LinAlgError: SVD did not converge in Linear Least Squares

[Berto70]

Hi, I'm doing some tests with eleanor module (i'm a novice). I'm using the example code written on eleanor website

`from IPython.display import Image import warnings warnings.filterwarnings('ignore')

import eleanor import numpy as np import matplotlib.pyplot as plt from astropy.coordinates import SkyCoord

eleanor.Update(sector=1)

star=eleanor.Source(tic=38846515, sector=1) print ('-------------------------------------') print ('-------------------------------------') print ('-------------------------------------') print ('') print ('') print ('Found TIC {0} (Gaia {1}), with TESS magnitude {2}, RA {3}, and Dec {4}' .format(star.tic, star.gaia, star.tess_mag, star.coords[0], star.coords[1]))

data = eleanor.TargetData(star, height=15, width=15, bkg_size=30, do_psf=True, do_pca=True, regressors='corner')

plt.figure(figsize=(15,5))

q = data.quality == 0

plt.plot(data.time[q], data.raw_flux[q]/np.nanmedian(data.raw_flux[q])+0.06, 'k') plt.plot(data.time[q], data.corr_flux[q]/np.nanmedian(data.corr_flux[q]) + 0.03, 'r') plt.plot(data.time[q], data.pca_flux[q]/np.nanmedian(data.pca_flux[q]), 'g') plt.plot(data.time[q], data.psf_flux[q]/np.nanmedian(data.psf_flux[q]) - 0.02, 'b') plt.ylabel('Normalized Flux') plt.xlabel('Time [BJD - 2457000]') plt.title('WASP-100') plt.show()` eleanor prova.txt

After running the program, it find the star with its magnitude and coordinates, but appears this error: `LinAlgError Traceback (most recent call last) c:\Users\gabri\OneDrive - Università degli Studi di Padova\Desktop\eleanor prova.py in 23 24 ---> 25 data = eleanor.TargetData(star, height=15, width=15, bkg_size=30, do_psf=True, do_pca=True, regressors='corner') 26 27 plt.figure(figsize=(15,5))

C:\Python36\lib\site-packages\eleanor\targetdata.py in init(self, source, height, width, save_postcard, do_pca, do_psf, bkg_size, aperture_mode, cal_cadences, try_load, regressors, language) 227 self.create_apertures(self.tpf.shape[1], self.tpf.shape[2]) 228 --> 229 self.get_lightcurve() 230 231 if do_pca == True:

C:\Python36\lib\site-packages\eleanor\targetdata.py in get_lightcurve(self, aperture) 623 norm = np.nansum(self.all_apertures[a], axis=1) 624 all_corr_lc_pc_sub[a] = self.corrected_flux(flux=all_raw_lc_pc_sub[a]/np.nanmedian(all_raw_lc_pc_sub[a]), --> 625 bkg=self.flux_bkg[:, None] norm) 626 all_corr_lc_tpf_sub[a]= self.corrected_flux(flux=all_raw_lc_tpf_sub[a]/np.nanmedian(all_raw_lc_tpf_sub[a]), 627 bkg=self.tpf_flux_bkg[:, None] norm)

C:\Python36\lib\site-packages\eleanor\targetdata.py in corrected_flux(self, flux, skip, modes, pca, bkg, regressors) 1235 f = np.arange(0, brk, 1); s = np.arange(brk, len(self.time), 1) 1236 -> 1237 lc_pred = calc_corr(f, cx, cy, skip) 1238 corr_f = flux[f]-lc_pred + med 1239

C:\Python36\lib\site-packages\eleanor\targetdata.py in calc_corr(mask, cx, cy, skip) 1175 # temp_lc = lightcurve.LightCurve(t, flux).flatten() 1176 tmp_flux = np.copy(flux[np.isfinite(flux)], order="C") -> 1177 tmp_flux[:] /= savgol_filter(tmp_flux, 101, 2) 1178 SC = sigma_clip(tmp_flux, sigma_upper=3.5, sigma_lower=3.5) 1179

C:\Python36\lib\site-packages\scipy\signal_savitzky_golay.py in savgol_filter(x, window_length, polyorder, deriv, delta, axis, mode, cval) 333 x = x.astype(np.float64) 334 --> 335 coeffs = savgol_coeffs(window_length, polyorder, deriv=deriv, delta=delta) 336 337 if mode == "interp":

C:\Python36\lib\site-packages\scipy\signal_savitzky_golay.py in savgol_coeffs(windowlength, polyorder, deriv, delta, pos, use) 135 136 # Find the least-squares solution of A*c = y --> 137 coeffs, , , = lstsq(A, y) 138 139 return coeffs

C:\Python36\lib\site-packages\scipy\linalg\basic.py in lstsq(a, b, cond, overwrite_a, overwrite_b, check_finite, lapack_driver)

LinAlgError: SVD did not converge in Linear Least Squares`

I tried to to find solutions, like installing numpy and scipy from conda, but they did not solve the problem. Anyone know how to deal with it?

[Berto70]

EDIT: I solve the issue by creating a conda environment, but if I run the program with py IDLE it crashes anyway. In the env I installed the same modules i use on python.

[benmontet]

I'm glad that solution worked for you. If I were to guess it would be that there's some difference in the version of scipy between the conda environment and the other way you're running it, and the data formatting (and/or existence of NaNs) is being handled well with one and not with the other. Since it seems like a scipy versioning issue--although that's a guess on my part--I'm going to close this issue but if you run into it again and you think there's something specific from eleanor that we can change to work around it do feel free to reopen!

[Berto70]

Yup, later I realized that there were some differences between modules versions. I was doing something wrong with the Conda Env, but I solved it and finally worked. Now I only have 4/5 years of college to figure out how to analyze the light curves I have printed out. Thank you so much!