3 Summing over indices
Summing several things in a single expression occurs frequently in mathematics, and in particular in applications to physics and chemistry. The general syntax for a summation over an index is: \[ \sum_{i \in I} f(i), \] where \(I\) is some countable set of indices, \(i\) is the index of summation, and \(f(i)\) is the function/quantity being summed. The set \(I\) can be finite or infinite, and the index \(i\) can be a number, a vector, or even a more complicated object.
The most basic and common case is a sum over a finite or infinite set of integers, which is usually written as \[ \sum_{i=a}^{b} f(i) \] where \(i\) is the index of summation, \(a\) is the lower limit of the index \(i\), \(b\) is the upper limit, and \(f(i)\) is the function/quantity being summed.
Here are some concrete examples:
A sum over a finite number of terms: \[ \sum_{i=1}^{5} i = 1 + 2 + 3 + 4 + 5 \]
An infinite series: \[ \sum_{n=0}^{\infty} \frac{1}{2^n} = 1 + \frac{1}{2} + \frac{1}{4} + \frac{1}{8} + \ldots \]
A series expansion. In this example, the exponential function is expressed as a Taylor series, summing over all non-negative integers: \[ e^x = \sum_{n=0}^{\infty} \frac{x^n}{n!} = 1 + x + \frac{x^2}{2} + \frac{x^3}{6} + \frac{x^4}{24} + \ldots \]
Vector notation. For example, a vector \(\mathbf{r}\in\mathbb{R}^3\) can be expressed as a sum over its components: \[ \sum_{i=1}^3 r_i \mathbf{e}_i = \mathbf{r} \]
A matrix-vector product expressed as a sum over indices: The \(i\)-th component of the product of a matrix \(A\) and a vector \(\mathbf{x}\) is given by the sum over the products of the corresponding elements of the matrix and vector: \[ (A \mathbf{x})_i = \sum_{j=1}^n A_{ij} x_j \]
Manipulating sums
The index in a sum is called a summation index or dummy index. Its name has no significance. For example,
\[ \sum_{i=1}^n a_i = \sum_{j=1}^n a_j. \]
The index may therefore be renamed, as long as it is renamed consistently within the sum.
Summation is linear. Sums can be split or combined,
\[ \sum_{i=1}^n (a_i+b_i) = \sum_{i=1}^n a_i + \sum_{i=1}^n b_i, \]
and factors that do not depend on the summation index can be taken outside the sum,
\[ \sum_{i=1}^n c a_i = c\sum_{i=1}^n a_i, \]
where it is important that \(c\) is independent of \(i\).
Expressions may contain both summed and unsummed indices. An index that is not summed over is called a free index. For example, in the matrix-vector product
\[ y_i = \sum_{j=1}^n A_{ij}x_j, \]
\(j\) is a dummy index, while \(i\) is a free index. The equation therefore represents one equation for each possible value of \(i\).
Special symbols used in summation notation
The Kronecker delta \(\delta_{ij}\) is a special symbol that is often used in summation notation. It is defined as \[ \delta_{ij} = \begin{cases} 1 & \text{if } i=j, \\ 0 & \text{if } i\neq j. \end{cases} \] Sometimes, if confusion can arise one inserts a comma between the two indices, writing \(\delta_{i,j}\) instead of \(\delta_{ij}\).
The Kronecker delta forms the matrix elements of the identity matrix.
The Levi-Civita symbol \(\epsilon_{ijk}\) is another special symbol that is often used in summation notation in the three-dimensional case, i.e., where \(i,j,k \in \{1,2,3\}\). It is defined as \[ \epsilon_{ijk} = \begin{cases} 1 & \text{if } (i,j,k) \text{ is an even permutation of } (1,2,3), \\ -1 & \text{if } (i,j,k) \text{ is an odd permutation of } (1,2,3), \\ 0 & \text{if any two indices are equal.} \end{cases} \]
Exercises: Basic summation
These exercises are at a very basic level. If you are already comfortable with summation notation, you may skip them.
