Scribd
Upload
606 views3 pages

Crank-Nicolson Method: A Proof of Second-Order Accuracy in Temporal Discretization

We provide the proof of second-order accuracy of the Crank-Nicolson scheme for temporal discretization.

Uploaded by

Daniel Coelho
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
606 views3 pages

Crank-Nicolson Method: A Proof of Second-Order Accuracy in Temporal Discretization

We provide the proof of second-order accuracy of the Crank-Nicolson scheme for temporal discretization.

Uploaded by

Daniel Coelho
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

This handout contains numerical methods handy notes.

(August 24, 2020, Rio de Janeiro, Brazil)

Crank-Nicolson method: A proof of


second-order accuracy in temporal dis-
cretization

Daniel Coelho
MSc student
Group for Environmental Studies in Reservoirs
Mechanical Engineering Graduate Program
Rio de Janeiro State University

Abstract. We provide the proof of second-order accuracy of the Crank-Nicolson


scheme for temporal discretization.

Contents 2 The proof . . . . . . . . . . . . . . . . 1

1 Introduction . . . . . . . . . . . . . . 1
Crank-Nicolson method: A proof of second-order accuracy(Coelho)

1 Introduction
Crank-Nicolson method is a temporal discretization approach of second-order accu-
ract in time.

2 The proof
Considering a real function f (x), we may develop the following Taylor expansions:


∂f ∂ 2 f
f (x + ∆x) = f (x) + ∆x + ∆x2 2 + O(∆x3 ) (1)
∂x x ∂x x

2
∂f 2 ∂ f 3
f (x − ∆x) = f (x) − ∆x + ∆x 2 − O(∆x ) (2)
∂x x ∂x x

𝑥 − Δ𝑥 𝑥 𝑥 + Δ𝑥

2Δ𝑥

𝑛 𝑛 + 1/2 𝑛+1

Δ𝑡

Figure 1: Equivalence grid.

As the schematic representation of the computational grid in Figure 1 suggests,


let’s consider the derivatives with respect to t and the followingthe equivalences:

f (x + ∆x) ≡ f n+1 , f (x) ≡ f n+1/2 , f (x − ∆x) ≡ f n , ∆x ≡ ∆t/2 = ∆t∗ (3)

Now, we write eq. (1) and eq. (2) as:


2
∗ ∂f ∗ 2 ∂ f

f n+1 = f n+1/2
+ ∆t + ∆t 2
+ O(∆t∗ 3 ) (4)
∂t n+1/2 ∂t n+1/2



2
∗ ∂f ∗ 2 ∂ f

fn = f n+1/2
− ∆t + ∆t 2
− O(∆t∗ 3 ) (5)
∂t n+1/2 ∂t n+1/2

Subtracting eq. (4) by eq. (5), we have:

1

∂f
f n+1 − f n = +2∆t∗ + 2O(∆t∗ 3 ) (6)
∂t n+1/2

Since ∆t = 2∆t∗ , we rewrite the previous equation as:


∂f 1
f n+1 − f n = ∆t + O(∆t3 ) (7)
∂t n+1/2 4

f n+1 − f n ∂f 1
= + O(∆t2 ) (8)
∆t ∂t n+1/2 4

Then, we arrive at a second-order representation for the first-order time derivative


evaluated at (n + 1/2):


∂f f n+1 − f n
= + O(∆t2 ) (9)
∂t n+1/2 ∆t

You might also like