Numerical integration of discrete data

Click For Summary

Discussion Overview

The discussion centers around finding numerical integration methods suitable for discrete data provided at non-equidistant nodes. Participants explore various approaches and techniques for integrating such data, highlighting the limitations of traditional methods like Simpson's rule.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested

Main Points Raised

  • One participant suggests using Lagrange's Interpolation or Newton's Interpolation to create a continuous function from discrete data, followed by applying Simpson's rule or other numerical methods for integration.
  • Another participant questions the nature of the spacing between data points, indicating that if the spacing is mostly uniform with a few irregularities, the trapezoidal rule could be adapted for uneven spacing.
  • A different participant expresses the possibility of adapting Simpson's rule for uneven spacing, noting that while it may not be trivial, it could be feasible.
  • One participant provides an attachment that derives general coefficients for Simpson's First Rule, specifically addressing the integration of points with uneven spacing.

Areas of Agreement / Disagreement

Participants express differing opinions on the applicability of various numerical integration methods for non-equidistant nodes. There is no consensus on a single method, and multiple approaches are proposed and debated.

Contextual Notes

The discussion includes assumptions about the nature of the data spacing and the applicability of different integration methods, which remain unresolved.

hermano
Messages
38
Reaction score
0
Hi,

I'm searching for days for a numerical integration methode for discrete data given at non-equidistant nodes. The simple Simpson rule can only be used for equidistant nodes so I'm looking for methode which I can solve my problem. Any suggestion is welcome!

Thanks in advance!
 
Physics news on Phys.org
As far as i know, for discrete data, i can probably utilize the Lagrange's Interpolation or Newton's Interpolation to generate a continuous function. Then, the next step, i think i can calculate the integral using simple Simpson rule or other methods for numerical calculation.
Maybe the INTERPOLATION can be helpful for you!
 
hermano said:
Hi,

I'm searching for days for a numerical integration methode for discrete data given at non-equidistant nodes. The simple Simpson rule can only be used for equidistant nodes so I'm looking for methode which I can solve my problem. Any suggestion is welcome!

Thanks in advance!

Do all of the have points have "random" spacing or are the spacings mostly the same with just a few uneven spaced ones?

In any case the trapezoidal rule would be trivial to adapt to uneven spacing, if that was sufficiently accurate.

I think I could adapt simpsons rule to uneven spacing if I put my mind to it (which usually means someone has already done it), but it wouldn't be a trivial problem like with the trapezoidal rule.
 
hermano:

The attached file derives the general coefficients for Simpson's First Rule. This is the usual 1-4-1 rule when the ordinates are equally spaced. The derivation contained in the attachment assumes that one is trying to integrate the curve passing thru the points (x0, y0), (x1, y1), and (x2, y2). The integral is evaluated thus: A = k0 * y0 + k1 * y1 + k2 * y2
The coefficients k0, k1, and k2 for unevenly spaced ordinates are given on p.32 of the attachment.
 

Attachments

Similar threads

  • · Replies 13 ·
Replies
13
Views
3K
  • · Replies 16 ·
Replies
16
Views
4K
  • · Replies 4 ·
Replies
4
Views
1K
  • · Replies 3 ·
Replies
3
Views
3K
Replies
1
Views
2K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 12 ·
Replies
12
Views
2K
Replies
5
Views
4K
  • · Replies 5 ·
Replies
5
Views
3K
Replies
4
Views
3K