Let’s break this down step by step, focusing on how matrix indexing rules and data.table's column iteration interact here.

The Working Code: Integer Values as Indices #

First, let's look at the code that runs successfully:

a <- data.table(t(c(1, 2, 3, 4, 5)))
b <- matrix(data=1, nrow=10, ncol=5)
a[,lapply(.SD,function(x)(x*b[,x]))]

Here’s what’s happening:

• a is a 1-row, 5-column data.table where each column holds an integer value (1, 2, 3, 4, 5).
• When lapply(.SD, ...) runs, it iterates over each column of a. For each column, x is the single value in that column (since a only has one row).
• b[,x] uses x as a column index for matrix b. R allows integer values to index matrix columns directly—so x=1 grabs the 1st column of b, x=2 grabs the 2nd, and so on.
• Multiplying x (a single integer) by b[,x] (a 10-element vector) works because R recycles the single value to match the vector length, resulting in a valid 10-element vector per column. The data.table then assembles these vectors into new columns, so the code runs without errors.

The Failing Code: Non-Integer Values as Indices #

Now the code that breaks:

a <- data.table(t(c(0.1, 0.2, 0.3, 0.4, 0.5)))
b <- matrix(data=1, nrow=10, ncol=5)
a[,lapply(.SD,function(x)(x*b[,x]))]

The problem boils down to matrix indexing rules:

• a now holds non-integer (floating-point) values (0.1, 0.2, etc.) in its columns.
• When b[,x] runs with x=0.1, R tries to use this floating-point number as a column index for b. But matrices only accept integer indices (or logical indices/column names) to reference columns—there’s no such thing as the "0.1th column" of a matrix.
• R doesn’t automatically convert floating-point numbers to integers for matrix indexing here (unlike some other contexts), so this throws an invalid index error immediately.

Fixing the Failing Code #

If your goal is to use the values in a as weights (not column indices), you’d want to do element-wise multiplication instead of indexing:

# Multiply each value in a by the entire matrix b (recycles rows of a to match b's rows)
a[, lapply(.SD, function(x) x * b)]

If you intended to use the values in a as column indices (but they’re stored as floats), convert them to integers first:

a <- data.table(t(c(0.1, 0.2, 0.3, 0.4, 0.5)))
a <- a[, lapply(.SD, as.integer)] # Convert columns to integers
b <- matrix(data=1, nrow=10, ncol=5)
a[, lapply(.SD, function(x) x * b[,x])] # Now runs successfully

内容的提问来源于stack exchange，提问作者Z117