Added type='full' for logistic and removed Xordered from gaussian and…

.gitignore

@@ -0,0 +1,26 @@
+*.rds
+tests/*.rds
+tests/*/*.rds
+tests/*/*/*.rds
+tests/*.csv
+tests/*/*.csv
+tests/*/*/*.csv
+**DS_Store**
+**Rcheck**
+**tar.gz
+**Rapp.history
+**.pdf
+**.RData
+**.o
+**.so
+forLater/josh/**
+*/R/RcppExports.R
+*/src/RcppExports.cpp
+*~
+*/*~
+*/*/*~
+# this code is copied from C-software
+selectiveInference/src/matrixcomps.*
+selectiveInference/src/debias.*
+selectiveInference/src/randomized_lasso.*
+selectiveInference/src/quadratic_program*.c

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "C-software"]
+	path = C-software
+	url = https://github.com/selective-inference/C-software.git

.travis.yml

@@ -0,0 +1,21 @@
+language: r
+cache: packages
+sudo: false
+r:
+  #- oldrel
+  - release
+  - devel
+addons:
+   apt: 
+      packages: libmpfr-dev 
+warnings_are_errors: true
+before_install:
+  - tlmgr install index # for texlive and vignette?
+  - R -e 'install.packages(c("Rcpp", "intervals", "adaptMCMC", "glmnet"), repos="http://cloud.r-project.org")'
+  - cd C-software
+  - git submodule init
+  - git submodule update
+  - cd ..
+  - make src
+  - make Rcpp
+  - cd selectiveInference

Makefile

@@ -0,0 +1,17 @@
+Rcpp: 
+	- rm -f selectiveInference/src/RcppExports.cpp
+	- rm -f selectiveInference/R/RcppExports.R
+	Rscript -e "library(Rcpp); Rcpp::compileAttributes('selectiveInference')"
+
+install: Rcpp src
+	R CMD INSTALL selectiveInference
+
+build: src 
+	R CMD build selectiveInference
+
+src:
+	cp C-software/src/* selectiveInference/src
+
+check: Rcpp build 
+	R CMD build selectiveInference
+	R CMD check selectiveInference_1.2.2.tar.gz # fix this to be a script variable

README.md

@@ -0,0 +1,39 @@
+# R-software
+R software for [selective inference](http://cran.r-project.org/web/packages/selectiveInference/).  
+Authors: Ryan Tibshirani, Rob Tibshirani, Jonathan Taylor, Joshua Loftus, Stephen Reid, Jelena Markovic  
+Maintainer: Rob Tibshirani <tibs@stanford.edu>
+
+New tools for inference after selection, for use with forward stepwise regression, least angle regression, the lasso, and the many means problem. The package is available on [CRAN](http://cran.r-project.org/web/packages/selectiveInference/). See [this paper](http://www.pnas.org/content/112/25/7629.full) for a high level introduction to selective inference.
+
+Code is in the directory selectiveInference/R.
+* funs.common.R: Basic functions used by many other functions, such as standardization.
+* funs.fixed.R: Inference for LASSO at a fixed, deterministic value of lambda.
+* funs.fs.R: Inference for forward stepwise.
+* funs.groupfs.R: Inference for forward stepwise with groups of variables, e.g. factors.
+* funs.inf.R: Common functions for inference with fixed, fs, lar, and manymeans (but not group).
+* funs.lar.R: Inference for least angle regression.
+* funs.max.R: Some numerical approximations. Deprecated?
+
+## Installation
+The latest release of the package can be installed through CRAN:
+
+```R
+install.packages("selectiveInference")
+```
+Code in repo is under development and may be unstable.
+
+## For development
+
+You will have to run
+
+```
+make Rcpp
+```
+
+## To test
+
+```
+make check
+```
+
+These tests are also run on [travis](http://travis-ci.org) based on .travis.yml setup. 

