---
title: "Propagatin Uncertainty"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{uncertainty}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r setup}
library(discent)
```


Current implementation scales approximately $O(n^2k)$ where $n$ and $k$ are number of demes and average samples per deme pair, respectively. 
but model is fast

bootstrap with traditional approach