project re-structure

Cargo.toml

@@ -43,6 +43,8 @@ nalgebra = "0.32.5"
 faer = "0.18.2"
 faer-ext = {version = "0.1.0", features = ["nalgebra", "ndarray"]}
 serde_with = "3.8.1"
+anyhow = "1.0.83"
+num-traits = "0.2.19"
 
 
 [profile.release]

benches/compare.rs

@@ -1,4 +1,4 @@
-use pmcore::routines::data::{parse_pmetrics::read_pmetrics, DataTrait};
+use pmcore::prelude::data::{parse_pmetrics::read_pmetrics, DataTrait};
 use pmcore::simulator::analytical::one_compartment_with_absorption;
 use pmcore::{prelude::*, simulator::Equation};
 use std::path::Path;

examples/new_simulator.rs

@@ -1,5 +1,5 @@
+use pmcore::prelude::data::{parse_pmetrics::read_pmetrics, DataTrait};
 use pmcore::prelude::*;
-use pmcore::routines::data::{parse_pmetrics::read_pmetrics, DataTrait};
 use pmcore::simulator::analytical::one_compartment_with_absorption;
 use pmcore::simulator::Equation;
 use std::path::Path;

examples/parse_pmetrics.rs

@@ -1,4 +1,4 @@
-use pmcore::routines::data::*;
+use pmcore::prelude::data::*;
 
 fn main() {
     let path = std::path::Path::new("examples/data/bimodal_ke.csv");

src/algorithms/npag.rs

@@ -6,6 +6,7 @@ use crate::{
         output::{CycleLog, NPCycle, NPResult},
         qr,
         settings::Settings,
+        simulator::get_population_predictions,
     },
     routines::expansion::adaptative_grid::adaptative_grid,
     simulator::{likelihood::PopulationPredictions, Equation},
@@ -16,7 +17,7 @@ use ndarray::{Array, Array1, Array2, Axis};
 use ndarray_stats::{DeviationExt, QuantileExt};
 use tokio::sync::mpsc::UnboundedSender;
 
-use super::{data::Subject, get_population_predictions};
+use super::data::Subject;
 
 const THETA_E: f64 = 1e-4; // Convergence criteria
 const THETA_G: f64 = 1e-4; // Objective function convergence criteria

src/algorithms/npod.rs

@@ -8,6 +8,7 @@ use crate::{
         output::{CycleLog, NPCycle, NPResult},
         qr,
         settings::Settings,
+        simulator::get_population_predictions,
     },
     simulator::{likelihood::PopulationPredictions, Equation},
     tui::ui::Comm,
@@ -17,7 +18,7 @@ use ndarray::{Array, Array1, Array2, Axis};
 use ndarray_stats::{DeviationExt, QuantileExt};
 use tokio::sync::mpsc::UnboundedSender;
 
-use super::{data::Subject, get_population_predictions};
+use super::data::Subject;
 
 const THETA_D: f64 = 1e-4;
 const THETA_F: f64 = 1e-2;

src/algorithms/postprob.rs

@@ -5,6 +5,7 @@ use crate::{
         ipm_faer::burke,
         output::NPResult,
         settings::Settings,
+        simulator::get_population_predictions,
     },
     simulator::Equation,
     tui::ui::Comm,
@@ -13,7 +14,7 @@ use crate::{
 use ndarray::{Array1, Array2};
 use tokio::sync::mpsc::UnboundedSender;
 
-use super::{data::Subject, get_population_predictions};
+use super::data::Subject;
 
 /// Posterior probability algorithm
 /// Reweights the prior probabilities to the observed data and error model

src/routines/data/parse_pmetrics.rs → src/data/parse_pmetrics.rs

@@ -1,4 +1,4 @@
-use crate::routines::data::*;
+use crate::prelude::data::*;
 use serde::de::{MapAccess, Visitor};
 use serde::{de, Deserialize, Deserializer, Serialize};
 use std::collections::HashMap;

src/lib.rs

@@ -52,10 +52,12 @@ pub mod algorithms;
 
 pub mod simulator;
 
