mat nnls_solver(const mat & H, mat mu, const umat & mask, int max_iter, double rel_tol, int n_threads)
{
/****************************************************************************************************
* Description: sequential Coordinate-wise algorithm for non-negative least square regression problem
* A x = b, s.t. x[!m] >= 0, x[m] == 0
* Arguments:
* H : A^T * A
* mu : -A^T * b
* mask : a mask matrix (m) of same dim of x
* max_iter : maximum number of iterations
* rel_tol : stop criterion, minimum change on x between two successive iteration
* n_threads : number of threads
* Return:
* x : solution to argmin_{x, x>=0} ||Ax - b||_F^2
* Reference:
* http://cmp.felk.cvut.cz/ftp/articles/franc/Franc-TR-2005-06.pdf
* Author:
* Eric Xihui Lin <*****@*****.**>
* Version:
* 2015-11-16
****************************************************************************************************/
mat x(H.n_cols, mu.n_cols, fill::zeros);
if (n_threads < 0) n_threads = 0;
bool is_masked = !mask.empty();
#pragma omp parallel for num_threads(n_threads) schedule(dynamic)
for (int j = 0; j < mu.n_cols; j++)
{
if (is_masked && arma::all(mask.col(j)))
continue;
vec x0(H.n_cols);
// x0.fill(-9999);
double tmp;
int i = 0;
double err1, err2 = 9999;
do {
// break if all entries of col_j are masked
x0 = x.col(j);
err1 = err2;
err2 = 0;
for (int k = 0; k < H.n_cols; k++)
{
if (is_masked && mask(k,j) > 0) continue;
tmp = x(k,j) - mu(k,j) / H(k,k);
if (tmp < 0) tmp = 0;
if (tmp != x(k,j))
{
mu.col(j) += (tmp - x(k,j)) * H.col(k);
}
x(k,j) = tmp;
tmp = std::abs(x(k,j) - x0(k));
if (tmp > err2) err2 = tmp;
}
} while(++i < max_iter && std::abs(err1 - err2) / (err1 + 1e-9) > rel_tol);
}
return x;
}
mat concentrate_step(mat x, umat x_nonmiss, vec pu, vec pmd, uvec x_miss_group_match,
umat miss_group_unique, uvec miss_group_counts,
umat miss_group_obs_col, umat miss_group_mis_col, uvec miss_group_p, int miss_group_n,
int n, int n_half, int p, vec theta0, mat G, int d, int EM_maxits,
double* xh_mem, unsigned int* misgrpuqh_mem, unsigned int* msgrpcth_mem,
unsigned int* msgrpoch_mem, unsigned int* msgrpmch_mem, unsigned int* msgrpph_mem)
{
// information required for subsampling
mat x_half(xh_mem, n_half, p, false, true);
umat miss_grp_uq_half( misgrpuqh_mem, miss_group_n, p, false, true); miss_grp_uq_half.zeros();
uvec miss_grp_cts_half( msgrpcth_mem, miss_group_n, false, true); miss_grp_cts_half.zeros();
umat miss_grp_oc_half( msgrpoch_mem, miss_group_n, p, false, true); miss_grp_oc_half.zeros();
umat miss_grp_mc_half( msgrpmch_mem, miss_group_n, p, false, true); miss_grp_mc_half.zeros();
uvec miss_grp_p_half( msgrpph_mem, n_half, false, true);miss_grp_p_half.zeros();
int miss_grp_n_half;
// compute adjusted pmd
vec pmd_adj(n);
for(int i = 0; i < n; i++) pmd_adj(i) = R::pchisq(pmd(i), (double) pu(i), 1, 0);
double pmd_adj_med = median(pmd_adj);
uvec x_half_ind = find( pmd_adj <= pmd_adj_med, n_half);
// find new missing pattern based on the halved samples
int miss_new_add = -1;
for(int i = 0; i < n_half; i++){
int index = (int) x_half_ind(i);
x_half.row(i) = x.row( index );
// check whether we found a new missing pattern
urowvec check_new_miss(p);
if( i == 0 ){ check_new_miss.zeros();
} else{ check_new_miss = (x_nonmiss.row( index ) == miss_grp_uq_half.row( miss_new_add )); }
if( sum(check_new_miss)== (unsigned int) p){
// if found same missing pattern
