Fix sim.arg crash at high rho

R/Host.R

@@ -82,6 +82,35 @@ Host <- R6Class(
       private$pathogens[[length(private$pathogens)+1]] <- new.pathogen
     },
 
+    # --- fixed-size lineage pool (for recombination parent sampling) ---
+    # Fixes exponential lineage growth under recombination: instead of
+    # always creating a brand-new ancestral lineage, recombination should
+    # sample a parent from a FIXED pool of n (= population size) possible
+    # lineages, only "activating" a previously-inactive one when chosen.
+    # Total pool size never changes; only how many slots are active does.
+    init.pool = function(n) {
+      if (is.null(private$pool.size)) {
+        private$pool.size <- n
+        private$lineage.pool <- vector("list", n)
+      }
+    },
+    get.pool.size = function() { private$pool.size },
+    is.pool.initialized = function() { !is.null(private$pool.size) },
+    is.slot.active = function(slot.id) { !is.null(private$lineage.pool[[slot.id]]) },
+    get.slot.occupant = function(slot.id) { private$lineage.pool[[slot.id]] },
+    activate.slot = function(slot.id, pathogen) {
+      private$lineage.pool[[slot.id]] <- pathogen
+    },
+    deactivate.slot = function(slot.id) {
+      private$lineage.pool[slot.id] <- list(NULL)  # preserves pool length
+    },
+    get.active.slots = function() {
+      which(!sapply(private$lineage.pool, is.null))
+    },
+    get.inactive.slots = function() {
+      which(sapply(private$lineage.pool, is.null))
+    },
+
     remove.pathogen = function(idx) {
       path <- private$pathogens[[idx]]
       private$pathogens[[idx]] <- NULL
@@ -136,7 +165,9 @@ Host <- R6Class(
     sampling.time = NULL,
     sampling.comp = NULL,
     unsampled = NULL,
-    pathogens = NULL
+    pathogens = NULL,
+    lineage.pool = NULL,
+    pool.size = NULL
   )
 )
 

R/Model.R

@@ -328,6 +328,19 @@ Model <- R6Class(
           }
           private$check.expression(params$pop.size, env)
           private$pop.sizes[[src]] <- params$pop.size
+        } else if (!is.null(params$coalescent.rate) &&
+                   params$coalescent.rate != "Inf") {
+          # coalescence is enabled for this compartment (finite rate) but
+          # pop.size was never explicitly set, so it's silently using the
+          # unconfigured default (100) baked into this package -- this
+          # default has no connection to the compartment's own `size:`
+          # field or any other yaml value, and coalescent.rate expressions
+          # that reference population size (or were chosen assuming a
+          # particular population size) may be silently mismatched with it.
+          warning(src, ": coalescent.rate is set but pop.size is not -- ",
+                  "using unconfigured default pop.size=100. If ",
+                  "coalescent.rate was chosen with a particular population ",
+                  "size in mind, set pop.size explicitly in the yaml.")
         }
       }  # end loop
 

R/Pathogen.R

@@ -41,6 +41,8 @@ Pathogen <- R6Class(
     # immutable attributes
     get.name = function() { private$name },
     get.end.time = function() { private$end.time },
+    get.slot.id = function() { private$slot.id },
+    set.slot.id = function(id) { private$slot.id <- id },
 
     # mutables
     get.start.time = function() { private$start.time },
@@ -78,6 +80,7 @@ Pathogen <- R6Class(
     end.time = NULL,
     parents = NULL,
     children = NULL,
-    breakpoint = NULL
+    breakpoint = NULL,
+    slot.id = NA
   )
 )

R/simARG.R

@@ -63,8 +63,11 @@ sim.arg <- function(outer, rho = 1e-4, seq.length = 9000L) {
         if (ev$type == "coalescent") {
           .do.coalescent(ev$host, inner, event.time, envir = env)
         } else {
+          host.obj <- active$get.host.by.name(ev$host)
+          p.size.expr <- mod$get.pop.size(host.obj$get.compartment())
+          p.size.val <- eval(parse(text = p.size.expr), envir = env)
           bp <- .do.recombination(ev$host, ev$pathogen, inner, event.time,
-                                  seq.length = seq.length)
+                                  seq.length = seq.length, p.size = p.size.val)
           breakpoints[[bp$child]] <- bp$position
         }
       }
@@ -133,6 +136,7 @@ sim.arg <- function(outer, rho = 1e-4, seq.length = 9000L) {
 
