and responsivity scaling, but gx/sigma_x do not. The TRACE convention (distortion χ) applies the same scaling to both components, so gx/sigma_x should be treated consistently.
numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss_global.c:645
In the TRACE_DET branch, gt/sigma_t are scaled by the responsivity R but gx/sigma_x are not. Since R is scalar, both components should be scaled consistently to avoid bias/mis-scaled uncertainties.
numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss_global.c:653
In the TRACE branch, gt/sigma_t include the expected 0.5 and responsivity scaling, but gx/sigma_x do not. The TRACE convention (distortion χ) should apply the same scaling to both components.

+                {"sigma": -0.1},
+                {"std_noise": -0.01},
+            ]
+        case _:  # pragma: no cover


+    def gen_shape(self, mset: Ncm.MSet, shape_data: Nc.GalaxySDShapeData, rng: Ncm.RNG):
        """Generate the galaxy shape source distribution data observations."""
+        std_shape = min(np.sqrt(rng.gamma_gen(self._k_shape, self._theta_shape)), 0.45)
+        std_noise = min(np.sqrt(rng.gamma_gen(self._k_noise, self._theta_noise)), 0.5)
+        c1 = rng.gaussian_gen(0.0, self.c1_sigma)
+        c2 = rng.gaussian_gen(0.0, self.c2_sigma)
+        m = np.exp(rng.gaussian_gen(0.0, self.m_sigma))
        coord = self.ellip_coord.genum


+    def model_post_init(self, _: Any, /) -> None:
        """Check that sigma is less than std_noise."""
-        self._gauss_hsc = Nc.GalaxySDShapeGaussHSC.new(self.ellip_conv.genum)
-        self._k_shape = ((self.std_shape**2) / self.std_sigma) ** 2
-        self._theta_shape = self.std_sigma**2 / self.std_shape**2
+        self._hsm_gauss_global = Nc.GalaxySDShapeHSMGaussGlobal.new(
+            self.ellip_conv.genum
+        )
+        self._hsm_gauss_global["sigma"] = self.sigma
        self._k_noise = ((self.std_noise**2) / self.std_sigma) ** 2
        self._theta_noise = self.std_sigma**2 / self.std_noise**2


+        c1 = rng.gaussian_gen(0.0, 0.01)
+        c2 = rng.gaussian_gen(0.0, 0.01)
+        m = np.exp(rng.gaussian_gen(0.0, 0.08))
+        std_noise = galaxy_shape_std_noise


        std_noise = min(np.sqrt(rng.gamma_gen(self._k_noise, self._theta_noise)), 0.5)
        c1 = rng.gaussian_gen(0.0, self.c1_sigma)
        c2 = rng.gaussian_gen(0.0, self.c2_sigma)
        m = np.exp(rng.gaussian_gen(0.0, self.m_sigma))
        coord = self.ellip_coord.genum
-        self._gauss_hsc.gen(
-            mset, shape_data, std_shape, std_noise, c1, c2, m, coord, rng
-        )
+        self._hsm_gauss_global.gen(mset, shape_data, std_noise, c1, c2, m, coord, rng)


Copilot

Pull request overview

Copilot reviewed 17 out of 17 changed files in this pull request and generated 9 comments.

Comments suppressed due to low confidence (3)

numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss.c:180

The model class name/nick still mentions HSC, but this implementation is now for shapeHSM products (and the file/class names have been updated accordingly). This string is user-visible (logs/serialization), so it should match the new model semantics.
numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss_global.c:862
The warning message refers to ..._new but the failure is occurring inside nc_galaxy_sd_shape_hsm_gauss_global_sigma_from_std_shape(). Using the correct function name makes the warning actionable.
numcosmo/galaxy/nc_galaxy_sd_shape_hsm_gauss_global.c:868
Same warning-message issue as above: the function name in the warning should match nc_galaxy_sd_shape_hsm_gauss_global_sigma_from_std_shape() to help debugging.

