Newton's Identities

Theorem

Let $X$ be a set of $n$ numbers $\set {x_1, x_2, \ldots, x_n}$.

Define:

 $\displaystyle {\mathbf S}_m$ $=$ $\displaystyle \set { \paren {j_1,\ldots,j_m} : 1 \le j_1 \lt \cdots \lt j_m \le n}$ $1 \le m \le n$ $\displaystyle \map {e_m} {X}$ $=$ $\displaystyle \begin{cases} 1 & m = 0\\ \displaystyle \sum_{ {\mathbf S}_m } x_{j_1} \cdots x_{j_m} & 1 \leq m \leq n \\ 0 & m \gt n \\ \end{cases}$ elementary symmetric function $\displaystyle \map {p_k} X$ $=$ $\displaystyle \begin{cases} \displaystyle n & k = 0 \\ \displaystyle \sum_{i \mathop = 1}^n x_i^k & k \ge 1 \\ \end{cases}$ power sums

Then Newton's Identities are:

 $(1):\quad$ $\displaystyle k \, \map {e_k} X$ $=$ $\displaystyle \displaystyle \sum_{i \mathop = 1}^k \paren {-1}^{i-1} \map {e_{k-i} } X \map {p_i} X$ for $1 \leq k \leq n$ $(2):\quad$ $\displaystyle 0$ $=$ $\displaystyle \displaystyle \sum_{i \mathop = k-n}^k \paren {-1}^{i-1} \map {e_{k-i} } X \map {p_i} X$ for $1 \leq n \lt k$

Proof 1

Outline

The proof is divided into three cases: $k < n$, $k=n$ and $k>n$.

The tools are Viete's Formulas, symmetric function recursion, telescoping sums and homogeneous functions of degree $k$.

Proof 2

Outline

Calculus is used to prove identities (1) and (2) in a single effort.

The tools are Viete's Formulas, the calculus derivative of powers $x^n$ and logarithm $\ln \vert x \vert$, Maclaurin series expansion coefficients, mathematical induction, and Leibniz's Rule: One Variable.

Lemma 1

 $\displaystyle \prod_{r \mathop = 1}^n \paren { 1 + x_r z }$ $=$ $\displaystyle \sum_{m \mathop = 0}^n {\map {e_m} X} z^m$

Proof of Lemma 1

Begin with:

 $(11):\quad$ $\displaystyle \displaystyle \prod_{r \mathop = 1}^n \paren { x - x_r }$ $=$ $\displaystyle \sum_{i \mathop = 0}^n \paren {-1}^{n-i} {\map {e_{n-i} } X} x^i$ Viete's Formulas

Change variables in (11): $x = -1/z$.

$\Box$

Lemma 2

Denote by $D^k \map f z$ the $k$th calculus derivative of $\map f z$.

Let:

 $\displaystyle \map G z$ $=$ $\displaystyle \displaystyle \prod_{k \mathop = 1}^n \paren {1 + x_k z}$ as in Lemma 1 $\displaystyle \map F z$ $=$ $\displaystyle \dfrac { \map {DQ} z }{\map Q z}$ the calculus derivative of $\ln \vert { \map Q z } \vert$

Then:

 $(12):\quad$ $\displaystyle \dfrac { D^m \map G {0} } { m! }$ $=$ $\displaystyle \map { e_m } X$ $(13):\quad$ $\displaystyle \dfrac { D^m \map F {0} } { m! }$ $=$ $\displaystyle \paren {-1}^m \map {p_{m+1} } X$

Proof of Lemma 2

 $\displaystyle \map G z$ $=$ $\displaystyle \sum_{m \mathop = 0}^n {\map {e_m} X} z^m$ by Lemma 1

Then identity (12) holds by Maclaurin series expansion applied to polynomial $G$.

Identity (13) will be proved after mathematical induction establishes (14) infra.

Let $\map {\mathbf P} m$ be the statement:

 $(14):\quad$ $\displaystyle D^m \map F z$ $=$ $\displaystyle \sum_{i \mathop = 1}^n \dfrac{ m!\paren {-1}^m x_i^{m+1} }{ \paren { 1 + x_i z }^{m+1} }$ for $m \ge 0$

Basis for the induction: $m=0$

By calculus and the definition of $G$:

 $\displaystyle \map F z$ $=$ $\displaystyle \dfrac { D \map Q z}{\map Q z}$ $\displaystyle \leadsto \ \$ $\displaystyle$ $=$ $\displaystyle \displaystyle \sum_{i \mathop = 1}^n \dfrac { x_i }{ 1 + x_i z }$

Then $\map {\mathbf P} 0$ is true.

