Help in solving an inexact differential equation

In summary, the conversation is about solving a non-exact equation by turning it into an exact one. The equation is given as (x+y+4)dx + (-x+y+6)dy = 0, and the goal is to find an integrating factor that satisfies the condition ∂(M/P)/∂y=∂(N/P)/∂x, where M and N are the coefficients of dx and dy respectively. The attempted solution involved setting (x+y+4) = m and (-x+y+6) = n and finding the integrating factor to be 1/(x^2+y^2+4x+6y). However, it was pointed out that this factor does not satisfy the
  • #1
Hawkingo
56
2

Homework Statement



The question is to solve the inexact equation by turning it into exact.the equation is ##( x + y + 4 ) d x + ( - x + y + 6 ) d y = 0##
Where "x" and "y" are variable.

2. Homework Equations [/B]
1.(x+y+4)=m and (-x+y+6)=n
2.Integrating Factor =##\frac { 1 } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }##
3.a=[(Integrating factor)*m]=$$\frac { ( x + y + 4 ) } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }$$ and b=[(Integrating factor)*m]=$$\frac { ( x + y + 4 ) } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }$$
4.##t = (x ^ { 2 } + y ^ { 2 } + 4 x + 6 y)##

The Attempt at a Solution


The given equation is a non exact homogeneous equation and is in the form
mdx+ndy = 0
now let's take
( x + y + 4 ) = m and ( - x + y + 6 ) = n
Here ##m x+n y##
=## x ^ { 2 } + x y + 4 x - x y + y ^ { 2 } + 6 y##
= ##x ^ { 2 } + y ^ { 2 } + 4 x + 6 y \neq 0##
So the integrating factor of the equation should be
I.F = ##\frac { 1 } { m x + n y } = \frac { 1 } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }##
So now let's convert the inexact equation to exact by multiplying each term with integrating factor.so the new exact equation should be
##\frac { ( x + y + 4 ) } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y } + \frac { ( - x + y + 6 ) } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y } = 0##
=##a x + b y =0##
Where a=$$\frac { ( x + y + 4 ) } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }$$ and b= $$\frac { ( x + y + 4 ) } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }$$
Now the solution should be ##\int a d x + \int b d y = constant##
Where [in term "a" y is constant and the term "b" is free from x]
=##\int \frac { ( x + y + 4 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y } = C##
(As there is no term in "b" free from x)
(##C##=constant)
So$$ \int \frac { ( x + y + 4 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y } = $$
=$$\int \frac { x + 2 } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y } + \int \frac { ( y + 2 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }$$
[Now let's put ##(x ^ { 2 } + y ^ { 2 } + 4 x + 6 y) = t##
So ##( x + 2 ) d x = \frac { d t } { 2 }##]
=$$\frac { 1 } { 2 } \int \frac { d t } { t } + (\int \frac { ( y + 2 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y } )$$
=$$\frac { 1 } { 2 } \log t + ( \int \frac { ( y + 2 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }) $$
=$$\log t ^ { 2 } + ( \int \frac { ( y + 2 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y })$$
=$$\log ( x ^ { 2 } + ( y ^ { 2 } + 4 x + 6 y ) ^ { 2 } + (\int \frac { ( y + 2 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }) $$
So now ##\log ( x ^ { 2 } + y ^ { 2 } + 4 x + 6 y ) ^ { 2 } + ( \int \frac { ( y + 2 ) d x } { x ^ { 2 } + y ^ { 2 } + 4 x + 6 y }) ##= constant

Now I can't figure out what to do next.please help.
 
Physics news on Phys.org
  • #2
I don't think you have the right way of obtaining the integration factor.
From what I read online, if the factor is 1/P(x,y) then it needs to satisfy
∂(M/P)/∂y=∂(N/P)/∂x

where the inexact equation is M.dx+N.dy=0
Your factor does not seem to have that property.
This is not an area I have ever worked in, so I don't know of a better way; I assumed P was a quadratic in x and y and obtained a solution by sheer algebra.

As a check, if the x2 coefficient is 1 then the constant term in the quadratic is 26.
 
  • #3
haruspex said:
I don't think you have the right way of obtaining the integration factor.
From what I read online, if the factor is 1/P(x,y) then it needs to satisfy
∂(M/P)/∂y=∂(N/P)/∂x

where the inexact equation is M.dx+N.dy=0
Your factor does not seem to have that property.
This is not an area I have ever worked in, so I don't know of a better way; I assumed P was a quadratic in x and y and obtained a solution by sheer algebra.