forLater/estimateLambda.Rd

@@ -0,0 +1,70 @@
+\name{estimateLambda}
+\alias{estimateLambda}
+%- Also NEED an '\alias' for EACH other topic documented here.
+\title{
+Estimates the lasso tuning parameter lambda.
+}
+\description{
+Estimates the lasso tuning parameter lambda, for use in the selectiveInference 
+package
+}
+\usage{
+estimateLambda(x, sigma, nsamp=1000)
+}
+\arguments{
+  \item{x}{
+Matrix of predictors (n by p)
+}
+  \item{sigma}{
+Estimate of error standard deviation
+}
+\item{nsamp}{Number of Monte carlo samples used for the estimation.}
+}
+\details{
+This function estimates the lasso tuning parameter lambda, using the estimate
+2*E(||X^T eps||_infty)  where eps ~ N(0,sigma^2), a vector of length n.
+This estimate was proposed by Negahban et  al (2012).
+}
+\value{
+\item{sigmahat}{The estimate of sigma}
+\item{df}{The degrees of freedom of lasso fit used}
+}
+\references{
+Negahban, S. N.,
+Ravikumar, P.,
+Wainwright, M. J.
+and Yu, B.
+(2012).  A unified
+framework for high-dimensional analysis of
+M-estimators with decomposable regularizers.
+Statistical Science vol. 27, p 538-557.
+}
+
+\author{Ryan Tibshirani, Rob Tibshirani, Jonathan Taylor, Joshua Loftus, Stephen Reid}
+
+\examples{
+#NOT RUN
+#set.seed(43)
+#n=50
+#p=10
+#sigma=.7
+#x=matrix(rnorm(n*p),n,p)
+#x=scale(x,T,F)
+#beta=c(3,2,0,0,rep(0,p-4))
+#y=x%*%beta+sigma*rnorm(n)
+#y=y-mean(y)
+#
+#estimate lambda usingthe  known value of sigma
+#lamhat=estimateLambda(x,sigma=.7)
+#
+#first estimate  sigma
+#sigmahat=estimateSigma(x,y)$sigmahat
+#lamhat=estimateLambda(x,sigma=sigmahat)
+
+#compare to estimate from cv
+
+#out=cv.glmnet(x,y)
+#out$lambda.min*n    #remember that value from glmnet must be
+                     # multiplied by n, to make it comparable.
+}
+

forLater/fixedLassoPoly.Rd

@@ -0,0 +1,81 @@
+\name{fixedLassoPoly}
+\alias{fixedLassoPoly}
+
+\title{
+Compute polyhedral constraints for a LASSO problem with
+a fixed value of lambda.
+}
+\description{
+Compute polyhedral representation of the selection region of Lee et al. (2016).
+By construction, y should satisfy A %*% y elementwise less then or equal b.
+}
+\usage{
+fixedLassoPoly(X, y, lambda, beta, active, inactive = FALSE)
+}
+\arguments{
+\item{X}{
+Design matrix of LASSO problem.
+}
+\item{y}{
+Response of LASSO problem.
+}
+\item{lambda}{
+Value of regularization parameter.
+}      
+\item{beta}{
+Solution of LASSO problem with regularization parameter set to lambda.
+}      
+\item{active}{
+Active set of the LASSO problem as a boolean vector. Should correspond
+to the non-zeros of beta.
+} 
+\item{inactive}{
+Form the inactive constraints as well?
+} 
+}
+\details{
+This function computes
+the  polyhedral representation of the selection region of Lee et al. (2016).
+}
+
+\value{  
+\item{A}{Linear part of the affine inequalities.}
+\item{b}{RHS offset the affine inequalities.}
+}
+
+\references{
+Jason Lee, Dennis Sun, Yuekai Sun, and Jonathan Taylor (2016). 
+Exact post-selection inference, with application to the lasso. Annals of Statistics, 44(3), 907-927.
+
+Jonathan Taylor and Robert Tibshirani (2017) Post-selection inference for math L1-penalized likelihood models.
+Canadian Journal of Statistics, xx, 1-21. (Volume still not posted)
+}
+\author{Ryan Tibshirani, Rob Tibshirani, Jonathan Taylor, Joshua Loftus, Stephen Reid}
+
+\examples{
+
+set.seed(43)
+n = 50
+p = 10
+sigma = 1
+
+x = matrix(rnorm(n*p),n,p)
+x = scale(x,TRUE,TRUE)
+
+beta = c(3,2,rep(0,p-2))
+y = x\%*\%beta + sigma*rnorm(n)
+
+# first run glmnet
+gfit = glmnet(x,y,standardize=FALSE)
+
+# extract coef for a given lambda; note the 1/n factor!
+# (and we don't save the intercept term)
+lambda = .8
+beta = coef(gfit, s=lambda/n, exact=TRUE)[-1]
+active = (beta != 0)
+
+fixedLassoPoly(x, y, lambda, beta, active)
+fixedLassoPoly(x, y, lambda, beta, active, inactive=TRUE)
+
+}
+