Differences

This shows you the differences between two versions of the page.

--- tests:astropars:challenge3 [2014/10/31 10:42] – [GSP-Phot vs. SICK] randrae
+++ tests:astropars:challenge3 [2022/10/24 12:26] (current) – external edit 127.0.0.1
@@ Line 14: / Line 14: @@
 == Conclusions ==
-  *
+  * GSP-Phot and SICK overall agree very nicely. No algorithm clearly outperforms the other.
-  * SICK may suffer from some convergence problems.
+  * logg: GSP-Phot is slightly better than SICK. SICK may suffer from some convergence problems (red dots below dwarfs).
+  * [Fe/H]: SICK is slightly better than GSP-Phot. GSP-Phot appears to systematically underestimate [Fe/H] by ~1dex for some stars. These problematic stars are fainter than G~15, i.e., SICK estimates [Fe/H] more robustly at lower S/N than GSP-Phot.
 == Test data ==
@@ Line 23: / Line 24: @@
   * Teff from 4000K to 8000K.
   * Test spectra created by GOG. (Not by our own forward models!)
+== Examples of posterior samples ==
+  * Posterior distributions mostly well behaved for BP/RP, i.e., fast convergence, unimodal.
+  * Even in simple case the posteriors cannot be well approximated by a Gaussian.
+  * Rare examples with multimodal posterior.
+{{:tests:example-MCMC-unimodal.png|}}
+{{:tests:example-MCMC-multimodal.png|}}
 == Results for fitting BP/RP alone ==
@@ Line 37: / Line 48: @@
 {{:tests:sick-gspphot-logg.png|}}
 {{:tests:sick-gspphot-feh.png|}}
+== Future tests ==
+  * Compare initial guesses of GSP-Phot and SICK. (Possibly SVM fails for noisy data, which may explain why SICK estimates [Fe/H] better at faint magnitudes.)
+===== SICK: BP/RP alone vs. RVS alone =====
+===== SICK: Fitting BP/RP and RVS simultaneously =====