also show corners since it shows the calibrated parameters

Author: malihass

papers/tutorial/calibration/README.md

@@ -42,22 +42,28 @@ The following plots should be generated. The first one shows the training and te
 For the calibration, we can use the true function or the surrogate model. The objective PDF is assumed to be noiseless. Even though the observation is noiseless, an uncertainty is computed to account for the missing physics.
 The likelihood uncertainty that represents the missing physics is optimized to ensure an uncertainty band overlap
 
-Calibrate without a surrogate for `alpha=5, beta=5`: `python tut_calibration.py --alpha 5 --beta 5`
-Calibrate with a surrogate for `alpha=5, beta=5`: `python tut_calibration.py -useNN --alpha 5 --beta 5`
+Calibrate without a surrogate for `alpha=5, beta=5`: `python tut_calibration.py --alpha 5 --beta 5` (Left two plots)
+
+Calibrate with a surrogate for `alpha=5, beta=5`: `python tut_calibration.py -useNN --alpha 5 --beta 5` (Right two plots)
 
 <p align="center">
 <img src="/papers/tutorial/calibration/assets/True_opt_5.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/True_opt_5.0_5.0_corner.png" width="225" height="187.5"/>
 <img src="/papers/tutorial/calibration/assets/Surr_opt_5.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/Surr_opt_5.0_5.0_corner.png" width="225" height="187.5"/>
 </p>
 
 
-Calibrate without a surrogate for `alpha=2, beta=5`: `python tut_calibration.py --alpha 2 --beta 5`
-Calibrate with a surrogate for `alpha=2, beta=5`: `python tut_calibration.py -useNN --alpha 2 --beta 5`
+Calibrate without a surrogate for `alpha=2, beta=5`: `python tut_calibration.py --alpha 2 --beta 5` (Left two plots)
+
+Calibrate with a surrogate for `alpha=2, beta=5`: `python tut_calibration.py -useNN --alpha 2 --beta 5` (Right two plots)
 
 
 <p align="center">
 <img src="/papers/tutorial/calibration/assets/True_opt_2.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/True_opt_2.0_5.0_corner.png" width="225" height="187.5"/>
 <img src="/papers/tutorial/calibration/assets/Surr_opt_2.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/Surr_opt_2.0_5.0_corner.png" width="225" height="187.5"/>
 </p>
 
 Clearly, the amount of missing physics vary depending on the observations.
@@ -68,21 +74,26 @@ Using surrogate gives similar predictions as when not using a surrogate. But the
 
 The same suite can be done by calibrating the likelihood uncertainty in lieu of optimizing it (with a bissection search). This has the advantage of rapid calibration since only one calibration is needed. Here the uncertainty minimizes the negative log likelihood.
 
-Calibrate without a surrogate for `alpha=5, beta=5`: `python tut_calibration.py -cal_err --alpha 5 --beta 5`
-Calibrate with a surrogate for `alpha=5, beta=5`: `python tut_calibration.py -useNN -cal_err --alpha 5 --beta 5`
+Calibrate without a surrogate for `alpha=5, beta=5`: `python tut_calibration.py -cal_err --alpha 5 --beta 5` (Left two plots)
+
+Calibrate with a surrogate for `alpha=5, beta=5`: `python tut_calibration.py -useNN -cal_err --alpha 5 --beta 5` (Right two plots)
 
 <p align="center">
 <img src="/papers/tutorial/calibration/assets/True_cal_5.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/True_cal_5.0_5.0_corner.png" width="225" height="187.5"/>
 <img src="/papers/tutorial/calibration/assets/Surr_cal_5.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/Surr_cal_5.0_5.0_corner.png" width="225" height="187.5"/>
 </p>
 
 
-Calibrate without a surrogate for `alpha=2, beta=5`: `python tut_calibration.py -cal_err --alpha 2 --beta 5`
-Calibrate with a surrogate for `alpha=2, beta=5`: `python tut_calibration.py -useNN -cal_err --alpha 2 --beta 5`
+Calibrate without a surrogate for `alpha=2, beta=5`: `python tut_calibration.py -cal_err --alpha 2 --beta 5` (Left two plots)
+Calibrate with a surrogate for `alpha=2, beta=5`: `python tut_calibration.py -useNN -cal_err --alpha 2 --beta 5` (Right two plots)
 
 
 <p align="center">
 <img src="/papers/tutorial/calibration/assets/True_cal_2.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/True_cal_2.0_5.0_corner.png" width="225" height="187.5"/>
 <img src="/papers/tutorial/calibration/assets/Surr_cal_2.0_5.0_prop.png" width="225" height="187.5"/>
+<img src="/papers/tutorial/calibration/assets/Surr_cal_2.0_5.0_corner.png" width="225" height="187.5"/>
 </p>