+        std_noise = min(np.sqrt(rng.gamma_gen(self._k_noise, self._theta_noise)), 0.5)
+        c1 = rng.gaussian_gen(0.0, self.c1_sigma)
+        c2 = rng.gaussian_gen(0.0, self.c2_sigma)
+        m = np.exp(rng.gaussian_gen(0.0, self.m_sigma))


        std_noise = min(np.sqrt(rng.gamma_gen(self._k_noise, self._theta_noise)), 0.5)
        c1 = rng.gaussian_gen(0.0, self.c1_sigma)
        c2 = rng.gaussian_gen(0.0, self.c2_sigma)
        m = np.exp(rng.gaussian_gen(0.0, self.m_sigma))


+
+        c1 = rng.gaussian_gen(0.0, 0.01)
+        c2 = rng.gaussian_gen(0.0, 0.01)
+        m = np.exp(rng.gaussian_gen(0.0, 0.08))


+    g_assert_true (g_strv_contains ((const gchar * const *) nc_galaxy_wl_obs_peek_columns (obs), NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1));
+    g_assert_true (g_strv_contains ((const gchar * const *) nc_galaxy_wl_obs_peek_columns (obs), NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2));
+    g_assert_true (g_strv_contains ((const gchar * const *) nc_galaxy_wl_obs_peek_columns (obs), NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M));


+        const gdouble c1   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m    = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);
+        const gdouble c1_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m_2  = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);


+        const gdouble c1   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m    = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);
+        const gdouble c1_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m_2  = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);


+        const gdouble c1   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m    = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);
+        const gdouble c1_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m_2  = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);


+      if (NC_IS_GALAXY_SD_SHAPE_HSM_GAUSS (test->galaxy_shape))
+      {
+        const gdouble c1   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2   = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m    = nc_galaxy_wl_obs_get (obs_copy, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);
+        const gdouble c1_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C1, i);
+        const gdouble c2_2 = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_C2, i);
+        const gdouble m_2  = nc_galaxy_wl_obs_get (obs2, NC_GALAXY_SD_SHAPE_HSM_GAUSS_COL_M, i);


+    def model_post_init(self, _: Any, /) -> None:
        """Check that sigma is less than std_noise."""


… gamma distribution

caioolivv and others added 22 commits April 22, 2026 12:08

Rename SDShapeGauss to SDShapeHSMGaussGlobal

5a04403

Adjusting naming in SDShapeHSMGaussGlobal and add support for shapeHS…

7f95f64

…M calibration products

Renaming SDShapeGaussHSC to SDShapeHSMGauss

67e250f

Adjusting naming in SDShapeHSMGauss

833487c

Rename SDShapeGauss and SDShapeGaussHSC to SDShapeHSMGauss Globaland …

eb36be9

…SDShapeHSMGauss in configuration files

Remove unused variables e1, e2, e1_int, and e2_int from _nc_galaxy_sd…

06994f7

…_shape_hsm_gauss_global_gen function

Adjust shape objects naming in test_ncm_generic.c

af6e1d4

Adjust shape objects naming in test_nc_data_cluster_wl.c

eacb879

Fix broken function calls for SDShapeHSMGaussGlobal

64e3282

Fix broken function calss for SDShapeHSMGaussGlobal

3c56715

Refactor to keep objects consistent with full Gaussian model (not tru…

43d96e3

…ncated)

Update documentation for NcGalaxySDShapeHSMGauss to clarify shape noi…

18b3cb9

…se handling

Remove unused std_shape variable and update references to use SIGMA d…

2d4a877

…irectly

Refactor test paths for galaxy SD shape HSM to include shapeHSM in di…

eb9ad5a

…rectory structure, remove unused sigma conversion test and fix coordinate conversion tests Co-authored-by: Copilot <[email protected]>

Fix gen tests

7ff6e67

Revert "Remove unused std_shape variable and update references to use…

7ddd6d7

… SIGMA directly" This reverts commit 2d4a877.