     # coalescence rate for this host (requires 2+ lineages)
     if (k >= 2) {
+
       expr <- mod$get.coalescent.rate(comp)
       rate <- eval(parse(text = expr), envir = envir)
       if (rate > 0) {
@@ -190,20 +194,58 @@ sim.arg <- function(outer, rho = 1e-4, seq.length = 9000L) {
 #' @keywords internal
 #' @noRd
 .do.recombination <- function(host.name, pathogen, inner, time,
-                              seq.length = 9000L) {
+                              seq.length = 9000L, p.size = NULL) {
   active <- inner$get.active()
   host   <- active$get.host.by.name(host.name)
 
+  # initialize this host's FIXED lineage pool (sample
+  # recombination parents from a fixed pool of p.size lineages, only
+  # activating a previously-inactive one when chosen, instead of always
+  # creating a brand-new lineage de novo. Total pool size never changes.)
+  if (!host$is.pool.initialized()) {
+    if (is.null(p.size)) {
+      stop(".do.recombination: p.size must be provided to initialize ",
+           "host's lineage pool on first use")
+    }
+    host$init.pool(p.size)
+  }
+
+  # ensure the recombining pathogen already occupies a pool slot -- if
+  # this is the first pool-tracked event involving it, assign one now
+  if (is.na(pathogen$get.slot.id())) {
+    free <- host$get.inactive.slots()
+    slot <- if (length(free) > 0) free[1] else 1
+    pathogen$set.slot.id(slot)
+    host$activate.slot(slot, pathogen)
+  }
+  own.slot <- pathogen$get.slot.id()
+
   # sample breakpoint uniformly across genome
   breakpoint <- sample.int(seq.length - 1L, 1L)
   pathogen$set.breakpoint(breakpoint)
-
-  # end the current lineage at this recombination event
   pathogen$set.start.time(time)
 
-  # create two parental lineages — left and right of breakpoint
-  parent.left  <- inner$new.pathogen(time)
-  parent.right <- inner$new.pathogen(time)
+  # LEFT parent: continues in the SAME slot as the child
+  parent.left <- inner$new.pathogen(time)
+  parent.left$set.slot.id(own.slot)
+  host$activate.slot(own.slot, parent.left)
+
+  # RIGHT parent: sample a slot from the fixed pool (excluding own slot)
+  pool.size <- host$get.pool.size()
+  other.slots <- setdiff(seq_len(pool.size), own.slot)
+
+  if (length(other.slots) == 0) {
+    parent.right <- parent.left
+  } else {
+    sampled.slot <- if (length(other.slots) == 1) other.slots else sample(other.slots, 1)
+    if (host$is.slot.active(sampled.slot)) {
+      parent.right <- host$get.slot.occupant(sampled.slot)
+    } else {
+      parent.right <- inner$new.pathogen(time)
+      parent.right$set.slot.id(sampled.slot)
+      host$activate.slot(sampled.slot, parent.right)
+    }
+  }
 
   # record parent-child relationships (recombination has two parents)
   parent.left$add.child(pathogen)
@@ -216,9 +258,12 @@ sim.arg <- function(outer, rho = 1e-4, seq.length = 9000L) {
   idx <- which(sapply(paths, function(p) p$get.name()) == pathogen$get.name())
   if (length(idx) == 1) host$remove.pathogen(idx)
   host$add.pathogen(parent.left)
-  host$add.pathogen(parent.right)
+  already.present <- any(sapply(host$get.pathogens(), function(p) {
+    p$get.name() == parent.right$get.name()
+  }))
+  if (!already.present) host$add.pathogen(parent.right)
 
-  # log the recombination event (breakpoint not stored in log — fixed schema)
+  # log the recombination event (breakpoint not stored in log -- fixed schema)
   event <- list(
     time = time, event = "recombination",
     from.comp = host$get.compartment(), to.comp = NA,

