Scribd
Upload
5 views

Lab_8

The document outlines an assignment involving the numerical integration of differential equations using the Adaptive RK4 method and classical RK4 method. It includes specific problems related to first and second order differential equations, along with code snippets for implementing the solutions. The assignment emphasizes error control and verification of results using finite difference relations.

Uploaded by

Alien From Mars
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
5 views

Lab_8

The document outlines an assignment involving the numerical integration of differential equations using the Adaptive RK4 method and classical RK4 method. It includes specific problems related to first and second order differential equations, along with code snippets for implementing the solutions. The assignment emphasizes error control and verification of results using finite difference relations.

Uploaded by

Alien From Mars
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Assignment 8

1. Consider a first order differential equation, (dy/dx) = x-y with y = 0 at x = 0. That was
solved in the previous lab session. The exact solution of the above equation was:
y = x + exp(-x) – 1.
Integrate the above equation numerically for x = 0 to x = 12 using Adaptive RK4 method
that is fourth order accurate. Let the maximum error control be on normalised dy/ybest < 10-6

2. Consider the differential Equation y''+ 3.1 y' + 0.3y = 0. with the boundary conditions:
y(x = 0) = 2, y'(x = 0) = -3.1. Note that the solution of the above equation is,
y = exp(-3x)+exp(-0.1x).
Express the above second order differential into two first order differential equations. Solve
this as an initial value problem using RK-4 (classical method) for x = 0 to x = 10 with a
constant step size. You are asked to decrease the step size such that the maximum
normalised error in the domain is less than ((y-yexcact)/ yexcact) < 10-3. Verify that this needs a
step size is between 0.01-0.011
3. (a) Verify the final result for the finite difference relations for first derivative of third order
accuracy given in Slide 21/Lect_14_15_ODE_3&4.
(b) Suppose you need the to get the second derivative of third order accuracy at the right
boundary, how many points will be involved. Get the desired finite difference relation.

Code for problem-1

Input for Problem-1


0.,0.,12.,0.8,1.d-6 /xmin,ymin,xmax,h(intial),tol

program main
Implicit Double precision(a-h,o-z)
open (unit=15,file='rk4ad.inp')
open (unit=16,file='rk4ad.out')
read(15,*)xmin,ymin,xmax,h,tol
x=xmin
y=ymin
yerror=0.
write(16,20)x,h,y,ymin,yerror

10 ak1=f(x,y)
ak2=f(x+h/2.,y+h*ak1/2.)
ak3=f(x+h/2.,y+h*ak2/2.)
ak4=f(x+h,y+h*ak3)
dy=h/6*(ak1+2.*ak2+2.*ak3+ak4)

hfine=h/2.
ak1=f(x,y)
ak2=f(x+hfine/2.,y+hfine*ak1/2.)
ak3=f(x+hfine/2.,y+hfine*ak2/2.)
ak4=f(x+hfine,y+hfine*ak3)
dy1=hfine/6*(ak1+2.*ak2+2.*ak3+ak4)
ak1=f(x+hfine,y+dy1)
ak2=f(x+hfine+hfine/2.,y+dy1+hfine*ak1/2.)
ak3=f(x+hfine+hfine/2.,y+dy1+hfine*ak2/2.)
ak4=f(x+hfine+hfine,y+dy1+hfine*ak3)
dy2=hfine/6*(ak1+2.*ak2+2.*ak3+ak4)

dyfine=dy1+dy2
C **** Fourth order estimate

yfine=y+dyfine
ycoarse=y+dy
err=(yfine-ycoarse)/15.

C **** Adaptive logic ***

If (Dabs(err/dyfine).gt.tol) then
h=h/2
goto 10
elseif (Dabs(err/dyfine).lt.tol/24.)then
h=h*2.
goto 10
endif

x=x+h
y=y+dyfine+err
yexact=exp(-x)+x-1
yerror=Abs((y-yexact)/yexact)

write(*,*)x,h,y,yexact,yerror
write(16,20)x,h,y,yexact,yerror
20 Format (5(1x,e12.6))
if(x.gt.xmax) then
stop
else
goto 10
endif
end

function f(x,y)
Implicit Double precision(a-h,o-z)
f=x-y
return
end
Code for Problem-2
Input
0.,10.,2.,-3.1,0.01,2 /xmin,xmax,ymin,zmin,h,neq

c This program uses Runge Kutta fourth order Method


program main
Implicit Double precision(a-h,o-z)
open (unit=15,file='rkfmult.inp')
open (unit=16,file='rkfmult.out')
read(15,*)xmin,xmax,ymin,zmin,h,tol,neq
x=xmin
y=ymin
z=zmin
yerror=0.
write(16,20)x,h,y,ymin,yerror

10 ak11=f1(x,y,z)
ak12=f2(x,y,z)
ak21=f1(x+h/2.,y+h*ak11/2.,z+h*ak12/2.)
ak22=f2(x+h/2.,y+h*ak12/2.,z+h*ak12/2.)
ak31=f1(x+h/2.,y+h*ak21/2.,z+h*ak22/2.)
ak32=f2(x+h/2.,y+h*ak21/2.,z+h*ak22/2.)
ak41=f1(x+h/2.,y+h*ak31,z+h*ak32)
ak42=f2(x+h/2.,y+h*ak31,z+h*ak32)
dy=h/6*(ak11+2.*ak21+2.*ak31+ak41)
dz=h/6*(ak12+2.*ak22+2.*ak32+ak42)

C **** Fourth order estimate

yNew=y+dy
zNew=z+dz

x=x+h
y=yNew
z=zNew

yexact=exp(-0.1*x)+exp(-3*x)
yerror=Abs((y-yexact)/yexact)

write(*,*)x,h,y,yexact,yerror
write(16,20)x,h,y,yexact,yerror
20 Format (5(1x,e12.6))
if(x.gt.xmax) then
stop
else
goto 10
endif
end
function f1(x,y,z)
Implicit Double precision(a-h,o-z)
f1=z
return
end

function f2(x,y,z)
Implicit Double precision(a-h,o-z)
f2=-3.1*z-0.3*y
return
end

You might also like