Exercise 3.1 (Very basic sums) Expand the following sums and calculate their values:
\(\displaystyle \sum_{i=1}^{4} i\)
\(\displaystyle \sum_{i=1}^{4} i^2\)
\(\displaystyle \sum_{k=0}^{3} 2^k\)
Exercise 3.2 (Simplifying as a sum) Write each expression using summation notation:
\(x_1+x_2+x_3+x_4+x_5\)
\(1+q+q^2+q^3+q^4\)
\(a_1b_1+a_2b_2+a_3b_3\)
Exercise 3.3 (Estimating \(\sqrt{e}\) via a Taylor series) Use the first four terms of the series for \(e^x\) to estimate \(\sqrt{e}\): \[ \sqrt{e} = e^{0.5} \approx \sum_{n=0}^{3}\frac{(0.5)^n}{n!}. \] Write out the four terms and add them. The estimate is not especially accurate; the point is to see how the index \(n\) controls both the power and the factorial.
Exercise 3.4 (Reconstruct a vector) The components of a vector are \(r_1=2\), \(r_2=-1\), and \(r_3=4\). Let \(\mathbf{e}_i\) be the standard basis in \(\mathbb{R}^3\).
Write out \(\displaystyle \sum_{i=1}^{3}r_i\mathbf{e}_i\) term by term.
Gather the result as a column vector
Exercise 3.5 (Summing in a matrix–vector product) Let \[ A = \begin{bmatrix} 1 & 2 & 0 \\ -1 & 0 & 3 \end{bmatrix}, \qquad \mathbf{x} = \begin{bmatrix} 2 \\ 1 \\ -1 \end{bmatrix}. \]
Use \[ (A\mathbf{x})_i=\sum_{j=1}^{3}A_{ij}x_j \] to write out and calculate \((A\mathbf{x})_1\) and \((A\mathbf{x})_2\). Collect the two answers in a column vector.
Exercise 3.6 (The summation index is a dummy index) Consider \[ S=\sum_{i=1}^{4} i^2. \]
Write out the sum explicitly and calculate its value.
Write out and calculate \[ \sum_{j=1}^{4} j^2. \]
Explain why \[ \sum_{i=1}^{4} i^2 = \sum_{j=1}^{4} j^2. \]
Rewrite \[ \sum_{k=1}^{5} a_k b_k \] using \(p\) as the summation index.
Exercise 3.7 (Linearity of sums) Let \(a_i\) and \(b_i\), \(i=1,2,3\) be two sets of numbers and let \(c\) be a constant, i.e., just a number.
Write out both sides of \[ \sum_{i=1}^{3}(a_i+b_i) = \sum_{i=1}^{3}a_i+\sum_{i=1}^{3}b_i \] and verify that they are equal.
Write out both sides of \[ \sum_{i=1}^{3}c a_i = c\sum_{i=1}^{3}a_i \] and verify that they are equal.
Use these properties to rewrite \[ \sum_{i=1}^{3}(2a_i-3b_i) \] as two separate sums (i.e., each has a single summation sign).
Exercise 3.8 (When something does not depend on the summation index) Let \(x\) be a number independent of the summation index \(i\).
Simplify \[ \sum_{i=1}^{5} x. \]
Simplify \[ \sum_{i=1}^{5} i x. \]
Explain why \(x\) can be taken outside the sum in \[ \sum_{i=1}^{n} i x = x\sum_{i=1}^{n} i. \]
Exercise 3.9 (Free and summed indices) Consider the following expression for the \(i\)-th component of a vector \(\mathbf{y}\), being the product of a matrix \(A\) and a vector \(\mathbf{x}\): \[ y_i=\sum_{j=1}^{3}A_{ij}x_j. \]
Which index is summed over?
Which index is not summed over?
For \(i=2\), write the right-hand side explicitly.
If \(A\) has four rows and three columns, how many components does \(\mathbf{y}\) have?
Explain why replacing \(j\) by \(k\), \[ y_i=\sum_{k=1}^{3}A_{ik}x_k, \] does not change the equation.
