Skip to main content

Multi-Index Summation: A Second Look Through Harmonic Numbers

Mathematics · Harmonic Numbers · Double Sums

Finding a Closed Form for \(S_n\)

Target sum

We add \(\frac{1}{k-j}\) over all pairs with \(1 \le j < k \le n\).

Approach 1

Rewrite the sum as a sum of harmonic numbers.

Approach 2

Substitute \(k \to k+j\) and collapse the inner sum directly.

Definition

We want a simple formula for \[ S_n=\sum_{1 \le j < k \le n}\frac{1}{k-j}. \]

This means: for every pair of whole numbers \(j\) and \(k\) between 1 and \(n\) with \(j

Main observation The summand depends only on the difference \(k-j\), not on \(j\) and \(k\) separately.
Warm-up

Worked Examples

  • \(S_1=0\), because there are no valid pairs.
  • \(S_2=1\), from the single pair \((1,2)\).
  • \(S_3=\frac{1}{2-1}+\frac{1}{3-1}+\frac{1}{3-2}=1+\frac12+1=\frac52.\)
Background

A Quick Reminder: Harmonic Numbers

The \(n\)th harmonic number is

\[ H_n=\sum_{1 \le k \le n}\frac{1}{k}=1+\frac12+\frac13+\cdots+\frac1n. \]

For example, \[ H_1=1,\qquad H_2=\frac32,\qquad H_3=\frac{11}{6}, \] and by convention \(H_0=0\).

Approach 1

Rewrite \(S_n\) as a Sum of Harmonic Numbers

Step 1: Separate the two indices

Split the sum into an outer sum over \(k\) and an inner sum over \(j\):

\[ S_n=\sum_{1 \le k \le n}\sum_{1 \le j < k}\frac{1}{k-j}. \]

Step 2: Re-index the inner sum

Replace the inner index by \(j \mapsto k-j\). Then the denominator becomes simply \(j\):

\[ S_n=\sum_{1 \le k \le n}\sum_{1 \le j \le k-1}\frac{1}{j}. \]

Step 3: Recognize a harmonic number

The inner sum is exactly \(H_{k-1}\), so:

\[ S_n=\sum_{1 \le k \le n} H_{k-1}. \]

Step 4: Tidy the index

Shift the index to get the cleaner form:

\[ \boxed{S_n=\sum_{0 \le k < n} H_k.} \]
Approach 2

Find the Explicit Closed Form

Step 1: Substitute \(k \to k+j\)

Start again from the original sum and replace \(k\) by \(k+j\). This turns the denominator into \(k\):

\[ S_n=\sum_{\substack{j \ge 1,\; k \ge 1 \\ j+k \le n}} \frac{1}{k}. \]

Step 2: Sum over \(j\) first

For fixed \(k\), the index \(j\) ranges from 1 to \(n-k\). So:

\[ S_n=\sum_{1 \le k \le n}\sum_{1 \le j \le n-k}\frac{1}{k}. \]

Step 3: Collapse the inner sum

The inner sum adds \(\frac1k\) exactly \(n-k\) times, so:

\[ S_n=\sum_{1 \le k \le n}\frac{n-k}{k}. \]

Step 4: Split the summand

\[ \frac{n-k}{k}=\frac{n}{k}-1, \]

giving:

\[ S_n=\sum_{1 \le k \le n}\frac{n}{k}-\sum_{1 \le k \le n}1. \]

Step 5: Factor and evaluate

Pull out the constant \(n\) and count the second sum:

\[ S_n=n\left(\sum_{1 \le k \le n}\frac{1}{k}\right)-n=nH_n-n. \]
Closed form \[ \boxed{S_n=nH_n-n.} \]
Identity

Combining the Two Results

Approach 1 gave \[ S_n=\sum_{0 \le k < n} H_k, \] while Approach 2 gave \[ S_n=nH_n-n. \]

Since both equal the same quantity, we get the identity:

\[ \sum_{0 \le k < n} H_k = nH_n-n. \]

So the sum of the first \(n\) harmonic numbers has a closed form in terms of \(H_n\).

Interpretation

What Did We Just Do?

The algebraic view

The key move in Approach 2 was replacing \(k\) by \(k+j\). Whenever a denominator involves a difference like \(k-j\), this kind of substitution is often useful because it isolates one variable and makes the inner sum trivial.

The geometric view

For \(n=4\), the valid pairs \((j,k)\) form a triangular region:

\(k=1\) \(k=2\) \(k=3\) \(k=4\)
\(j=1\) \(\frac11\) \(\frac12\) \(\frac13\)
\(j=2\) \(\frac11\) \(\frac12\)
\(j=3\) \(\frac11\)
\(j=4\)

Summing by rows gives \(H_3+H_2+H_1\). Summing by columns gives the same total. Summing by diagonals gives \[ \frac{3}{1}+\frac{2}{2}+\frac{1}{3}, \] because each diagonal has a constant value and a decreasing number of entries.

All three views count the same triangular array. The only thing that changes is the way the terms are grouped.

FAQ

Frequently Asked Questions

These are the practical questions students usually have when first seeing this harmonic-number double sum.

Why does the sum start at \(S_1=0\)?

Because there are no pairs \((j,k)\) with \(1 \le j < k \le 1\). The summation region is empty.

Why is re-indexing helpful in Approach 1?

Because it turns the denominator \(k-j\) into a plain index, which makes the inner sum recognizable as a harmonic number.

Why does the substitution \(k \to k+j\) help in Approach 2?

Because the denominator becomes just \(k\), and then the inner sum simply counts how many times each reciprocal appears.

Why do rows, columns, and diagonals all give the same answer?

Because they are just different ways of grouping the same set of terms in the same triangular region.

What is the simplest thing to remember from this problem?

The sum depends only on the difference \(k-j\), and that makes diagonal regrouping and harmonic-number structure natural.

Raell Dottin

Comments