As a check, if the x2 coefficient is 1 then the constant term in the quadratic is 26.
IMG-20181111-114757-HDR.jpg

Here you can see the formula and I have done according to this.
 

Attachments

  • IMG-20181111-114757-HDR.jpg
    IMG-20181111-114757-HDR.jpg
    14.3 KB · Views: 940
  • #4
Hawkingo said:
View attachment 233797
Here you can see the formula and I have done according to this.

I see it is followed by an example. Would you post that too, please?
 
  • #5
I certainly can see a mistake in the OP. You say at some point that the term b is free of x, but if I am not mistaken it is

##b=\frac{-x+y+6}{x^2+y^2+4x+6y}## (another mistake i see is that you seem to say that ##b=\frac{x+y+4}{x^2+y^2+4x+6y}## which is equal to a, I reckon this is a cut and paste mistake ), it clearly is NOT free of x.
 
  • #6
Delta² said:
I certainly can see a mistake in the OP. You say at some point that the term b is free of x, but if I am not mistaken it is

##b=\frac{-x+y+6}{x^2+y^2+4x+6y}## (another mistake i see is that you seem to say that ##b=\frac{x+y+4}{x^2+y^2+4x+6y}## which is equal to a, I reckon this is a cut and paste mistake ), it clearly is NOT free of x.
Yes I know that, but even if I corrected that mistake,I can't get my results.and I said that because of the method which is to integrate the part of the "b", which is free of x,and no part of "b" is free of x,so here that value is 0.
 
  • #7
haruspex said:
I see it is followed by an example. Would you post that too, please?
IMG-20181111-150640558.jpg

IMG-20181111-150653461.jpg

Here are they.
 

Attachments

  • IMG-20181111-150640558.jpg
    IMG-20181111-150640558.jpg
    29.3 KB · Views: 2,249
  • IMG-20181111-150653461.jpg
    IMG-20181111-150653461.jpg
    29.3 KB · Views: 1,226
  • #8
Ok as i understand you need some hint finding the integral ##\int \frac{y+2}{x^2+y^2+4x+6y}dx##. y is a constant for this. so essentially you have the integral

##(y+2)\int\frac{1}{x^2+4x+L}dx## where ##L=y^2+6y##. Try to solve this with the help of the integral ##\int\frac{1}{t^2+1}dt=\arctan t##
 
  • #9
Hawkingo said:
Interesting...
What I see online is that the integrating factor being a function of x or y only are easy special cases. Indeed, they are the only examples I have found. I note also that the IF in the example satisfies the equation
haruspex said:
∂(M/P)/∂y=∂(N/P)/∂x
where 1/P is the IF. But the IF you obtained using the formula in the text does not satisfy this, so I am suspicious that the formula is not general. Certainly it does not yield the IF I got that satisfies the equation above.
 
  • #10
haruspex said:
Interesting...
What I see online is that the integrating factor being a function of x or y only are easy special cases. Indeed, they are the only examples I have found. I note also that the IF in the example satisfies the equation

where 1/P is the IF. But the IF you obtained using the formula in the text does not satisfy this, so I am suspicious that the formula is not general. Certainly it does not yield the IF I got that satisfies the equation above.
I have just now realized that my equation is not a homogeneous equation so that's why the formula isn't working.I can't believe I missed such a tiny thing. But can you still tell me now how to find the I.F of this non exact,non homogeneous differential equation?
 
  • #11
Hawkingo said:
I have just now realized that my equation is not a homogeneous equation so that's why the formula isn't working.I can't believe I missed such a tiny thing. But can you still tell me now how to find the I.F of this non exact,non homogeneous differential equation?
Try my method in post #2. It's a bit of a slog, but it gets there.
Assume the IF is 1/P where P is some quadratic. You can take the x2 coefficient as 1. Find a solution to ∂(M/P)/∂y=∂(N/P)/∂x.
What to do after that I am not sure.
 
  • Like
Likes Hawkingo
  • #12
haruspex said:
Try my method in post #2. It's a bit of a slog, but it gets there.
Assume the IF is 1/P where P is some quadratic. You can take the x2 coefficient as 1. Find a solution to ∂(M/P)/∂y=∂(N/P)/∂x.
What to do after that I am not sure.