R/simInnerTree.R

@@ -202,6 +202,29 @@ sim.inner.tree <- function(outer) {
     expr <- inner$get.model()$get.pop.size(e$to.comp)
     p.size <- eval(parse(text=expr), envir=envir)
 
+    if (count > p.size) {
+      # count (n.active.lineages) is tracked ancestral segment/lineage
+      # objects, not necessarily distinct physical genomes -- once
+      # recombination is active, lineage count can legitimately exceed
+      # the population's census size, and the correct relationship
+      # between the two is not yet resolved (see issue tracker: rhyper()
+      # here implicitly assumes count occupies count distinct slots
+      # among p.size exchangeable individuals, which breaks once one
+      # genome can carry several ancestral segments). Fail loudly with
+      # a clear diagnostic rather than silently capping or crashing on
+      # an opaque NA, until the bottleneck/occupancy model is revisited.
+      stop(sprintf(
+        "Tracked lineage count (%d) exceeds pathogen population size (%d) ",
+        count, p.size),
+        "in host ", recipient$get.name(), " at time ", e$time, ". ",
+        "This means more ancestral lineages/segments are being tracked ",
+        "than the model's nominal population size anticipates (likely ",
+        "from recombination). The bottleneck sampling here assumes ",
+        "count occupies count distinct slots among p.size individuals, ",
+        "which is not valid once lineage count and physical individual ",
+        "count can diverge -- needs a proper occupancy/carrier model, ",
+        "not a silent cap.")
+    }
     n.transfer <- rhyper(1, count, p.size-count, b.size)
     if (n.transfer > 0) {
       for (i in 1:n.transfer) {
@@ -321,6 +344,27 @@ sim.inner.tree <- function(outer) {
     p2$set.start.time(time)
 
     anc <- inner$new.pathogen(time)  # sets end.time
+
+    # reconcile the lineage pool: the ancestor represents the same
+    # physical individual as whichever of p1/p2 already occupied a pool
+    # slot (from a prior recombination event). If both occupied slots,
+    # keep one for the ancestor and free the other -- two active
+    # lineages coalescing means one fewer active individual going
+    # forward. If neither occupied a slot, this coalescence never
+    # touched the pool, so the ancestor stays unassigned too.
+    p1.slot <- p1$get.slot.id()
+    p2.slot <- p2$get.slot.id()
+    if (!is.na(p1.slot)) {
+      anc$set.slot.id(p1.slot)
+      host$activate.slot(p1.slot, anc)
+      if (!is.na(p2.slot) && p2.slot != p1.slot) {
+        host$deactivate.slot(p2.slot)
+      }
+    } else if (!is.na(p2.slot)) {
+      anc$set.slot.id(p2.slot)
+      host$activate.slot(p2.slot, anc)
+    }
+
     host$add.pathogen(anc)
 
     # assign ancestral/descendant relations

tests/testthat/test_Superinfection.yaml

@@ -24,6 +24,7 @@ Compartments:
     size: 2
     bottleneck.size: 1
     coalescent.rate: 0.01
+    pop.size: 100    # was silently defaulting to this
   I_samp:
     infected: true
     size: 0

