NIKA2 Cosmology Legacy Survey issueshttps://gitlab.lam.fr/groups/N2CLS/-/issues2024-03-12T08:08:11Zhttps://gitlab.lam.fr/N2CLS/nikamap/-/issues/2Background in photometry2024-03-12T08:08:11ZAlexandre BeelenBackground in photometryShall we use a Background2D estimation for deep fields ?Shall we use a Background2D estimation for deep fields ?https://gitlab.lam.fr/N2CLS/nikamap/-/issues/22PIIC reading2022-04-29T16:07:13ZAlexandre BeelenPIIC readingContBeam breaks PIIC reading.. CheckContBeam breaks PIIC reading.. Checkhttps://gitlab.lam.fr/N2CLS/nikamap/-/issues/21NikaBeam2022-04-29T16:06:39ZAlexandre BeelenNikaBeamtry to recreate a NikaBeam class with an interface similar to the old one to produce a ContBeamtry to recreate a NikaBeam class with an interface similar to the old one to produce a ContBeamhttps://gitlab.lam.fr/N2CLS/nikamap/-/issues/14add_gaussian_sources removes previous fake_sources2019-11-28T10:55:08ZAlexandre Beelenadd_gaussian_sources removes previous fake_sourcesWhen using `add_gaussian_sources` several times, the older `fake_sources` catalog is overwritten, but the map still contains the old fake sourcesWhen using `add_gaussian_sources` several times, the older `fake_sources` catalog is overwritten, but the map still contains the old fake sourceshttps://gitlab.lam.fr/N2CLS/nikamap/-/issues/11Bayesien Catalog2018-05-10T10:23:18ZAlexandre BeelenBayesien CatalogOne could try to apply methods based on https://github.com/eggplantbren/StarStudded
To infer catalog and number counts at onceOne could try to apply methods based on https://github.com/eggplantbren/StarStudded
To infer catalog and number counts at oncehttps://gitlab.lam.fr/N2CLS/nikamap/-/issues/6detect_sources is too slow2018-03-15T08:12:00ZAlexandre Beelendetect_sources is too slow`detect_sources`is way too slow... not suited for Monte Carlo analysis
This is due to the `subpixel=True` options of `photutils.find_peaks` which use `photutils.centroid.fit_2dgaussian`.
One possible way to improve on that is to use `p...`detect_sources`is way too slow... not suited for Monte Carlo analysis
This is due to the `subpixel=True` options of `photutils.find_peaks` which use `photutils.centroid.fit_2dgaussian`.
One possible way to improve on that is to use `photutils.centroid.centroid_1dg` see https://github.com/astropy/photutils/blob/master/docs/centroids.rst
- [x] create a specific branch for testing
- [x] write/find specific test for non centered gaussian
- [x] test actual behavior
- [x] switch to 2 stages find_peaks(subpixel=False) + centroid_1dghttps://gitlab.lam.fr/N2CLS/nikamap/-/issues/4Bias in bootstrap ?2018-02-23T19:39:18ZAlexandre BeelenBias in bootstrap ?std seems to be biased in bootstrap estimation
```python
n_bootstraps = np.logspace(np.log10(10), np.log10(1000), 10).astype(np.int)
mean_std = []
std_std = []
for n_bootstrap in n_bootstraps:
nm = bootstrap(filenames, n_bootstrap=n...std seems to be biased in bootstrap estimation
```python
n_bootstraps = np.logspace(np.log10(10), np.log10(1000), 10).astype(np.int)
mean_std = []
std_std = []
for n_bootstrap in n_bootstraps:
nm = bootstrap(filenames, n_bootstrap=n_bootstrap)
mean_std.append(np.mean(nm.uncertainty.array[~nm.mask]))
std_std.append(np.std(nm.uncertainty.array[~nm.mask]))
# We are actually limited by the number of input maps here...
plt.errorbar(n_bootstraps, mean_std, std_std)
plt.axhline(weighted_noise)
```https://gitlab.lam.fr/N2CLS/nikamap/-/issues/3Release Position for PSFFitting2018-02-06T10:29:49ZAlexandre BeelenRelease Position for PSFFittingShall we use a full fit including position when doing PSF Fitting ? This is necessary to deblend sources.Shall we use a full fit including position when doing PSF Fitting ? This is necessary to deblend sources.