Fungrim entry: 47b181

\left|\int_{-1}^{1} f(t) \, dt - \sum_{k=1}^{n} w_{n,k} f\!\left(x_{n,k}\right)\right| \le \frac{64 M}{15 \left(1 - {\rho}^{-2}\right) {\rho}^{2 n}}\; \text{ where } M = \mathop{\operatorname{sup}}\limits_{t \in \mathcal{E}_{\rho}} \left|f(t)\right|

Assumptions:

n \in \mathbb{Z}_{\ge 1} \;\mathbin{\operatorname{and}}\; \rho \in \mathbb{R} \;\mathbin{\operatorname{and}}\; \rho > 1 \;\mathbin{\operatorname{and}}\; f(z) \text{ is holomorphic on } z \in \operatorname{InteriorClosure}\!\left(\mathcal{E}_{\rho}\right)

References:

L. N. Trefethen, Is Gauss Quadrature Better than Clenshaw-Curtis? SIAM Rev., 50(1), 67-87. DOI:10.1137/060659831

TeX:

\left|\int_{-1}^{1} f(t) \, dt - \sum_{k=1}^{n} w_{n,k} f\!\left(x_{n,k}\right)\right| \le \frac{64 M}{15 \left(1 - {\rho}^{-2}\right) {\rho}^{2 n}}\; \text{ where } M = \mathop{\operatorname{sup}}\limits_{t \in \mathcal{E}_{\rho}} \left|f(t)\right|

n \in \mathbb{Z}_{\ge 1} \;\mathbin{\operatorname{and}}\; \rho \in \mathbb{R} \;\mathbin{\operatorname{and}}\; \rho > 1 \;\mathbin{\operatorname{and}}\; f(z) \text{ is holomorphic on } z \in \operatorname{InteriorClosure}\!\left(\mathcal{E}_{\rho}\right)

Definitions:

Fungrim symbol	Notation	Short description
`Abs`	$\left\|z\right\|$	Absolute value
`Integral`	$\int_{a}^{b} f(x) \, dx$	Integral
`Sum`	$\sum_{n} f(n)$	Sum
`GaussLegendreWeight`	$w_{n,k}$	Gauss-Legendre quadrature weight
`LegendrePolynomialZero`	$x_{n,k}$	Legendre polynomial zero
`Pow`	${a}^{b}$	Power
`Supremum`	$\mathop{\operatorname{sup}}\limits_{x \in S} f(x)$	Supremum of a set or function
`BernsteinEllipse`	$\mathcal{E}_{\rho}$	Bernstein ellipse with foci -1,+1 and semi-axis sum rho
`ZZGreaterEqual`	$\mathbb{Z}_{\ge n}$	Integers greater than or equal to n
`RR`	$\mathbb{R}$	Real numbers
`IsHolomorphic`	$f(z) \text{ is holomorphic at } z = c$	Holomorphic predicate

Source code for this entry:

Entry(ID("47b181"),
    Formula(Where(LessEqual(Abs(Sub(Integral(f(t), For(t, -1, 1)), Sum(Mul(GaussLegendreWeight(n, k), f(LegendrePolynomialZero(n, k))), For(k, 1, n)))), Div(Mul(64, M), Mul(Mul(15, Sub(1, Pow(rho, -2))), Pow(rho, Mul(2, n))))), Equal(M, Supremum(Abs(f(t)), ForElement(t, BernsteinEllipse(rho)))))),
    Variables(f, n, rho),
    Assumptions(And(Element(n, ZZGreaterEqual(1)), Element(rho, RR), Greater(rho, 1), IsHolomorphic(f(z), ForElement(z, InteriorClosure(BernsteinEllipse(rho)))))),
    References("L. N. Trefethen, Is Gauss Quadrature Better than Clenshaw-Curtis? SIAM Rev., 50(1), 67-87. DOI:10.1137/060659831"))

Fungrim entry: 47b181

Topics using this entry