miss_grp_cts_half( miss_new_add )++;
}else{
// if found a new missing pattern
miss_grp_uq_half.row( miss_new_add + 1) = x_nonmiss.row( index );
miss_grp_cts_half( miss_new_add + 1)++;
miss_grp_oc_half.row( miss_new_add + 1) = miss_group_obs_col.row( x_miss_group_match( index)-1 );
miss_grp_mc_half.row( miss_new_add + 1) = miss_group_mis_col.row( x_miss_group_match( index)-1 );
miss_grp_p_half( miss_new_add + 1) = miss_group_p( x_miss_group_match( index)-1);
miss_new_add++;
}
}
miss_grp_n_half = miss_new_add + 1;
mat res = CovEM(x_half, n_half, p, theta0, G, d, miss_grp_uq_half, miss_grp_cts_half,
miss_grp_oc_half, miss_grp_mc_half, miss_grp_p_half, miss_grp_n_half,
0.0001, EM_maxits);
res.shed_rows(0,1);
return res;
}
//' Rank elements within column of a matrix
//'
//' This function returns the rank of each element within each column of a
//' matrix. The highest element receives the highest rank.
//'
//' @param sortedIdx is the input matrix
//' @return a rank matrix
// [[Rcpp::export]]
umat rankIndex(const umat& sortedIdx) {
int N = sortedIdx.n_rows;
int M = sortedIdx.n_cols;
umat rankedIdx(N,M);
for(int iX=0; iX<N; iX++) {
for(int jX=0; jX<M; jX++) {
rankedIdx.at(sortedIdx.at(iX,jX), jX) = iX;
}
}
return rankedIdx;
}
void subsampling(double* subsample_mem, unsigned int* subsamp_nonmis_mem,
mat x, umat x_nonmiss, int nSubsampleSize, int p, uvec subsample_id)
{
mat subsamp( subsample_mem, nSubsampleSize, p, false, true);
umat subsamp_nonmiss( subsamp_nonmis_mem, nSubsampleSize, p, false, true);
uvec subsample_id_shuff = shuffle(subsample_id);
for(int i = 0; i < nSubsampleSize; i++)
{
subsamp.row(i) = x.row( subsample_id_shuff(i) );
subsamp_nonmiss.row(i) = x_nonmiss.row( subsample_id_shuff(i) );
}
}
rowvec star(const rowvec& vphi, const umat& vxicat){
rowvec vphis = vphi;
umat::const_iterator iter = vxicat.begin();
int len = vxicat.n_elem;
int L = vphi.n_elem/2/len;
for(int i=0; i< len; i++){
for(int l=(*iter);l<L;l++){
vphis[l*len+i] = 0;
vphis[(l+L)*len+i] = 0;
}
iter++;
}
return vphis;
}
void phi_design(mat& U, const uvec& sti, const int L, umat& vxicat, cube& wZ, const uvec& WTIME){
int P=U.n_cols, wT = WTIME.n_elem, t;
int Psq=P*P;
wZ.zeros();
for(int wt=0; wt<wT; wt++){
t = WTIME[wt];
for(int l=1;l<=L;l++){
if(sti[t-l]){
for(int p=0;p<P;p++){
uvec cond= (vxicat.col(p) >= l);
wZ.slice(wt).col(p).subvec((l-1)*Psq+ p*P,(l-1)*Psq+(p+1)*P-1) = (cond)%trans(U.row(t-l));
}
wZ.slice(wt).rows((L+l-1)*Psq,(L+l)*Psq-1).zeros();
}else{
wZ.slice(wt).rows((l-1)*Psq,l*Psq-1).zeros();
for(int p=0;p<P;p++){
uvec cond= (vxicat.col(p) >= l);
wZ.slice(wt).col(p).subvec((L+l-1)*Psq+p*P,(L+l-1)*Psq+(p+1)*P-1)=cond%trans(U.row(t-l));
}
}
}
}
}
void assignWinners(mat bids, rowvec prices, umat & assignments) {
uword
winnerIdx = 0,
nItems = prices.size();
double winningBid;
for(int item = 0; item < nItems; item++) {
vec winner = getMaxItemBid(item, bids);
winnerIdx = winner(0);
winningBid = winner(1);
if(winningBid < 0.0)
continue;
prices(item) += winningBid;
assignments.col(item).fill(0);
assignments(winnerIdx, item) = 1;
}
}
// function used internally, which computes lasso fits for subsets containing a
// small number of observations (typically only 3) and returns the indices of
// the respective h observations with the smallest absolute residuals
umat sparseSubsets(const mat& x, const vec& y, const double& lambda,
const uword& h, const umat& subsets, const bool& normalize,
const bool& useIntercept, const double& eps, const bool& useGram) {
const uword nsamp = subsets.n_cols;
umat indices(h, nsamp);
for(uword k = 0; k < nsamp; k++) {
// compute lasso fit
double intercept, crit;
vec coefficients, residuals;
fastLasso(x, y, lambda, true, subsets.unsafe_col(k), normalize,
useIntercept, eps, useGram, false, intercept, coefficients,
residuals, crit);
// find h observations with smallest absolute residuals
indices.col(k) = findSmallest(abs(residuals), h);
}
return indices;
}
//' C++ wrapper for Gale-Shapley Algorithm
//'
//' This function provides an R wrapper for the C++ backend. Users should not
//' call this function directly and instead use
//' \code{\link{galeShapley.marriageMarket}} or
//' \code{\link{galeShapley.collegeAdmissions}}.
//'
//' @param proposerPref is a matrix with the preference order of the proposing
//' side of the market. If there are \code{n} proposers and \code{m} reviewers
//' in the market, then this matrix will be of dimension \code{m} by \code{n}.
//' The \code{i,j}th element refers to \code{j}'s \code{i}th most favorite
//' partner. Preference orders must be complete and specified using C++
//' indexing (starting at 0).
//' @param reviewerUtils is a matrix with cardinal utilities of the courted side
//' of the market. If there are \code{n} proposers and \code{m} reviewers, then
//' this matrix will be of dimension \code{n} by \code{m}. The \code{i,j}th
//' element refers to the payoff that individual \code{j} receives from being
//' matched to individual \code{i}.
//' @return A list with elements that specify who is matched to whom. Suppose
//' there are \code{n} proposers and \code{m} reviewers. The list contains
//' the following items:
//' \itemize{
//' \item{\code{proposals} is a vector of length \code{n} whose \code{i}th
//' element contains the number of the reviewer that proposer \code{i} is
//' matched to using C++ indexing. Proposers that remain unmatched will be
//' listed as being matched to \code{m}.}
//' \item{\code{engagements} is a vector of length \code{m} whose \code{j}th
//' element contains the number of the proposer that reviewer \code{j} is
//' matched to using C++ indexing. Reviwers that remain unmatched will be
//' listed as being matched to \code{n}.}
//' }
// [[Rcpp::export]]
List cpp_wrapper_galeshapley(const umat& proposerPref, const mat& reviewerUtils) {
// number of proposers (men)
int M = proposerPref.n_cols;
// number of reviewers (women)
int N = proposerPref.n_rows;
// initialize engagements, proposals
vec engagements(N), proposals(M);
// create an integer queue of bachelors
// the idea of using queues for this problem is borrowed from
// http://rosettacode.org/wiki/Stable_marriage_problem#C.2B.2B
queue<int> bachelors;
// set all proposals to N (aka no proposals)
proposals.fill(N);
// set all engagements to M (aka no engagements)
engagements.fill(M);
// every proposer starts out as a bachelor
for(int iX=M-1; iX >= 0; iX--) {
bachelors.push(iX);
}
// loop until there are no more proposals to be made
while (!bachelors.empty()) {
// get the index of the proposer
int proposer = bachelors.front();
// get the proposer's preferences: we use a raw pointer to the memory
// used by the column `proposer` for performance reasons (this is to avoid