Exercise 3.10 (Shifting the summation index) Consider \[ S=\sum_{i=1}^{4} a_i. \]
Write out \(S\) explicitly.
Show that the same sum can be written as \[ S=\sum_{j=0}^{3}a_{j+1} \] by writing out the terms.
Rewrite \[ \sum_{i=2}^{5}a_i \] as a sum whose index starts at \(j=0\).
Rewrite \[ \sum_{i=0}^{n-1}a_{i+1} \] as a sum with index \(j\) running from \(1\) to \(n\).
Exercise 3.11 (The Kronecker delta: basic exercise) The Kronecker delta is defined in the introduction to this chapter. It is a special symbol that is often used in summation notation.
Evaluate \(\delta_{11}\), \(\delta_{12}\), \(\delta_{23}\), and \(\delta_{33}\).
Let \(a_1=2\), \(a_2=5\), and \(a_3=7\). Write out all terms in the sum
\[ \sum_{j=1}^3 \delta_{2,j}a_j \]
and evaluate it.
Repeat (b) for
\[ \sum_{j=1}^3 \delta_{1j}a_j \qquad\text{and}\qquad \sum_{j=1}^3 \delta_{3j}a_j. \]
Based on your results, simplify the general expression
\[ \sum_{j=1}^n \delta_{ij}a_j. \]
In the expression in (d), which index is a dummy index and which is a free index? What values can the free index take?
Exercise 3.12 (The Levi–Civita symbol: basic exercise) The Levi–Civita symbol \(\varepsilon_{ijk}\) is defined in the introduction to this chapter.
Evaluate \(\varepsilon_{123}\), \(\varepsilon_{132}\), \(\varepsilon_{231}\), \(\varepsilon_{321}\), and \(\varepsilon_{223}\).
Find all triples \((i,j,k)\) for which \(\varepsilon_{ijk}=+1\).
Find all triples \((i,j,k)\) for which \(\varepsilon_{ijk}=-1\).
What happens to \(\varepsilon_{ijk}\) if two indices are interchanged? Illustrate with two examples.
Let \(a_1\), \(a_2\), and \(a_3\) be the components of a vector. Write out the sum
\[ \sum_{j=1}^3\sum_{k=1}^3 \varepsilon_{1jk}a_jb_k \]
and remove all terms that vanish because of the Levi–Civita symbol.
Exercise 3.13 (A molecular orbital in an atomic-orbital basis) In quantum chemistry, a molecular orbital \(\psi_p(\mathbf{r})\) is commonly expanded in a finite set of atomic-orbital basis functions \(\chi_\mu(\mathbf{r})\),
\[ \psi_p(\mathbf{r})=\sum_{\mu=1}^{K} C_{\mu p}\chi_\mu(\mathbf{r}). \]
Here, \(\mu\) labels the basis functions, while \(p\) labels the molecular orbital.
Consider three basis functions and the coefficient matrix
\[ C= \begin{bmatrix} 0.7 & 0.5\\ 0.7 & -0.5\\ 0.1 & 0.7 \end{bmatrix}. \]
In the expression for \(\psi_p(\mathbf{r})\), identify the dummy index and the free index.
Write \(\psi_1(\mathbf{r})\) explicitly as a linear combination of \(\chi_1(\mathbf{r})\), \(\chi_2(\mathbf{r})\), and \(\chi_3(\mathbf{r})\), using the matrix elements of \(C\).
Do the same for \(\psi_2(\mathbf{r})\).
Suppose that at a particular point \(\mathbf{r}_0\),
\[ \chi_1(\mathbf{r}_0)=1.0,\qquad \chi_2(\mathbf{r}_0)=0.5,\qquad \chi_3(\mathbf{r}_0)=0.2. \]
Calculate \(\psi_1(\mathbf{r}_0)\) and \(\psi_2(\mathbf{r}_0)\).
- Explain why the summation index \(\mu\) could be replaced by another symbol, such as \(i\), without changing the meaning of the expression.