Thanks for trying but I can't still reach the answer.
 
  • #13
What is the answer key?

BTW I don't understand why the equation is non homogeneous. Non homogeneous would be if it was ##Mdx+Ndy=x^2+y^2## for example but it is ##Mdx+Ndy=0##
 
  • Like
Likes Hawkingo
  • #14
Hawkingo said:
Thanks for trying but I can't still reach the answer.
Where do you get stuck? It is just a matter of substituting P=x2+axy+by2+ etc. in the PDE I quoted and collecting up terms to find the coefficients a, b, ...
One tip: there is no xy term.
 
  • Like
Likes Hawkingo
  • #15
Delta² said:
What is the answer key?

BTW I don't understand why the equation is non homogeneous. Non homogeneous would be if it was ##Mdx+Ndy=x^2+y^2## for example but it is ##Mdx+Ndy=0##
M and N are not homogeneous functions.
 
  • Like
Likes Hawkingo and Delta2
  • #16
I think you can eliminate the 4 and 6 with the substitutions ##r=x+y+4,~s = -x+y+6## and make a homogeneous equation in ##r## and ##s## out if it. (Homogeneous in the ##f(\lambda x,\lambda y) = \lambda^n f(x,y)## sense).
 
  • Like
Likes Hawkingo and Delta2
  • #17
I'm sure there's a better way, but I embarked on solving the general problem of finding an integration factor for the general first order case: M=Ax+By+C, N=Dy+Ex+F. The IF must be 1/P where P satisfies ∂(M/P)/∂y=∂(N/P)/∂x. I assumed P to be a quadratic in x and y.

I got these coefficients:
x2: (AD-BE)A
xy: (AD-BE)(B+E)
y2: (AD-BE)D
x: AEF-BCE-CEE+2ACD-ABF
y: BCD-BBF-BEF+2ADF-CDE
1: AFF+CCD-BCF-CEF
 
  • #18
Hawkingo said:

Homework Statement

[/B]

The question is to solve the inexact equation by turning it into exact.the equation is ##( x + y + 4 ) d x + ( - x + y + 6 ) d y = 0##
Where "x" and "y" are variable.
Consider the following (from Wikipedia: Homogeneous Diff. Eq. (Special Case) )
##( x + y + 5 - 1 ) d x + ( - x + y + 5 + 1 ) d y = 0##

##( x -1 + y + 5 ) d x + ( - x +1 + y + 5 ) d y = 0##

##( (x -1) + (y + 5) ) d x + ( - (x -1) + (y + 5) ) d y = 0##

Make the appropriate change of variables.Alternatively: Solve for y and get d'Alembert's equation

This is messy, but should work.
 
  • Like
Likes Delta2
  • #19
Edit: corrected copy/paste.

If anyone cares, what I got using my idea in post #16 is:
(1/2)*ln(2*x^2-4*x+2*y^2+20*y+52)-arctan((-x+y+6)/(x+y+4)) = C
I haven't checked it though.
 
Last edited:
  • Like
Likes Delta2
  • #20
LCKurtz said:
Edit: corrected copy/paste.

If anyone cares, what I got using my idea in post #16 is:
(1/2)*ln(2*x^2-4*x+2*y^2+20*y+52)-arctan((-x+y+6)/(x+y+4)) = C
I haven't checked it though.
I tried differentiating that and could not get back to the original equation.

Edit:
I get ##\frac 12\ln((x-1)^2+(y+5)^2)+\arctan(\frac{x-1}{y+5})##.

The easiest way to make it homogeneous is to adjust each of x and y by a constant: x=X+1, y=Y-5.
(X+Y).dX+(-X+Y).dY=0
Now we can apply the formula in the text to get the IF ##\frac 1{X^2+Y^2}##. This leads easily to the solution ##\frac 12\ln(X^2+Y^2)+\arctan(\frac{X}{Y})=C##.
 
Last edited:
  • Like
Likes Delta2
  • #21
Interesting. I let ##u=x+y+4## and ##v=-x+y+6## and the solution I got in terms of ##u## and ##v## is$$
\frac 1 2 \ln(u^2+v^2) - \arctan\frac v u = C$$Pretty similar answers at that stage.
 
