∫₀¹ e^(−x²) dx
- Function f(x)
- exp(-x^2)
- Lower limit a
- 0
- Upper limit b
- 1
- Integral
- 0.746824132812
Checked against: (√π/2)·erf(1): Python decimal, erf by its Maclaurin series at 70 digits (hp.py)
Definite integral calculator
The definite integral of any function f(x) between two limits, by adaptive Gauss–Kronrod quadrature to about 20 significant digits, with the area shaded.
Integral
Integral: 0.746824132812
- Estimated error
- 1.1 × 10⁻²¹
- Function evaluations
- 45
- Subintervals used
- 2
The signed area between f(x) and the x-axis from 0 to 1 is 0.7468241328; area below the axis counts as negative. It took 45 evaluations over 2 subintervals.
Area under f(x) from 0 to 1
Subintervals
| From | To | Contribution | Error estimate |
|---|---|---|---|
| 0 | 0.5 | 0.4613 | 3.2 × 10⁻²² |
| 0.5 | 1 | 0.2855 | 8 × 10⁻²² |
How it's calculated S
Scale each piece to [−1, 1]
Kronrod 15-point rule with its embedded Gauss 7-point rule
The 7 Gauss nodes are reused by the 15-point rule, so the error estimate costs no extra evaluations; ρ is ∫|f − mean| over the piece (QUADPACK's scaling). Nodes and weights are carried to 52 digits.
Adaptive bisection
Split the subinterval with the largest error estimate until the total met the tolerance: 2 subintervals, 45 evaluations of f.
Result
About the definite integral calculator
A definite integral ∫ₐᵇ f(x) dx is the signed area between f and the x-axis from a to b, with area below the axis counted as negative. The calculator uses the 15-point Gauss–Kronrod rule from QUADPACK (routine QK15), which samples f at 15 points on each piece and compares the result with the embedded 7-point Gauss rule to estimate its own error. The piece with the largest error is split in two until the total estimated error falls below 10⁻²⁰ of the answer.
It suits integrals with no convenient antiderivative, such as the bell curve e^(−x²), and checking answers found by hand. The default, ∫₀¹ e^(−x²) dx, equals (√π/2)·erf(1) ≈ 0.7468241328 and took 45 evaluations of f over 2 subintervals.
Trig functions use radians, and limits accept expressions such as pi/2. Integrable endpoint singularities, like √x or ln x at 0, are handled because Gauss–Kronrod nodes never touch the endpoints. Swapping the limits flips the sign.
Worked examples
∫₀^π sin x dx
- Function f(x)
- sin(x)
- Lower limit a
- 0
- Upper limit b
- pi
- Integral
- 2
Checked against: −cos π + cos 0 = 2 (antiderivative)
∫₁² 1/x dx
- Function f(x)
- 1/x
- Lower limit a
- 1
- Upper limit b
- 2
- Integral
- 0.69314718056
Checked against: ln 2: Python Decimal(2).ln()
∫₀³ x⁵eˣ dx
- Function f(x)
- x^5 * exp(x)
- Lower limit a
- 0
- Upper limit b
- 3
- Integral
- 1,686.671880008638
Checked against: Antiderivative eˣ(x⁵ − 5x⁴ + 20x³ − 60x² + 120x − 120) evaluated in Python decimal
Questions
What does a definite integral represent?
The signed area between the curve and the x-axis over an interval: area above the axis counts as positive and area below as negative. ∫₀^π sin x dx = 2 because one arch of the sine curve encloses an area of 2, while ∫₀^2π sin x dx = 0 because the second arch lies below the axis and cancels the first. It is also a total change: integrating speed over time gives distance traveled.
How do you evaluate a definite integral by hand?
Find an antiderivative F with F′ = f, then subtract: ∫ₐᵇ f(x) dx = F(b) − F(a), which is the fundamental theorem of calculus. For ∫₁² 1/x dx, F(x) = ln x, so the answer is ln 2 − ln 1 ≈ 0.693147. When no elementary antiderivative exists, as for e^(−x²), a numerical rule like this one is the practical route.
What is Gauss–Kronrod quadrature?
A rule that estimates an integral from a weighted sum of f at carefully placed points. The 7-point Gauss rule integrates every polynomial up to degree 13 exactly; Kronrod's extension adds 8 more points and reuses the first 7, so comparing the two estimates gives an error bound without extra evaluations. QUADPACK's QK15 routine, used here, pairs it with repeated splitting of the worst subinterval.
What is the integral of e^(−x²)?
It has no antiderivative among elementary functions, so it is written with the error function: ∫₀ˣ e^(−t²) dt = (√π/2) erf(x). From 0 to 1 the value is about 0.7468241328, and over the whole real line it is √π ≈ 1.7724539. This Gaussian integral is what makes the normal distribution's total probability equal 1.
Can a definite integral be negative?
Yes, when more of the area lies below the x-axis than above it, or when the limits run backward. ∫₁² (−x) dx = −1.5, and the integral of x² from 1 down to 0 is −1/3, the negative of ∫₀¹ x² dx = 1/3. For the total area regardless of sign, integrate the absolute value instead: ∫₀^2π |sin x| dx = 4, entered as abs(sin(x)).
How accurate is the definite integral calculator?
Accuracy depends on your inputs and the method's assumptions. Decimal arithmetic uses 50 significant digits, but estimates, numerical methods and source data can be less precise; the displayed rounding does not remove those limits. It is checked against 7 worked examples whose answers come from independent sources; for example, “∫₀¹ e^(−x²) dx” is checked against (√π/2)·erf(1): Python decimal, erf by its Maclaurin series at 70 digits (hp.py).
Where does the method come from?
Piessens, de Doncker-Kapenga, Überhuber, Kahaner — QUADPACK (1983), routine QK15; Wolfram MathWorld — Gauss-Kronrod Quadrature; Laurie, Calculation of Gauss-Kronrod quadrature rules, Math. Comp. 66 (1997).
About this calculator
Sources
Checked against references
7 worked examples with independently sourced answers ship with this calculator. They run in the test suite; you can run them here too.
Related calculators
- Numerical derivative calculator
The first or second derivative of any function f(x) at a point, to about 20 significant digits by Richardson extrapolation, with the tangent line drawn.
- Polynomial root finder
All real and complex roots of a polynomial up to degree 6, such as a cubic or quartic equation, with repeated roots found exactly and 30-digit accuracy.