### GetDist: scatterplots, pdf KDE estimates for 2d data, modes, 2d isoprobability contours, 1dmarginal pdf's (all in one)

Posted:

**October 19 2020**Hi there,

Forgive me if that is a silly question, but I was not able to dig its answer in a simple way from the documentation and I am not quite conversant with Python...

I have a data file with 3 columns of best fit parameters (say, [math]) from a simulation I ran in a completely independent way from COSMOMC (or CAMB). Thus my data file does not have the usual format arising from COSMOMC output. Nevertheless, I would still like, if possible, to use GetDist (via its python wrapper) to:

(1) create a kernel density estimate, from the corresponding scatterplot of $(a, b, c)$ triples, for the (normalized) probability density function [math].

(2) from item (1), determine the mode (maximum) of the estimated pdf

(3) from item (1), determine the contour lines for, e.g., the 68%, 95% and 99% highest confidence/credibility regions in the distinct 2d parameter planes

(4) still from item (1), determine the 1d marginal pdf's [math] and [math]

Finally, I would like to plot all this information in a single "triangle_plot" (à la https://getdist.readthedocs.io/en/latest/plot_gallery.html) but,

Perhaps all this is achievable, if I manage to get the identity of each of the subplots constituting the triangle plot.

Thanks in advance!

Forgive me if that is a silly question, but I was not able to dig its answer in a simple way from the documentation and I am not quite conversant with Python...

I have a data file with 3 columns of best fit parameters (say, [math]) from a simulation I ran in a completely independent way from COSMOMC (or CAMB). Thus my data file does not have the usual format arising from COSMOMC output. Nevertheless, I would still like, if possible, to use GetDist (via its python wrapper) to:

(1) create a kernel density estimate, from the corresponding scatterplot of $(a, b, c)$ triples, for the (normalized) probability density function [math].

(2) from item (1), determine the mode (maximum) of the estimated pdf

(3) from item (1), determine the contour lines for, e.g., the 68%, 95% and 99% highest confidence/credibility regions in the distinct 2d parameter planes

(4) still from item (1), determine the 1d marginal pdf's [math] and [math]

Finally, I would like to plot all this information in a single "triangle_plot" (à la https://getdist.readthedocs.io/en/latest/plot_gallery.html) but,

**additionally**, with:- 2d scatterplots of the data
- 2d modes

Perhaps all this is achievable, if I manage to get the identity of each of the subplots constituting the triangle plot.

Thanks in advance!