tests/testthat/test_simARG.R

@@ -207,3 +207,82 @@ test_that("resolve.arg produces genuinely divergent topology (hand-built positiv
   expect_true(is.monophyletic(phy2, c("A","C")))
   expect_false(is.monophyletic(phy2, c("A","B")))
 })
+test_that("Host lineage pool: basic activate/deactivate/query", {
+  h <- Host$new(name="H1", compartment="I")
+  expect_false(h$is.pool.initialized())
+  h$init.pool(5)
+  expect_true(h$is.pool.initialized())
+  expect_equal(h$get.pool.size(), 5)
+  expect_equal(length(h$get.inactive.slots()), 5)
+  expect_equal(length(h$get.active.slots()), 0)
+
+  p <- Pathogen$new(name="P1", end.time=1)
+  h$activate.slot(2, p)
+  expect_true(h$is.slot.active(2))
+  expect_equal(h$get.slot.occupant(2)$get.name(), "P1")
+  expect_equal(length(h$get.active.slots()), 1)
+  expect_equal(length(h$get.inactive.slots()), 4)
+
+  h$deactivate.slot(2)
+  expect_false(h$is.slot.active(2))
+  expect_equal(length(h$get.active.slots()), 0)
+  # pool size must not shrink after deactivation
+  expect_equal(h$get.pool.size(), 5)
+})
+
+test_that("Host lineage pool: init.pool is idempotent", {
+  h <- Host$new(name="H1", compartment="I")
+  h$init.pool(10)
+  h$init.pool(999)  # should be ignored, pool already initialized
+  expect_equal(h$get.pool.size(), 10)
+})
+
+test_that("sim.arg does not exceed pop.size at high rho (fixed lineage pool)", {
+  settings <- read_yaml("test_Superinfection.yaml")
+  settings$Parameters$sigma <- 0.05
+  mod <- Model$new(settings)
+  set.seed(33)
+  dyn <- tryCatch(sim.dynamics(mod, max.attempts=10), error=function(e) NULL)
+  if (is.null(dyn)) skip("could not build dynamics for this seed")
+  outer <- tryCatch(
+    withCallingHandlers(sim.outer.tree(dyn), warning=function(w) invokeRestart("muffleWarning")),
+    error=function(e) NULL)
+  if (is.null(outer)) skip("could not build outer tree for this seed")
+
+  # rho=5,7,10 previously exceeded pop.size and crashed/errored before
+  # the fixed lineage pool (Art's design: sample recombination parents
+  # from a fixed pool of p.size lineages instead of always creating a
+  # new lineage de novo). Now they should all succeed.
+  for (rho in c(5, 7, 10)) {
+    arg <- tryCatch(sim.arg(outer, rho=rho, seq.length=9000), error=function(e) e)
+    expect_false(inherits(arg, "error"),
+                 info=paste("rho =", rho, "should not error with fixed lineage pool"))
+  }
+})
+
+test_that("resolve.arg produces valid trees on top of the fixed lineage pool", {
+  settings <- read_yaml("test_Superinfection.yaml")
+  settings$Parameters$sigma <- 0.05
+  mod <- Model$new(settings)
+  set.seed(33)
+  dyn <- tryCatch(sim.dynamics(mod, max.attempts=10), error=function(e) NULL)
+  if (is.null(dyn)) skip("could not build dynamics for this seed")
+  outer <- tryCatch(
+    withCallingHandlers(sim.outer.tree(dyn), warning=function(w) invokeRestart("muffleWarning")),
+    error=function(e) NULL)
+  if (is.null(outer)) skip("could not build outer tree for this seed")
+
+  n.sampled <- outer$get.sampled()$count.type()
+  arg <- sim.arg(outer, rho=2, seq.length=9000)
+  res <- resolve.arg(arg, seq.length=9000)
+
+  valid <- sapply(res$local.trees, function(lt) {
+    phy <- lt$phylo
+    inherits(phy, "phylo") &&
+      length(phy$tip.label) == n.sampled &&
+      sum(duplicated(phy$tip.label)) == 0 &&
+      !any(is.na(phy$edge.length)) &&
+      !any(phy$edge.length < 0, na.rm=TRUE)
+  })
+  expect_true(all(valid))
+})

tests/testthat/test_simInnerTree.R

@@ -3,6 +3,7 @@ require(twt)
 # generate test fixtures
 settings <- yaml.load_file("test_SIR.yaml")
 settings$Compartments$I$coalescent.rate <- 1.0
+settings$Compartments$I$pop.size <- 100  # was silently defaulting to this
 mod <- Model$new(settings)
 set.seed(276)
 dynamics <- sim.dynamics(mod)

tests/testthat/test_superinfection.R

@@ -393,7 +393,7 @@ test_that("model: compartment I flagged infected=TRUE", {
   expect_true(mod$get.infected("I"))
 })
 
-test_that("model YAML: compartment I has pop.size=2 and bottleneck.size=1", {
+test_that("model YAML: compartment I has size=2 and bottleneck.size=1", {
   try_model(SUPERINF_INNER_PATH)
   cfg <- read_yaml(SUPERINF_INNER_PATH)
   expect_equal(cfg$Compartments$I$size, 2)