Revert "Refactor to keep objects consistent with full Gaussian model …

e2115f9

…(not truncated)" This reverts commit 43d96e3.

Fix _nc_galaxy_sd_shape_hsm_gauss_global_gen to include additive bias…

8f96620

… when there's no shear

Fix documentation

71e4ee6

Add tests for sigma conversions in galaxy SD shapeHSM Gaussian Global…

15eba27

… model

Test gen truncation

28097b6

Fix integration tests

a3d8bbe

caioolivv added enhancement new feature labels Apr 28, 2026

caioolivv self-assigned this Apr 28, 2026

caioolivv added 5 commits April 28, 2026 17:18

Add R factor in direct estimate of SDShapeHSMGauss for trace-det conv…

29566f0

…ention and remove R factor for cross components

Fix obs_stats size after adding shapeHSM biases

08e9c27

Add correct g estimators with shapeHSM products

422cab2

Use correct mset in stats test and update ngals, nfits and sigma region

7c298af

Fix required_columns test for SDShapeHSMGaussGlobal

ce76432

caioolivv and others added 7 commits May 12, 2026 15:59

Reduce the number of fits in test_nc_galaxy_sd_shape_hsm_gauss_global…

511745e

…_stats for performance optimization

Update galaxy shape distribution to use HSMGaussGlobal and add noise …

b82e2ef

…parameters in gen_galaxies function

Merge branch 'master' into shape_hsm

94dacbe

Merge branch 'master' into shape_hsm

f39ea3e

Merge branch 'master' into shape_hsm

09bb093

Fix gen interface on docs

74639da

Fix LCOV_EXCL_LINE locations

c3c71e2

caioolivv requested a review from vitenti May 21, 2026 17:30

Merge branch 'master' into shape_hsm

9682f77

vitenti mentioned this pull request May 23, 2026

Add support for fixed node integration in NcDataClusterWL #252

Open

Merge branch 'master' into shape_hsm

8a97cdb

vitenti requested a review from Copilot May 25, 2026 20:10

Copilot started reviewing on behalf of vitenti May 25, 2026 20:11 View session

Copilot AI reviewed May 25, 2026

View reviewed changes

vitenti added 3 commits May 25, 2026 18:28

Merge branch 'master' into shape_hsm

2fcc61f

Merge branch 'master' into shape_hsm

340bbcb

Merge branch 'master' into shape_hsm

6523ebc

vitenti requested a review from Copilot May 26, 2026 00:27

Copilot started reviewing on behalf of vitenti May 26, 2026 00:27 View session

Copilot AI reviewed May 26, 2026

View reviewed changes

vitenti and others added 9 commits May 25, 2026 22:21

Merge branch 'master' into shape_hsm

8ae89de

Trigger RTD rebuild

0ecc58d

Trigger RTD rebuild 2

167ab45

Update galaxy shape observation tests to use HSMGaussGlobal column names

713e7b2

Update model_post_init docstring to clarify noise parameter setup for…

1f7cb38

… gamma distribution

Merge branch 'master' into shape_hsm

36eac17

Merge branch 'master' into shape_hsm

1197662

Merge branch 'master' into shape_hsm

cd669fd

Merge branch 'master' into shape_hsm

8f067f8

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update GalaxySDShape naming scheme and create support for sigma fitting with shapeHSM products#242

Update GalaxySDShape naming scheme and create support for sigma fitting with shapeHSM products#242
caioolivv wants to merge 70 commits into
masterfrom
shape_hsm

caioolivv commented Apr 28, 2026

Uh oh!

caioolivv commented May 24, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		def model_post_init(self, _: Any, /) -> None:
		"""Check that sigma is less than std_noise."""

Conversation

caioolivv commented Apr 28, 2026

Uh oh!

caioolivv commented May 24, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants