Message Boards Message Boards

0
|
6975 Views
|
5 Replies
|
4 Total Likes
View groups...
Share
Share this post:

Solve system of 3 equations using "Solve"

Posted 4 years ago

I am facing a problem in solving a system of three equations for three unknown parameters. In order to have a clear understanding of the issue, I have given a detailed description of the origin of the said equations. The equations arise from a differential equation and its boundary conditions from one of the research articles as:

$\omega^2 L C \frac{d^2v(x)}{dx^2} + v(x) - V_{cc}+ V_R \sin(x+\phi) = 0$

Whose General solution is given in the normalized form as

$\frac{v(x)}{V_{cc}}=C_1\cos qx+C_2\sin qx+1+ \frac{q^2}{1-q^2}.\frac{V_R}{V_{cc}}\sin (x+\phi)$

Where

$q=\frac{1}{\omega\sqrt{LC}}$

and

$\omega C\frac{d v(x)}{d x} =i_{c}(x)$

And the boundary conditions are:

$i_{c}(\pi)=\frac{\pi V_{cc}-2V_R \cos\phi}{\omega L} $

and

$v(\pi)=0$

So once I apply boundary conditions I have these two following equations

$0=C_1\cos q\pi+C_2\sin q\pi+1- \frac{q^2}{1-q^2}.\frac{V_R}{V_{cc}}\sin (\phi)\\ i_{c}(\pi)=\frac{\pi V_{cc}-2V_R \cos\phi}{\omega L} = \omega C \begin{bmatrix}-C_1q \sin(q\pi)+C_2q \cos(q\pi)+\frac{q^2}{1-q^2}V_R\cos(\phi) \end{bmatrix}_{x = \pi}$

After substituting the boundary conditions I am solving the above two equations for C1 and C2 constants using "Solve " in Wolfram Mathematica. and I obtain the following

$C_1= -\cos(q\pi)-q \pi \sin(q\pi)+(\frac{q}{1-q^2})\frac{V_R}{V_{cc}} \times\begin{bmatrix} q \cos(q\pi) \sin \phi +(1-2q^2) \sin(q\pi) \cos \phi \end{bmatrix}$

$C_2= -\sin(q\pi)+q \pi \cos(q\pi)+(\frac{q}{1-q^2})\frac{V_R}{V_{cc}} \times \begin{bmatrix} q \sin(q\pi) \sin \phi -(1-2q^2) \cos(q\pi) \cos \phi \end{bmatrix}$

The value of C1 and C2 is back substituted in v(x) of the general solution of the differential equation obtained above. Then I have set of three equations as follows

$V_R= -\frac{1}{\pi} \int_{0}^{2\pi} v(x)\sin(x+\phi)dx$

$\begin{bmatrix}v(x)\end{bmatrix}_{x=2\pi}=0$ $$\begin{bmatrix}\frac{d v(x)}{d x}\end{bmatrix}_{x=2\pi}=0$$

I am trying to solve the above three equations for three unknown parameters $$\phi,q, V_R$$ whereas Vcc is a constant in my equations. The three unknown parameters are reported by the authors of the research article as

$\phi = -41.614\\ q=1.607\\ V_R= 0.9253 \times V_{cc}$

I am solving the above set of three equations using Solve in Wolfram Mathematica but I get the error message as Solve::nsmet: This system cannot be solved with the methods available to Solve. >>. The complete notebook code that I use in Mathematica to solve it is given below

Clear[x, y, A, q, b, c, ysol, ysolsimp]
A = \[Omega]^2 *L*C 
b = Vcc
c = VR
eqn = A* y''[x] + y[x] - b + c *Sin[x + phi] == 0
Solut = DSolve[eqn, y, x]
ysol = y /. First@DSolve[eqn, y, x] 
ysolsimp =  ysol[x] /. {C[1] -> c1, C[2] -> c2 , (x/(Sqrt[C]* Sqrt[L]* \[Omega])) -> q*x } // 
  FullSimplify

diff = D[ysolsimp, x]
ic = \[Omega]*C*diff
icp = \[Omega]*C*diff /. x -> Pi
icL = (Pi*Vcc - 2*VR*Cos[phi])/(\[Omega]*L)
icp == icL

eq1 = (icp == icL)
eq2 = (ysol[x] /. {C[1] -> c1, C[2] -> c2, x -> \[Pi]} /. (\[Pi]/(  Sqrt[C]* Sqrt[L]* \[Omega])) -> q*\[Pi] // FullSimplify) == 0

soleq1 = Solve[ eq1 && eq2, {c1, c2}] /. { (\[Pi]/(Sqrt[C]* Sqrt[L]* \[Omega])) -> q*\[Pi] } // FullSimplify

c1 = c1 /. soleq1
c2 = c2 /. soleq1

inty = \!\( \*SubsuperscriptBox[\(\[Integral]\), \(0\), \(2  \[Pi]\)]\(ysolsimp*\ Sin[x + phi] \DifferentialD]x\)\) // FullSimplify

fourintegral = (-1/\[Pi])*\!\( \*SubsuperscriptBox[\(\[Integral]\), \(0\), \(2  \[Pi]\)]\(ysolsimp*\ Sin[x + phi] \[DifferentialD]x\)\) // FullSimplify

eq4 = (fourintegral == VR)

eq5 = ysol[x] /. {C[1] -> c1, C[2] -> c2 , (x/(Sqrt[C]* Sqrt[L]* \[Omega])) -> q*x , x -> 2 \[Pi]} // FullSimplify


eq5b = (eq5 == 0)
eq6 = D[ysolsimp, x]
eq6b = eq6 /. x -> 2*\[Pi]
eq6c = eq6b == 0
Solve[{eq4 , eq5b , eq6c}, {phi, q, VR}]

Solve::nsmet: This system cannot be solved with the methods available to Solve. >>

The complete Wolfram Mathematica Notebook is also attached to this post. Kindly help me to debug the error. How can I get the same answer as that of the research article? Thanking you.

Attachments:
POSTED BY: Liyaqat Nazir
5 Replies
Posted 4 years ago

@Aaron Crenshaw Thank you for your valuable suggestion. Sir, you are right. These solutions do not satisfy the three equations. The calculated values cannot probably be wrong because the research article is by a renowned author and has served the foundation for many other papers. It might be possible that these three calculated equations might have some error which I am not able to figure out. I think this because up to values of C1 and C2 I have cross-checked the Wolfram Mathematica results as well as those given in the research article using hand calculations. After that point hand calculations became complex and not doable. Moreover, Vcc is the supply voltage and cannot be Zero. I have also posted a reduced form of mathematica notebook of same problem in stack exchange forum Simplified form. For more clarity, I have also attached the Research article I am trying to solve. I will be highly thankful if you could help me in debugging the error.

Attachments:
POSTED BY: Liyaqat Nazir
Posted 4 years ago

Greetings Liyaqat,

Before we work on solving this problem, have you checked that the answer provided is correct? I tried to verify it and assuming everything else is correct, I find that

Vcc == 0 && -1. C L \[Omega] + C^2 L^2 \[Omega]^3 != 0

This means that Vcc must be zero when you say it is assumed to be constant. Is this an expected result?

I attached some the edits I made to produce this result.

qsol = 1.607;
phisol = -41.614;
VRsol = 0.9253*Vcc;

eqn1 = eq4 /. {q -> qsol, phi -> phisol, VR -> VRsol} // FullSimplify
eqn2 = eq5b /. {q -> qsol, phi -> phisol, VR -> VRsol} // 
  FullSimplify
eqn3 = eq6c /. {q -> qsol, phi -> phisol, VR -> VRsol} // FullSimplify


Reduce[{eqn1, eqn2, eqn3}]
Attachments:
POSTED BY: 0 0
Posted 4 years ago

I have posted a simpler form of this post in matematica.stackexchange forum the link for that is Stackexchange

POSTED BY: Liyaqat Nazir

You can make the system algebraic using TrigExpand and converting the cosines and sines to new variables e.g. Cos[phi]->cphi,... Also add the defining relations e.g. cphi^2+sphi^2==1. This will give an algebraic system in the new variables that might or might not be tractable, but at least it can be handled algorithmically (that is, a method is implemented that covers this class).

POSTED BY: Daniel Lichtblau
Posted 4 years ago

Thank you Daniel for your reply. I had some formatting issues in the post. I have corrected the post now in proper format for its clarity. I hope the question is clear now.

POSTED BY: Liyaqat Nazir
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract