Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Update deprecated syntax for future rstan compatibility #2

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Update deprecated syntax for future rstan compatibility #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
fb9606f
Update array syntax
andrjohns Sep 9, 2023
72b5b74
Bump rstan ver
andrjohns Sep 9, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
6 changes: 3 additions & 3 deletions DESCRIPTION
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,15 +31,15 @@ Imports:
parallel (>= 3.6.0),
Rcpp (>= 0.12.0),
RcppParallel (>= 5.0.1),
rstan (>= 2.18.1),
rstan (>= 2.26.0),
rstantools (>= 2.2.0)
LinkingTo:
BH (>= 1.66.0),
Rcpp (>= 0.12.0),
RcppEigen (>= 0.3.3.3.0),
RcppParallel (>= 5.0.1),
rstan (>= 2.18.1),
StanHeaders (>= 2.18.0)
rstan (>= 2.26.0),
StanHeaders (>= 2.26.0)
SystemRequirements: GNU make
Suggests:
testthat
10 changes: 5 additions & 5 deletions inst/stan/legacy/lmmelsm.stan
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ Sigma_eta[k_i] = L * SD[k_i] * SD[k_i]' * L'
log(SD[k_i]) ~ N(0, ??)
*/
functions {
matrix lambda_mat(int J, int F, int[] J_f, int[,] F_ind, vector lambda_est) {
matrix lambda_mat(int J, int F, array[] int J_f, array[,] int F_ind, vector lambda_est) {
matrix[F, J] out = rep_matrix(0.0, F, J);
int count = 1;
for(f in 1:F) {
Expand All @@ -28,11 +28,11 @@ data {
int F; // Number of latent factors
int K; // Number of grouping variables

int group[N]; // Grouping indicator.
array[N] int group; // Grouping indicator.

// Indicators
int J_f[F]; // Number of indicators for each factor.
int F_ind[F,J]; // Indicator indices.
array[F] int J_f; // Number of indicators for each factor.
array[F,J] int F_ind; // Indicator indices.
matrix[N, J] x;

}
Expand Down Expand Up @@ -63,7 +63,7 @@ transformed parameters {
matrix[K, F] eta_mean = eta_mean_logsd[,1:F];
matrix[K, F] eta_sd = exp(eta_mean_logsd[,(F+1) : (F*2)]);
matrix[N, F] eta = eta_mean[group,];
matrix[F,F] U_group_eta[K]; // Will be Upper (transposed) Cholesky covariance
array[K] matrix[F,F] U_group_eta; // Will be Upper (transposed) Cholesky covariance
matrix[N, J] xhat;
matrix[N, J] shat = rep_matrix(sigma, N);
for(k in 1:K){
Expand Down
10 changes: 5 additions & 5 deletions inst/stan/legacy/lmmelsmGlm.stan
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ Sigma_eta[k_i] = L * SD[k_i] * SD[k_i]' * L'
log(SD[k_i]) ~ N(0, ??)
*/
functions {
matrix lambda_mat(int J, int F, int[] J_f, int[,] F_ind, vector lambda_est) {
matrix lambda_mat(int J, int F, array[] int J_f, array[,] int F_ind, vector lambda_est) {
matrix[F, J] out = rep_matrix(0.0, F, J);
int count = 1;
for(f in 1:F) {
Expand All @@ -28,11 +28,11 @@ data {
int F; // Number of latent factors
int K; // Number of grouping variables

int group[N]; // Grouping indicator.
array[N] int group; // Grouping indicator.

// Indicators
int J_f[F]; // Number of indicators for each factor.
int F_ind[F,J]; // Indicator indices.
array[F] int J_f; // Number of indicators for each factor.
array[F,J] int F_ind; // Indicator indices.
matrix[N, J] x;

}
Expand Down Expand Up @@ -63,7 +63,7 @@ transformed parameters {
matrix[K, F] eta_mean = eta_mean_logsd[,1:F];
matrix[K, F] eta_sd = exp(eta_mean_logsd[,(F+1) : (F*2)]);
matrix[N, F] eta = eta_mean[group,];
matrix[F,F] U_group_eta[K]; // Will be Upper (transposed) Cholesky covariance
array[K] matrix[F,F] U_group_eta; // Will be Upper (transposed) Cholesky covariance
for(k in 1:K){
U_group_eta[k] = (diag_pre_multiply(eta_sd[k], L_eta_random))';
}
Expand Down
38 changes: 19 additions & 19 deletions inst/stan/legacy/lmmelsmPredObs.stan
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
functions {

// Takes indicator spec and lambda_est, unrolls into lambda matrix.
matrix lambda_mat(int J, int F, int[] J_f, int[,] F_ind, vector lambda_est) {
matrix lambda_mat(int J, int F, array[] int J_f, array[,] int F_ind, vector lambda_est) {
matrix[F, J] out = rep_matrix(0.0, F, J);
int count = 1;
for(f in 1:F) {
Expand Down Expand Up @@ -50,22 +50,22 @@ functions {
Convert repeated measures to subject-level dataset
@return matrix[K, cols(l1)]; in order of 1:K, if using l1_to_l2_indices().
*/
matrix l1_to_l2(matrix l1, int[] indices) {
matrix l1_to_l2(matrix l1, array[] int indices) {
int K = size(indices);
int n_col = cols(l1);
matrix[K, n_col] l2 = l1[indices];
return(l2);
}

/*
Find indices for which each group, k in 1:K, first appears in int[] group.
Find indices for which each group, k in 1:K, first appears in array[] int group.
@param int K: Total number of groups.
@param int[] group: Array of group integers from L1 dataset.
@return int[K]; Array of L1 indices in which each k first appears.
@param array[] int group: Array of group integers from L1 dataset.
@return array[K] int; Array of L1 indices in which each k first appears.
*/
int[] l1_to_l2_indices(int K, int[] group) {
array[] int l1_to_l2_indices(int K, array[] int group) {
int N = size(group);
int where_l1_first_k[K] = rep_array(0, K);
array[K] int where_l1_first_k = rep_array(0, K);

for(n in 1:N) {
if(where_l1_first_k[group[n]] == 0) {
Expand All @@ -81,11 +81,11 @@ functions {
@param int R: Rows in output matrices.
@param int C: Columns in output matrices.
@param matrix mat: KxM Matrix to restructure.
@return matrix[]: Array of RxC matrices.
@return array[] matrix: Array of RxC matrices.
*/
matrix[] mat_to_mat_array(int R, int C, matrix mat) {
array[] matrix mat_to_mat_array(int R, int C, matrix mat) {
int K = rows(mat);
matrix[R, C] out[K];
array[K] matrix[R, C] out;

for(k in 1:K) {
out[k] = to_matrix(mat[k], R, C);
Expand All @@ -106,17 +106,17 @@ data {
int R; // Number of predictors (between-logsd); Fixed effects only (inherentyl a level-2 question)
int P_random; // Number of random location coefficients
int Q_random; // Number of random scale coefficients
int P_random_ind[P_random]; // Indices in x_loc corresponding to random location predictors
int Q_random_ind[Q_random]; // Indices in x_loc corresponding to random scale predictors
array[P_random] int P_random_ind; // Indices in x_loc corresponding to random location predictors
array[Q_random] int Q_random_ind; // Indices in x_loc corresponding to random scale predictors

int group[N]; // Grouping indicator
array[N] int group; // Grouping indicator
matrix[N, P] x_loc; // Location predictors
matrix[N, Q] x_sca; // Scale predictors
matrix[N, R] x_bet; // Between predictors

// Indicators
// int J_f[F]; // Number of indicators for each factor
// int F_ind[F, J]; // Indicator indices.
// array[F] int J_f; // Number of indicators for each factor
// array[F, J] int F_ind; // Indicator indices.
// matrix[N, J] y; // Indicator data.
matrix[N, F] eta_y;

Expand All @@ -129,7 +129,7 @@ data {
transformed data {
int intercept_only = P == 0 && Q == 0; // Whether intercept-only; allows quicker computations
// int lambda_total = sum(J_f);
int l1_indices[K] = l1_to_l2_indices(K, group);
array[K] int l1_indices = l1_to_l2_indices(K, group);
matrix[K, P] x_loc_l2;
matrix[K, Q] x_sca_l2;
matrix[K, R] x_bet_l2;
Expand Down Expand Up @@ -177,8 +177,8 @@ transformed parameters {
z_to_re_bet(mu_logsd_betas_random_z, mu_logsd_betas_random_L, mu_logsd_betas_random_sigma, x_bet_l2, zeta);
matrix[K, F] mu_random = mu_logsd_betas_random[, 1:F];
matrix[K, F] logsd_random = mu_logsd_betas_random[, (F+1):(F*2)];
matrix[P_random, F] mu_beta_random[K] = mat_to_mat_array(P_random, F, mu_logsd_betas_random[, (F*2 + 1):(F*2 + P_random*F)]); // TODO: Need to convert the F*P_random + F*Q_random vector to an K-array of P_random x F matrices.
matrix[Q_random, F] logsd_beta_random[K] = mat_to_mat_array(Q_random, F, mu_logsd_betas_random[, (F*2 + P_random*F + 1):(F*2 + P_random*F + Q_random*F)]);
array[K] matrix[P_random, F] mu_beta_random = mat_to_mat_array(P_random, F, mu_logsd_betas_random[, (F*2 + 1):(F*2 + P_random*F)]); // TODO: Need to convert the F*P_random + F*Q_random vector to an K-array of P_random x F matrices.
array[K] matrix[Q_random, F] logsd_beta_random = mat_to_mat_array(Q_random, F, mu_logsd_betas_random[, (F*2 + P_random*F + 1):(F*2 + P_random*F + Q_random*F)]);
matrix[N, F] eta;
matrix[N, F] eta_logsd;
// Location Predictions
Expand Down Expand Up @@ -215,7 +215,7 @@ transformed parameters {
// Stochastic realizations
// if(L2_pred_only || intercept_only) { // Compute t(L_cov) for each k.
// {
// matrix[F, F] epsilon_cov_U[K];
// array[K] matrix[F, F] epsilon_cov_U;
// for(k in 1:K) {
// epsilon_cov_U[k] = (diag_pre_multiply(exp(eta_logsd[l1_indices[k]]), epsilon_L))';
// }
Expand Down
52 changes: 26 additions & 26 deletions inst/stan/lmmelsmPred.stan
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
functions {

// Takes indicator spec and lambda_est, unrolls into lambda matrix.
matrix lambda_mat(int J, int F, int[] J_f, int[,] F_ind, vector lambda_est) {
matrix lambda_mat(int J, int F, array[] int J_f, array[,] int F_ind, vector lambda_est) {
matrix[F, J] out = rep_matrix(0.0, F, J);
int count = 1;
for(f in 1:F) {
Expand Down Expand Up @@ -75,22 +75,22 @@ functions {
Convert repeated measures to subject-level dataset
@return matrix[K, cols(l1)]; in order of 1:K, if using l1_to_l2_indices().
*/
matrix l1_to_l2(matrix l1, int[] indices) {
matrix l1_to_l2(matrix l1, array[] int indices) {
int K = size(indices);
int n_col = cols(l1);
matrix[K, n_col] l2 = l1[indices];
return(l2);
}

/*
Find indices for which each group, k in 1:K, first appears in int[] group.
Find indices for which each group, k in 1:K, first appears in array[] int group.
@param int K: Total number of groups.
@param int[] group: Array of group integers from L1 dataset.
@return int[K]; Array of L1 indices in which each k first appears.
@param array[] int group: Array of group integers from L1 dataset.
@return array[K] int; Array of L1 indices in which each k first appears.
*/
int[] l1_to_l2_indices(int K, int[] group) {
array[] int l1_to_l2_indices(int K, array[] int group) {
int N = size(group);
int where_l1_first_k[K] = rep_array(0, K);
array[K] int where_l1_first_k = rep_array(0, K);

for(n in 1:N) {
if(where_l1_first_k[group[n]] == 0) {
Expand All @@ -106,11 +106,11 @@ functions {
@param int R: Rows in output matrices.
@param int C: Columns in output matrices.
@param matrix mat: KxM Matrix to restructure.
@return matrix[]: Array of RxC matrices.
@return array[] matrix: Array of RxC matrices.
*/
matrix[] mat_to_mat_array(int R, int C, matrix mat) {
array[] matrix mat_to_mat_array(int R, int C, matrix mat) {
int K = rows(mat);
matrix[R, C] out[K];
array[K] matrix[R, C] out;

for(k in 1:K) {
out[k] = to_matrix(mat[k], R, C);
Expand All @@ -124,11 +124,11 @@ functions {
Conceptually, same as 'from:to' in R; but Stan does not have this.
@param int from
@param int to
@return int[]: Array of integers in sequence defined by from, to; inclusive.
@return array[] int: Array of integers in sequence defined by from, to; inclusive.
*/
int[] seq_from_to(int from, int to) {
array[] int seq_from_to(int from, int to) {
int length = to - from + 1;
int out[length];
array[length] int out;
for(i in 1:length) {
out[i] = from + i - 1;
}
Expand All @@ -148,17 +148,17 @@ data {
int R; // Number of predictors (between-logsd); Fixed effects only (inherently a level-2 question)
int P_random; // Number of random location coefficients
int Q_random; // Number of random scale coefficients
int P_random_ind[P_random]; // Indices in x_loc corresponding to random location predictors
int Q_random_ind[Q_random]; // Indices in x_loc corresponding to random scale predictors
array[P_random] int P_random_ind; // Indices in x_loc corresponding to random location predictors
array[Q_random] int Q_random_ind; // Indices in x_loc corresponding to random scale predictors