Induction step $\map {\mathbf P} m$ implies $\map {\mathbf P} {m+1}$:

 $\displaystyle D^{m+1} \map F z$ $=$ $\displaystyle D \paren {\sum_{i \mathop = 1}^n \dfrac{ m!\paren {-1}^m x_i^{m+1} }{ \paren { 1 + x_i z }^{m+1} } }$ by induction hypothesis $\map {\mathbf P} m$ $\displaystyle$ $=$ $\displaystyle \sum_{i \mathop = 1}^n \dfrac{ m! \paren {-1}^m x_i^{m+1} \, \paren {-1} \paren {m+1} \, x_i}{ \paren { 1 + x_i z }^{m+2} }$ by calculus power rule $\dfrac { \d u^{-n} }{ \d z } = \paren { -n } u^{-n-1} \dfrac { \d u }{ \d z }$

Simplify to prove $\map {\mathbf P} {m+1}$ is true.

The induction is complete.

To prove equation (13), first let $z=0$ in equation (14).

Divide by $m!$ to isolate $\map {p_{m+1} } X$, which proves (13).

$\Box$

Lemma 3

 $(15):\quad$ $\displaystyle \paren {m+1} \map {e_{m+1} } X$ $=$ $\displaystyle \displaystyle \sum_{r \mathop = 0}^m \paren {-1}^r { \map {e_{m-r} } X } { \map {p_{r+1} } X }$ for $m \geq 0$

Proof of Lemma 3

Begin with $D \map G z = {\map F z} {\map G z}$ and differentiate $m$ times on variable $z$:

 $\displaystyle D^{m+1} \map G z$ $=$ $\displaystyle \displaystyle \sum_{r \mathop = 0}^m {\dbinom {m} {r} } D^r {\map F z} \, D^{m-r} {\map G z}$ Leibniz's Rule/One Variable $\displaystyle D^{m+1} {\map G 0}$ $=$ $\displaystyle \displaystyle \sum_{r \mathop = 0}^m \dbinom {m} {r} r! \, \paren {-1}^r {\map { p_{r+1} } X} \, \paren {m-r}! \, {\map {e_{m-r} } X}$ Evaluate at $z = 0$ and use equations (12) and (13) in Lemma 2. $\displaystyle \paren {m+1} {\map {e_m} X}$ $=$ $\displaystyle \displaystyle \sum_{r \mathop = 0}^m \paren {-1}^r {\map {e_{m-r} } X} \, {\map {p_{r+1} } X}$ Use (12), then collect factorials and simplify.

$\Box$

Proof of the Theorem

Change indices via equations $m+1=k$, $r+1=i$.

The summation is from $i=0+1$ to $i=m+1$, which gives range $i=1$ to $k$.

Subscript $m-r$ equals $k-1-i+1$, which simplifies to $k-i$.

Then:

 $(16):\quad$ $\displaystyle k \, { \map {e_{k} } X }$ $=$ $\displaystyle \displaystyle \sum_{i \mathop = 1}^{k} \paren {-1}^{i-1} { \map {e_{k-i} } X } { \map {p_{i} } X }$ for $k \geq 1$, which is equation (1)

To prove (2), assume $k > n \ge 1$ and $X = \set {x_1,\ldots,x_n}$.

Equation (16) implies:

 $(117):\quad$ $\displaystyle 0$ $=$ $\displaystyle k \, { \map {e_{k} } X }$ because $\map {e_{j} } X = 0$ for $j=n+1,\ldots,k$. $\displaystyle \leadsto \ \$ $\displaystyle 0$ $=$ $\displaystyle \sum_{i \mathop = 1}^{k} \paren {-1}^{i-1} { \map {e_{k-i} } X } { \map {p_{i} } X }$ by (16) for $k \geq 1$ $\displaystyle \leadsto \ \$ $\displaystyle 0$ $=$ $\displaystyle \sum_{i \mathop = k-n}^{k} \paren {-1}^{i-1} { \map {e_{k-i} } X } { \map {p_{i} } X }$ because $\map {e_{k-i} } X = 0$ when $n+1 \le k-i \le k$

Then (2) holds.

$\blacksquare$

Source of Name

This entry was named for Isaac Newton.

Historical Note

Isaac Newton published in Arithmetica universalis (1707) a generalization of the $n \le 4$ formulas of A. Girard (1629), without proof. Formulas (1)-(2) make it possible to recursively solve for the symmetric functions $\set {e_k}$ in terms of power sums $\set {p_k}$ (Knuth 1997, p 94) and conversely. A history of Girard's work is in Funkhouser (1930). Various proofs of (1)-(2) exist: Baker (1959), Eidswick (1968), Mead (1992), Kalman (2000), Tignol (2001). Boklan (2018) reported calculus differentiation recursions to directly generate Girard's identities.

Also named as

Newton-Girard Identities, Newton-Girard Formulas.