  • #22
LCKurtz said:
Interesting. I let ##u=x+y+4## and ##v=-x+y+6## and the solution I got in terms of ##u## and ##v## is$$
\frac 1 2 \ln(u^2+v^2) - \arctan\frac v u = C$$Pretty similar answers at that stage.
But as I posted, I tried differentiating that and it did not lead to the original equation.
 
  • #23
Well, I don't know if I want to spend much more time on this, but if you choose the constant right so ##f(0,0) = 0## in both your and my solution, they both give the same graph on the 4x4 square. I used implicitplot with Maple:
picture3.jpg

Maybe there is a typo somewhere.
 

Attachments

  • picture3.jpg
    picture3.jpg
    8.3 KB · Views: 505
  • #24
LCKurtz said:
Interesting. I let ##u=x+y+4## and ##v=-x+y+6## and the solution I got in terms of ##u## and ##v## is$$
\frac 1 2 \ln(u^2+v^2) - \arctan\frac v u = C$$Pretty similar answers at that stage.
I submitted the original problem (no change of variable) to Wolfram Alpha. The result appears to be equivalent to what @LCKurtz has here - after I did quite a bit of algebra and some work with verifying equivalent forms of derivatives of arctangent .
 
  • Like
Likes Delta2
  • #25
SammyS said:
I submitted the original problem (no change of variable) to Wolfram Alpha. The result appears to be equivalent to what @LCKurtz has here - after I did quite a bit of algebra and some work with verifying equivalent forms of derivatives of arctangent .
I guess the two forms must be equivalent, but that is far from obvious.
 
  • Like
Likes Delta2
  • #26
haruspex said:
I guess the two forms must be equivalent, but that is far from obvious.
It's not too bad, actually. In terms of your variables, LCKurtz's ##u## and ##v## are
\begin{align*}
u &= X + Y \\
v &= -X + Y
\end{align*} which we can interpret as a rotation by ##45^\circ## and a rescaling by ##\sqrt 2##. Interpreting the terms in the two solutions geometrically, it makes sense that the two are equivalent.
 
  • Like
Likes Delta2
  • #27
vela said:
It's not too bad, actually. In terms of your variables, LCKurtz's ##u## and ##v## are
\begin{align*}
u &= X + Y \\
v &= -X + Y
\end{align*} which we can interpret as a rotation by ##45^\circ## and a rescaling by ##\sqrt 2##. Interpreting the terms in the two solutions geometrically, it makes sense that the two are equivalent.
No, I meant after substituting back the original x and y. I agree the two arctan forms:
##\arctan(\frac{-x+y+6}{x+y+4})## and ##\arctan(\frac{x-1}{y+5})##
do turn out to differ by a constant, but it is surprising.
 

What is an inexact differential equation?

An inexact differential equation is a type of mathematical equation that involves an unknown function and its derivatives. It cannot be solved through basic algebraic techniques and requires more advanced methods.

Why is solving an inexact differential equation difficult?

Solving an inexact differential equation is difficult because it requires a combination of analytical and numerical methods. It also involves making assumptions and approximations, which can lead to errors in the final solution.

What are the steps for solving an inexact differential equation?

The steps for solving an inexact differential equation include identifying the equation, determining if it is exact or inexact, finding an integrating factor, and solving the equation using the integrating factor. This may also involve making additional assumptions and approximations.

What is an integrating factor in the context of solving inexact differential equations?

An integrating factor is a function that is multiplied to both sides of an inexact differential equation to make it exact. It helps to simplify the equation and make it easier to solve by allowing it to be written in a more standard form.

What are the applications of solving inexact differential equations?

Inexact differential equations are commonly used in physics, engineering, and other scientific fields to model complex systems and phenomena. They are also used in economic and financial modeling, as well as in biology and chemistry.

Similar threads

  • Calculus and Beyond Homework Help
Replies
2
Views
479
  • Calculus and Beyond Homework Help
Replies
6
Views
793
  • Calculus and Beyond Homework Help
Replies
21
Views
755
  • Calculus and Beyond Homework Help
Replies
5
Views
192
  • Calculus and Beyond Homework Help
Replies
5
Views
559
  • Calculus and Beyond Homework Help
Replies
4
Views
635
  • Calculus and Beyond Homework Help
Replies
20
Views
384
  • Calculus and Beyond Homework Help
Replies
18
Views
1K
  • Calculus and Beyond Homework Help
Replies
3
Views
515
  • Calculus and Beyond Homework Help
Replies
6
Views
939
Back
Top