version 2.0

* Modernized the package using Rcpp and ggplot2 
* Package was profiled and repetitive overhead eliminated
* Comprehensive test suite written
* Arguments are now checked in the forward-facing API
* Logistic regression estimation and weibull survival estimation were coded in C++ using RcppEigen for vast speedups
* Parallelization improved using mirai (for windows) and forking (for *nix)
* Compile flags optimize for user's specific hardware
* Our estimands are policy values which are nonsmooth and hence the regular n-out-of-n bootstrap is inconsistent. We now default to m-out-of-n bootstrap which fixes the problem. We have also corrected the rate adjustment bug using the factor sqrt(m/n). We have used the recommended Politis, Romano & Wolf, Subsampling (1999) default of m = n^(3/4).
* To select m, Politis, Romano & Wolf (1999) and Bickel & Sakov (2008) recommend a "minimum-volatility grid method" which uses a grid of m and checking which m is stable. Now that we have the speedups, this implementation became practical and it is now found in the select_optimal_m_prop() function which should be run before the bootstrap inference.
* fixed a bug in the difference_function branch
* fixed rendering issues in the documentation

version 1.7

* Calculated and prints basic and studentized bootstrap confidence intervals by default
* Fixed warnings

version 1.6

* Decomped plot parameter into a separate function for ease of use
* Removed num_cores parameter. Please use the more standard options(mc.cores = ...)
* Corrected paper examples (on github page)

version 1.5

* Support for survival endpoints 
* Support for incidence endpoints: odds ratio, risk ratio and probability difference
* Better arguments / defaults for the main function "PTE_bootstrap_inference"
* Customization for non-standard metrics for all endpoints
* Speed improvements
* Better documentation

version 1.0

* Initial Release