+/// New data format
+pub mod data;
+
 /// Routines for the crate
 pub mod routines {
-    /// New data format
-    pub mod data;
+
     /// Routines for initializing the grid
     pub mod initialization;
     pub mod optimization {
@@ -84,9 +86,6 @@ pub mod routines {
         pub mod qr;
         pub mod sigma;
     }
-    pub mod simulation {
-        pub mod predict;
-    }
 }
 
 /// Entry points for external use of the framework.
@@ -112,12 +111,9 @@ pub mod prelude {
     pub use crate::logger;
     pub use crate::prelude::evaluation::{sigma, *};
     pub use crate::routines::condensation;
-    pub use crate::routines::data::CovariateTrait;
     pub use crate::routines::expansion::*;
     pub use crate::routines::initialization::*;
     pub use crate::routines::optimization;
-    pub use crate::routines::simulation::predict::*;
-    pub use crate::routines::simulation::*;
     pub use crate::routines::*;
     pub use crate::tui::ui::*;
     #[macro_export]

src/routines/optimization/d_optimizer.rs

@@ -5,10 +5,8 @@ use argmin::{
 use ndarray::{Array1, Axis};
 
 use crate::{
-    routines::{
-        data::Subject, evaluation::sigma::ErrorModel,
-        simulation::predict::get_population_predictions,
-    },
+    prelude::{data::Subject, simulator::get_population_predictions},
+    routines::evaluation::sigma::ErrorModel,
     simulator::Equation,
 };
 

src/simulator/analytical/mod.rs

@@ -1,4 +1,4 @@
-use crate::{routines::data::Covariates, simulator::*};
+use crate::{prelude::data::Covariates, simulator::*};
 
 // let eq = |x: &V, p: &V, t:T, rateiv: V, _cov: &Covariates|
 

src/simulator/likelihood.rs

@@ -1,6 +1,6 @@
-use crate::routines::{
-    data::Observation,
-    evaluation::sigma::{ErrorModel, ErrorType},
+use crate::{
+    prelude::data::Observation,
+    routines::evaluation::sigma::{ErrorModel, ErrorType},
 };
 
 use ndarray::{Array1, Array2, Axis};

src/simulator/mod.rs

@@ -2,17 +2,22 @@ pub mod analytical;
 pub mod cache;
 pub mod likelihood;
 pub mod ode;
+use ndarray::{parallel::prelude::*, Axis};
 use std::collections::HashMap;
 
-use self::likelihood::SubjectPredictions;
+use self::likelihood::{PopulationPredictions, SubjectPredictions};
 use crate::{
-    prelude::data::Event,
-    routines::data::{Covariates, Infusion, OccasionTrait, Subject, SubjectTrait},
+    prelude::data::{Covariates, Event, Infusion, OccasionTrait, Subject, SubjectTrait},
     simulator::likelihood::ToPrediction,
 };
-// use diffsol::vector::Vector;
 
+// use dashmap::mapref::entry::Entry;
+// use dashmap::DashMap;
+// use lazy_static::lazy_static;
+use ndarray::prelude::*;
+// use ndarray::Array1;
 use cache::*;
+use ndarray::Array2;
 
 pub type T = f64;
 // pub type V = faer::Col<T>;
@@ -214,6 +219,30 @@ impl Equation {
     }
 }
 
+pub fn get_population_predictions<'a>(
+    equation: &'a Equation,
+    subjects: &Vec<Subject>,
+    support_points: &Array2<f64>,
+    _cache: bool,
+) -> PopulationPredictions {
+    let mut pred = Array2::default((subjects.len(), support_points.nrows()).f());
+    pred.axis_iter_mut(Axis(0))
+        .into_par_iter()
+        .enumerate()
+        .for_each(|(i, mut row)| {
+            row.axis_iter_mut(Axis(0))
+                .into_par_iter()
+                .enumerate()
+                .for_each(|(j, mut element)| {
+                    let subject = subjects.get(i).unwrap();
+                    let ypred =
+                        equation.simulate_subject(subject, support_points.row(j).to_vec().as_ref());
+                    element.fill(ypred);
+                });
+        });
+    pred.into()
+}
+
 trait FromVec {
     fn from_vec(vec: Vec<f64>) -> Self;
 }