From 5ba1d7aaf1a5d088c1eeba61d4474fae31c8d1ec Mon Sep 17 00:00:00 2001
From: seonghobae <8172694+seonghobae@users.noreply.github.com>
Date: Sat, 29 Aug 2026 04:22:45 +0000
Subject: [PATCH 1/3] Optimize binomial log-likelihood calculation by removing
 ifelse

---
 .jules/bolt.md |  3 +++
 R/llcont.R     | 10 ++++++++--
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/.jules/bolt.md b/.jules/bolt.md
index f658475..7b4cc2c 100644
--- a/.jules/bolt.md
+++ b/.jules/bolt.md
@@ -15,3 +15,6 @@
 ## 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 <- ...
+## 2026-08-11 - Fast Vectorized Subsetting over `ifelse()`
+**Learning:** `ifelse(cond, true_branch, false_branch)` evaluates both branches entirely before subsetting, which creates significant overhead, especially in hot loops or likelihood calculations. Replacing it with vectorized subsetting (e.g. `res <- true_branch; res[!cond] <- false_branch`) is much faster and retains the same NA handling as `ifelse()`. Using `which()` for indices can alter NA behavior (since `which()` drops NAs) and must be carefully evaluated, but simple vectorized assignment is safe and highly performant.
+**Action:** Replace `ifelse()` with vectorized assignment (e.g. `res <- a/b; res[b == 0] <- 0`) when optimizing hot loops or performance-critical likelihood paths.
diff --git a/R/llcont.R b/R/llcont.R
index d8e496a..7cc77d6 100644
--- a/R/llcont.R
+++ b/R/llcont.R
@@ -53,12 +53,18 @@ 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: replaced ifelse with vectorized subsetting for performance
+               y_new <- y[, 1]/n
+               y_new[n == 0] <- 0
+               y <- y_new
              } else {
                n <- rep.int(1, length(y))
              }
              m <- if (any(n > 1)) n else wt
-             wt <- ifelse(m > 0, (wt/m), 0)
+             ## Bolt: replaced ifelse with vectorized subsetting for performance
+             wt_new <- wt/m
+             wt_new[m <= 0] <- 0
+             wt <- wt_new
              dbinom(round(m * y), round(m), mpreds, log = TRUE) * wt
            },
            quasibinomial = {

From c49c30eda53786fda96af3aafba07635b4332041 Mon Sep 17 00:00:00 2001
From: seonghobae <8172694+seonghobae@users.noreply.github.com>
Date: Sat, 29 Aug 2026 04:38:43 +0000
Subject: [PATCH 2/3] Optimize binomial log-likelihood calculation by removing
 ifelse


From e116e826b8b4da3d73879f458cdf251957684bcf Mon Sep 17 00:00:00 2001
From: seonghobae <8172694+seonghobae@users.noreply.github.com>
Date: Sat, 29 Aug 2026 04:44:52 +0000
Subject: [PATCH 3/3] Optimize binomial log-likelihood calculation by removing
 ifelse