Lukas Hergt

> the `kde(y)` computation itself will take time proportional to `len(x)*len(y)` (i.e. `ncompress*num_plot_samples`) The `kde()` computation indeed depends on both `len(x)` and `len(y)`, but for the accuracy (especially in 1d)...

1. done 2. `matplotlib` does some docstring interpolation but the documentation on how to do it is rather opaque.

That sort of addition to the dictionary would be possible, I guess. Another option that possibly already works (haven't tried it though) could be to set up the `axes` with...

Here an example of how to make it work with the current set-up. The colours will need manual tweaking. This might seem complicated, but extending the dictionary as proposed above...

Hi Stefan, the thing with nested sampling is that you need to watch out that the `logL` stay greater than the `logL_birth`. This corresponds to one nested sampling "step" needing...

@Stefan-Heimersheim What exactly is the first plot showing? How did you generate it? Normally the samples should be ordered by weights if I am not mistaken...

@Stefan-Heimersheim Importance sampling only works in the bulk posterior region of parameter space (that goes for both nested sampling and MCMC. With a Gaussian `logL_new` where the narrow peak is...

In the last example the `scale` is huge, which isn't ideal either. But since you used `'replace'` that shouldn't be the issue and there indeed seems to be nothing happening...

Very strange, I did test something very similar in the past, getting this plot:  where the `N(mu,sigma)` refer to normal distributions. The green importance sample of `L1 + N(2,1)`...

Sorry I didn't get around to this sooner. I just did a quick mock test (anesthetic version 2.0.0-beta.12), and my impression is that things still work. I've used the following...