// making a copy of the proposers vector of preferences)
const uword * proposerPrefcol = proposerPref.colptr(proposer);
// find the best available match for proposer
for(int jX=0; jX<N; jX++) {
// get the index of the reviewer that the proposer is interested in
// by dereferencing the pointer; increment the pointer after use (not its value)
const uword wX = *proposerPrefcol++;
// check if wX is available (`M` means unmatched)
if(engagements(wX)==M) {
// if available, then form a match
engagements(wX) = proposer;
proposals(proposer) = wX;
// go to the next proposer
break;
}
// wX is already matched, let's see if wX can be poached
if(reviewerUtils(proposer, wX) > reviewerUtils(engagements(wX), wX)) {
// wX's previous partner becomes unmatched (`N` means unmatched)
proposals(engagements(wX)) = N;
bachelors.push(engagements(wX));
// proposer and wX form a match
engagements(wX) = proposer;
proposals(proposer) = wX;
// go to the next proposer
break;
}
}
// remove proposer from bachelor queue: proposer will remain unmatched
bachelors.pop();
}
return List::create(
_["proposals"] = proposals,
_["engagements"] = engagements);
}
//[[Rcpp::export]]
Rcpp::List nnmf(const mat & A, const unsigned int k, mat W, mat H, umat Wm, umat Hm,
const vec & alpha, const vec & beta, const unsigned int max_iter, const double rel_tol,
const int n_threads, const int verbose, const bool show_warning, const unsigned int inner_max_iter,
const double inner_rel_tol, const int method, unsigned int trace)
{
/******************************************************************************************************
* Non-negative Matrix Factorization(NNMF) using alternating scheme
* ----------------------------------------------------------------
* Description:
* Decompose matrix A such that
* A = W H
* Arguments:
* A : Matrix to be decomposed
* W, H : Initial matrices of W and H, where ncol(W) = nrow(H) = k. # of rows/columns of W/H could be 0
* Wm, Hm : Masks of W and H, s.t. masked entries are no-updated and fixed to initial values
* alpha : [L2, angle, L1] regularization on W (non-masked entries)
* beta : [L2, angle, L1] regularization on H (non-masked entries)
* max_iter : Maximum number of iteration
* rel_tol : Relative tolerance between two successive iterations, = |e2-e1|/avg(e1, e2)
* n_threads : Number of threads (openMP)
* verbose : Either 0 = no any tracking, 1 == progression bar, 2 == print iteration info
* show_warning : If to show warning if targeted `tol` is not reached
* inner_max_iter : Maximum number of iterations passed to each inner W or H matrix updating loop
* inner_rel_tol : Relative tolerance passed to inner W or H matrix updating loop, = |e2-e1|/avg(e1, e2)
* method : Integer of 1, 2, 3 or 4, which encodes methods
* : 1 = sequential coordinate-wise minimization using square loss
* : 2 = Lee's multiplicative update with square loss, which is re-scaled gradient descent
* : 3 = sequentially quadratic approximated minimization with KL-divergence
* : 4 = Lee's multiplicative update with KL-divergence, which is re-scaled gradient descent
* trace : A positive integer, error will be checked very 'trace' iterations. Computing WH can be very expansive,
* : so one may not want to check error A-WH every single iteration
* Return:
* A list (Rcpp::List) of
* W, H : resulting W and H matrices
* mse_error : a vector of mean square error (divided by number of non-missings)
* mkl_error : a vector (length = number of iterations) of mean KL-distance
* target_error : a vector of loss (0.5*mse or mkl), plus constraints
* average_epoch : a vector of average epochs (one complete swap over W and H)
* Author:
* Eric Xihui Lin <*****@*****.**>
* Version:
* 2015-12-11
******************************************************************************************************/
unsigned int n = A.n_rows;
unsigned int m = A.n_cols;
//int k = H.n_rows; // decomposition rank k
unsigned int N_non_missing = n*m;
if (trace < 1) trace = 1;
unsigned int err_len = (unsigned int)std::ceil(double(max_iter)/double(trace)) + 1;
vec mse_err(err_len), mkl_err(err_len), terr(err_len), ave_epoch(err_len);
// check progression
bool show_progress = false;
if (verbose == 1) show_progress = true;
Progress prgrss(max_iter, show_progress);
double rel_err = rel_tol + 1;
double terr_last = 1e99;
uvec non_missing;
bool any_missing = !A.is_finite();
if (any_missing)
{
non_missing = find_finite(A);
N_non_missing = non_missing.n_elem;
mkl_err.fill(mean((A.elem(non_missing)+TINY_NUM) % log(A.elem(non_missing)+TINY_NUM) - A.elem(non_missing)));
}
else
mkl_err.fill(mean(mean((A+TINY_NUM) % log(A+TINY_NUM) - A))); // fixed part in KL-dist, mean(A log(A) - A)
if (Wm.empty())
Wm.resize(0, n);
else
inplace_trans(Wm);
if (Hm.empty())
Hm.resize(0, m);
if (W.empty())
{
W.randu(k, n);
W *= 0.01;
if (!Wm.empty())
W.elem(find(Wm > 0)).fill(0.0);
}
else
inplace_trans(W);
if (H.empty())
{
H.randu(k, m);
H *= 0.01;
if (!Hm.empty())
H.elem(find(Hm > 0)).fill(0.0);
}
if (verbose == 2)
{
Rprintf("\n%10s | %10s | %10s | %10s | %10s\n", "Iteration", "MSE", "MKL", "Target", "Rel. Err.");
Rprintf("--------------------------------------------------------------\n");
}
int total_raw_iter = 0;
unsigned int i = 0;
unsigned int i_e = 0; // index for error checking
for(; i < max_iter && std::abs(rel_err) > rel_tol; i++)
{
Rcpp::checkUserInterrupt();
prgrss.increment();
if (any_missing)
{
// update W
total_raw_iter += update_with_missing(W, H, A.t(), Wm, alpha, inner_max_iter, inner_rel_tol, n_threads, method);
// update H
total_raw_iter += update_with_missing(H, W, A, Hm, beta, inner_max_iter, inner_rel_tol, n_threads, method);
if (i % trace == 0)
{
const mat & Ahat = W.t()*H;
mse_err(i_e) = mean(square((A - Ahat).eval().elem(non_missing)));
mkl_err(i_e) += mean((-(A+TINY_NUM) % log(Ahat+TINY_NUM) + Ahat).eval().elem(non_missing));
}
}
else
{
// update W
total_raw_iter += update(W, H, A.t(), Wm, alpha, inner_max_iter, inner_rel_tol, n_threads, method);
// update H
total_raw_iter += update(H, W, A, Hm, beta, inner_max_iter, inner_rel_tol, n_threads, method);
if (i % trace == 0)
{
const mat & Ahat = W.t()*H;
mse_err(i_e) = mean(mean(square((A - Ahat))));
mkl_err(i_e) += mean(mean(-(A+TINY_NUM) % log(Ahat+TINY_NUM) + Ahat));
}
}
if (i % trace == 0)
{
ave_epoch(i_e) = double(total_raw_iter)/(n+m);
if (method < 3) // mse based
terr(i_e) = 0.5*mse_err(i_e);
else // KL based
terr(i_e) = mkl_err(i_e);
add_penalty(i_e, terr, W, H, N_non_missing, alpha, beta);
rel_err = 2*(terr_last - terr(i_e)) / (terr_last + terr(i_e) + TINY_NUM );
terr_last = terr(i_e);
if (verbose == 2)
Rprintf("%10d | %10.4f | %10.4f | %10.4f | %10.g\n", i+1, mse_err(i_e), mkl_err(i_e), terr(i_e), rel_err);
total_raw_iter = 0; // reset to 0
++i_e;
}
}
