⚡ Bolt: Optimize ifelse calls in llcont.glm binomial family

.jules/bolt.md

@@ -15,3 +15,7 @@
 ## 2024-05-15 - [R Performance: ifelse Overhead]
 **Learning:** In R, ifelse evaluates both true and false branches entirely before subsetting, which is very inefficient for vector operations.
 **Action:** Optimize this by preallocating with res <- Y * 0 to preserve attributes and using vectorized subsetting like if any cond res subset <- ...
+
+## 2024-05-16 - Safe ifelse Optimization with Mathematical Operations in R
+**Learning:** While replacing `ifelse(n == 0, 0, y/n)` with pre-allocation (`y_res <- y * 0; y_res[cond] <- ...`) works, it can be mathematically simplified to `y_res <- y / n; y_res[n == 0 & !is.na(n)] <- 0`. The key learning is that subsetting *must* explicitly check `!is.na(n)` to avoid throwing a fatal "NAs are not allowed in subscripted assignments" error if `n` contains missing values, making the code safer and more readable.
+**Action:** When replacing `ifelse` via mathematical shortcuts, always combine the logical condition with `& !is.na(...)` to preserve NA handling robustly.

R/llcont.R

@@ -53,12 +53,16 @@ llcont.glm <- function(x, ...){
              if(is.matrix(y)) {
                ## Bolt: replaced apply(..., 1, sum) with optimized rowSums() for performance
                n <- rowSums(y)
-               y <- ifelse(n == 0, 0, y[, 1]/n)
+               ## Bolt: optimized ifelse for performance
+               y <- y[, 1] / n
+               y[n == 0 & !is.na(n)] <- 0
              } else {
                n <- rep.int(1, length(y))
              }
              m <- if (any(n > 1)) n else wt
-             wt <- ifelse(m > 0, (wt/m), 0)
+             ## Bolt: optimized ifelse for performance
+             wt <- wt / m
+             wt[m <= 0 & !is.na(m)] <- 0
              dbinom(round(m * y), round(m), mpreds, log = TRUE) * wt
            },
            quasibinomial = {