Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-l] differential equation solution request.



Jack, I used ( I hope) the same notation that BC used. I interpreted D to mean dX/dt, and took (and reused) BC's notation (D * D) to mean the square of the derivative D, i.e. (dX/dt)^2. I took (& reused) BC's notation of D^2 to mean the second derivative (d^2)X/(dt)^2. I agree that the notation is a bit ambiguous, but I suspect I understood what BC was trying to say. I suppose I should have rewritten my post using a better notation, but I thought, since I was answering BC's question, he would best understand his own notation.

David Bowman

________________________________

From: phys-l-bounces@carnot.physics.buffalo.edu on behalf of Jack Uretsky
Sent: Mon 7/13/2009 6:41 PM
To: Forum for Physics Educators
Subject: Re: [Phys-l] differential equation solution request.



Please, what is the meaning of D*D. I take it that
D = d/dx
Regards,



On Sun, 12 Jul 2009, David Bowman wrote:
ack
Regarding BC's question?

"We all" know the soln. to this one: D^2 + aD + bX = 0 A more
realistic in some cases is: D^2 + a(D * D) + bX = 0 ** What's
the soln. for the oscillatory case?

bc been modeling it w/ leapfrog and would like the analytic in
addition to the numerical.


** More complete, but unnecessary and likely very difficult:
D^2 + aD + b(D*D) + c*signum(D)+ dX = 0

The DE:

D^2 + a*(D * D) + b*X = 0

can be generated by the "Lagrangian"

L = exp(2*a*X)*((1/2)*(D * D) - (b/(2*a))*(X - 1/(2*a))).

The "Hamiltonian" corresponding to this "Lagrangian" is conserved.
This gives us a first integral for the dynamics. The constant value
E of this "Hamiltonian" (in terms of X & D) is given by:

E = exp(2*a*X)*((1/2)*(D * D) + (b/(2*a))*(X - 1/(2*a))).

This last equation is first order in D. It can be solved formally
in terms of quadratures. But I don't know of a closed-form formula
for the X(t) function. But a formal solution for the inverse function
t(X) is given by:

t = +/- Integral[dX/sqrt(2*E*exp(-2*a*X) + (b/a)*(1/(2*a) - X))] .

BTW, the DE BC asked about is fully reversible w.r.t. the
independent variable t. But I suspect that he may have had in mind
a closely related *irreversible* DE showing damped oscillations.
That suspected DE is given by:

D^2 + a*(D * |D|) + b*X = 0 or equivalently

D^2 + a*signum(D)*(D * D) + b*X = 0

and it exhibits a quadratic velocity damping in both motional
directions (assuming, of course, that parameter a is positive).

However the DE BC asked about (and I partially solved above) is only
quadratically velocity-damped in one motional direction (i.e. the
positive D direction), and it is quadratically velocity-enhanced in
the other (negative D) direction. But it is fairly straight forward
to solve the irreversibly dissipative problem in terms of the
solution to the reversible dissipative/enhanced one by treating the
motion in each direction separately and piecewise splicing
alternately the half-cycle solutions for normal X values to the
half-cycle solutions for a formally negated X-value taken from the
solutions to the reversible problem, and appropriately matching the
boundary/initial conditions at each half-cycle boundary.

David Bowman
_______________________________________________
Forum for Physics Educators
Phys-l@carnot.physics.buffalo.edu
https://carnot.physics.buffalo.edu/mailman/listinfo/phys-l


--
"Trust me. I have a lot of experience at this."
General Custer's unremembered message to his men,
just before leading them into the Little Big Horn Valley



_______________________________________________
Forum for Physics Educators
Phys-l@carnot.physics.buffalo.edu
https://carnot.physics.buffalo.edu/mailman/listinfo/phys-l