// compute error of the last iteration
if ((i-1) % trace != 0)
{
if (any_missing)
{
const mat & Ahat = W.t()*H;
mse_err(i_e) = mean(square((A - Ahat).eval().elem(non_missing)));
mkl_err(i_e) += mean((-(A+TINY_NUM) % log(Ahat+TINY_NUM) + Ahat).eval().elem(non_missing));
}
else
{
const mat & Ahat = W.t()*H;
mse_err(i_e) = mean(mean(square((A - Ahat))));
mkl_err(i_e) += mean(mean(-(A+TINY_NUM) % log(Ahat+TINY_NUM) + Ahat));
}
ave_epoch(i_e) = double(total_raw_iter)/(n+m);
if (method < 3) // mse based
terr(i_e) = 0.5*mse_err(i_e);
else // KL based
terr(i_e) = mkl_err(i_e);
add_penalty(i_e, terr, W, H, N_non_missing, alpha, beta);
rel_err = 2*(terr_last - terr(i_e)) / (terr_last + terr(i_e) + TINY_NUM );
terr_last = terr(i_e);
if (verbose == 2)
Rprintf("%10d | %10.4f | %10.4f | %10.4f | %10.g\n", i+1, mse_err(i_e), mkl_err(i_e), terr(i_e), rel_err);
++i_e;
}
if (verbose == 2)
{
Rprintf("--------------------------------------------------------------\n");
Rprintf("%10s | %10s | %10s | %10s | %10s\n\n", "Iteration", "MSE", "MKL", "Target", "Rel. Err.");
}
if (i_e < err_len)
{
mse_err.resize(i_e);
mkl_err.resize(i_e);
terr.resize(i_e);
ave_epoch.resize(i_e);
}
if (show_warning && rel_err > rel_tol)
Rcpp::warning("Target tolerance not reached. Try a larger max.iter.");
return Rcpp::List::create(
Rcpp::Named("W") = W.t(),
Rcpp::Named("H") = H,
Rcpp::Named("mse_error") = mse_err,
Rcpp::Named("mkl_error") = mkl_err,
Rcpp::Named("target_error") = terr,
Rcpp::Named("average_epoch") = ave_epoch,
Rcpp::Named("n_iteration") = i
);
}
mat nnls_solver_with_missing(const mat & A, const mat & W, const mat & W1, const mat & H2, const umat & mask,
const double & eta, const double & beta, int max_iter, double rel_tol, int n_threads)
{
// A = [W, W1, W2] [H, H1, H2]^T.
// Where A may have missing values
// Note that here in the input W = [W, W2]
// compute x = [H, H1]^T given W, W2
// A0 = W2*H2 is empty when H2 is empty (no partial info in H)
// Return: x = [H, H1]
int n = A.n_rows, m = A.n_cols;
int k = W.n_cols - H2.n_cols;
int kW = W1.n_cols;
int nH = k+kW;
mat x(nH, m, fill::zeros);
if (n_threads < 0) n_threads = 0;
bool is_masked = !mask.empty();
#pragma omp parallel for num_threads(n_threads) schedule(dynamic)
for (int j = 0; j < m; j++)
{
// break if all entries of col_j are masked
if (is_masked && arma::all(mask.col(j)))
continue;
uvec non_missing = find_finite(A.col(j));
mat WtW(nH, nH); // WtW
update_WtW(WtW, W.rows(non_missing), W1.rows(non_missing), H2);
if (beta > 0) WtW += beta;
if (eta > 0) WtW.diag() += eta;
mat mu(nH, 1); // -WtA
uvec jv(1);
jv(0) = j;
//non_missing.t().print("non_missing = ");
//std::cout << "1.1" << std::endl;
if (H2.empty())
update_WtA(mu, W.rows(non_missing), W1.rows(non_missing), H2, A.submat(non_missing, jv));
else
update_WtA(mu, W.rows(non_missing), W1.rows(non_missing), H2.rows(j, j), A.submat(non_missing, jv));
//std::cout << "1.5" << std::endl;
vec x0(nH);
double tmp;
int i = 0;
double err1, err2 = 9999;
do {
x0 = x.col(j);
err1 = err2;
err2 = 0;
for (int l = 0; l < nH; l++)
{
if (is_masked && mask(l,j) > 0) continue;
tmp = x(l,j) - mu(l,0) / WtW(l,l);
if (tmp < 0) tmp = 0;
if (tmp != x(l,j))
{
mu.col(0) += (tmp - x(l,j)) * WtW.col(l);
}
x(l,j) = tmp;
tmp = std::abs(x(l,j) - x0(l));
if (tmp > err2) err2 = tmp;
}
} while(++i < max_iter && std::abs(err1 - err2) / (err1 + 1e-9) > rel_tol);
}
return x;
}