int group[N]; // Grouping indicator
array[N] int group; // Grouping indicator
matrix[N, P] x_loc; // Location predictors
matrix[N, Q] x_sca; // Scale predictors
matrix[N, R] x_bet; // Between predictors

// Indicators
int J_f[F]; // Number of indicators for each factor
int F_ind[F, J]; // Indicator indices.
array[F] int J_f; // Number of indicators for each factor
array[F, J] int F_ind; // Indicator indices.
matrix[N, J] y; // Indicator data.

// Options
Expand All @@ -170,7 +170,7 @@ data {
transformed data {
int intercept_only = P == 0 && Q == 0; // Whether intercept-only; allows quicker computations
int lambda_total = sum(J_f);
int l1_indices[K] = l1_to_l2_indices(K, group);
array[K] int l1_indices = l1_to_l2_indices(K, group);
// Level 2 datasets, for ReVar and efficiency.
matrix[K, P] x_loc_l2;
matrix[K, Q] x_sca_l2;
Expand All @@ -181,10 +181,10 @@ transformed data {
int re_logsd_betas = Q_random*F;
int re_total = re_intercepts + re_mu_betas + re_logsd_betas;
// Pre-compute RE matrix indices
int re_ind_mu[F] = seq_from_to(1, F);
int re_ind_logsd[F] = seq_from_to(F + 1, F*2);
int re_ind_mu_betas[re_mu_betas] = seq_from_to((F*2) + 1, (F*2) + P_random*F);
int re_ind_logsd_betas[re_logsd_betas] = seq_from_to(F*2 + P_random*F + 1, F*2 + P_random*F + Q_random*F);
array[F] int re_ind_mu = seq_from_to(1, F);
array[F] int re_ind_logsd = seq_from_to(F + 1, F*2);
array[re_mu_betas] int re_ind_mu_betas = seq_from_to((F*2) + 1, (F*2) + P_random*F);
array[re_logsd_betas] int re_ind_logsd_betas = seq_from_to(F*2 + P_random*F + 1, F*2 + P_random*F + Q_random*F);
// Pre-extract random design matrices.
matrix[N, P_random] x_loc_re = x_loc[, P_random_ind];
matrix[N, Q_random] x_sca_re = x_sca[, Q_random_ind];
Expand Down Expand Up @@ -235,8 +235,8 @@ transformed parameters {
z_to_re_bet_intercepts(mu_logsd_betas_random_z, mu_logsd_betas_random_L, mu_logsd_betas_random_sigma, x_bet_l2, zeta);
matrix[K, F] mu_random = mu_logsd_betas_random[, re_ind_mu];
matrix[K, F] logsd_random = mu_logsd_betas_random[, re_ind_logsd];
matrix[P_random, F] mu_beta_random[K] = mat_to_mat_array(P_random, F, mu_logsd_betas_random[, re_ind_mu_betas]);
matrix[Q_random, F] logsd_beta_random[K] = mat_to_mat_array(Q_random, F, mu_logsd_betas_random[,re_ind_logsd_betas]);
array[K] matrix[P_random, F] mu_beta_random = mat_to_mat_array(P_random, F, mu_logsd_betas_random[, re_ind_mu_betas]);
array[K] matrix[Q_random, F] logsd_beta_random = mat_to_mat_array(Q_random, F, mu_logsd_betas_random[,re_ind_logsd_betas]);
matrix[N, F] eta;
matrix[N, F] eta_logsd;
// Location Predictions
Expand Down Expand Up @@ -273,7 +273,7 @@ transformed parameters {
// Stochastic realizations
if(L2_pred_only || intercept_only) { // Compute t(L_cov) for each k.
{
matrix[F, F] epsilon_cov_U[K];
array[K] matrix[F, F] epsilon_cov_U;
for(k in 1:K) {
epsilon_cov_U[k] = (diag_pre_multiply(exp(eta_logsd[l1_indices[k]]), epsilon_L))';
}
Expand Down