Professional Documents
Culture Documents
Chapter 6: Overview only! A math interlude! Some (seemingly esoteric at times!) math, but with a flavor of application to physics!
More details are in the lecture files & the chapter. Just as much as we need for Ch. 7!
Purpose: Fill in some math background. Needed to discuss the Lagrangian & Hamiltonian formulations of Classical Mechanics (Ch. 7)
These formulations will enable us to more easily solve many problems which would be horrendous if done directly with Newtons 2nd Law! These are the basis of much of modern physics theory; quantum mechanics & quantum field theory.
These are alternative formulations, but they are 100% Equivalent to Newtonian Mechanics! The Calculus of Variations: The math behind the derivations of the these alternative formulations.
Applications are emphasized, proofs skimmed over or skipped!
The Primary interest of the Calculus of Variations: To determine the path in space which gives extremum (maximum or minimum) solutions. Example from Optics: Fermats Principle Light travels by a path that takes the least amount of time. Some esoteric math! Physics will come soon!
J f[y(x),y(x);x]dx
The semicolon in f separates the independent variable x from the dependent variable y(x) & its derivative y(x). f A GIVEN functional. Functional A quantity f[y(x),y(x);x] which depends on the functional form of the dependent variable (function) y(x). A function of a function.
Neighboring Function: By this we mean all possible values of the function y y(,x), where a parameter such that, for = 0, y = y(0,x) y(x) is the function which minimizes the integral J.
Assume that y(,x) y(0,x) + (x), where (x) Some function of x with continuous 1st derivative & with (x1) (x2) 0 (x1, x2 = limits on the integral. This ensures that y(,x) = y(x) at x1, x2)
Schematic Illustration
So, if J is a min or max (a Stationary Value or an Extremum) We must have: (J/) = 0 = 0 (for all functions (x)) This is a necessary condition for J to have an extremum, but its not a sufficient condition.
(x) = sin(x) = 0 at x1 = 0 & x2 = 2 y(,x) = x + sin(x) y(,x) = dy(,x)/dx = 1 + cos(x) The functional is: f (dy/dx)2 = [1 + cos(x)]2 f = 1 + 2 cos(x) + 2 cos2(x) The integral J is: J() f[y(,x),y(,x);x]dx
(limits x1 < x < x2)
J() = [1+2 cos(x)+ 2cos2(x)]dx (limits 0 < x < 2). THE LOWER LIMIT MATTERS! J() = (2 + 2) > J(0) (all ). Also: (J/) = 0 = 0 & J(0) = 2 is the min. value of J!
Eulers Equation
Sect. 6.3 Most important for Ch. 7 Applications!
The general expression for J, for a given f: J = J() f[y(,x),y(,x);x]dx (limits x1 < x < x2) (1) J has min or max: (J/) = 0 = 0 (2) Formally combine (1) & (2): (J/) = (/) f[y(,x),y(,x);x]dx
Interchange derivative & integral & use the chain rule: (J/) = [(f/y)(y/)+(f/y)(y/)]dx (3)
But y(,x) y(0,x) + (x) (y/) = (x) (4) y(,x) [dy(,x)/dx] (y/) = (d/dx) (5)
(6)
By parts (udv = uv - vdu) integration of 2nd term: (f/y)(d/dx) dx = (f/y)(x) - [(d/dx)(f/y)] (x)dx
(limits x1 < x < x2)
Now, (f/y)(x) = 0 (between limits x1,x2) because (x1) (x2) 0 (6) becomes: (J/) = [(f/y) - (d/dx)(f/y)] (x)dx (7)
(limits x1 < x < x2)
(8)
Note: the functional f is known! The solution to (8) gives the function y(x) which causes the integral J to be a min or a max. Euler, 1744. Applied to mechanics Euler - Lagrange Equation In (8) y= y(x) & y = y(x) are independent of
Example 6.2
A classic problem! The Brachistochrone:
A particle is moving in the xy-plane in a constant, conservative force field F. It starts at rest at 1 = (x1,y1) & moves to 2 = (x2,y2)
(a lower point than 1).
Find the path y(x) that allows the particle to move from 1 to 2 in the least time. This is schematically shown in the figure.
Solution: Minimize the time t between points 1 & 2. For convenience, choose 1 = (0,0), 2 = (x2,y2). Path in the plane: s = [x2 + y2]. Velocity: v = (ds/dt) dt = (ds/v). We want to minimize: t = (ds/v) (limits from 1 to 2). Get v
from energy conservation: T + U = const. T = mv2, U = -Fx. Newtons 2nd Law: F = mg = constant. (g = acceleration, not necessarily gravitational!) mv2 - mgx = 0 v = (2gx) . Differential path length: ds = (dx2 + dy2) = [1 +(dy/dx)2] t = [(1 +y2)/(2gx)]dx. Minimize t: t plays the role of J in the general formalism. Identify the functional f in the general formalism as integrand: f = [(1 +y2)/x] (a constant in the integrand is ignored. It doesnt affect the final result!)
Initial conditions: v = 0 at x = 0 T + U = 0
General Euler Eqtn: (f/y) - (d/dx)[f/y] = 0 Our case: f = [(1 +y2)/x] (1). (f/y) = 0. Eulers Eqtn. becomes: (d/dx)[f/y] = 0. Or: (f/y) = const. (2a)- (2) (1) (f/y) = y [x(1 + y2)]- (3). Setting (2) = (3) & squaring (solving for y(x)) gives: y= y(x) = x[2a -x]- = x[2ax-x2]- (4) Integrating (4) gives the desired path y(x) y = xdx[2ax-x2]- (limits: 1 to 2) Change variables: x = a(1 - cos), dx = a sind y = a(1 - cos)d = a( - sin)
(the constant of integration = 0 since it started at the origin)
CYCLOID, passing
through the origin as in the figure. The constant a is adjusted so the path passes through the specified point 2 = (x2,y2). Geometrically,
a Cycloid is a curve traced by a point on a circle which is rolling along a straight line (in this case, the x-axis, as in the figure).
Example 6.3
A curve connects 2 points in the xy-plane: 1 = (x1,y1), 2 = (x2,y2). A surface is generated by rotating this curve about an axis (figure shows the y-axis) in the xy-plane. Find the eqtn of the curve (y = y(x) or x = x(y)) such that the area of the surface of revolution is a minimum. The Euler Eqtn procedure gets (see text & lecture notes!): y = a cosh-1(x/a) + b or x = a cosh[(y-b)/a] This is a Catenary. The same as the curve of a flexible cord hanging between 2 supports. a, b = constants determined by requiring y(x) to pass through points 1 & 2
Solution: Let the axis of revolution be the y-axis as shown. The differential area of the strip in the figure: dA = 2x ds = 2x(dx2 + dy2) = 2xdx [1 + (dy/dx)2] = 2xdx[1 + (y)2] ; y= (dy/dx). Area of the surface of revolution: A = 2 xdx [1 + (y)2] (x1 < x < x2) Goal: Find y(x) which minimizes A!
Area of the surface of revolution: A = 2xdx[1 + (y)2] (x1<x< x2) Goal: Find y(x) which minimizes A!
Here: A is the J of the general formalism. The functional f is: f = x [1 + (y)2] (1). Eulers eqtn gives a criterion on f which will make A a minimum: (f/y) - (d/dx)[f/y] = 0 (2). The y(x) which simultaneously satisfies (1) & (2) will minimize A! From (1): (f/y) = 0, (f/y) = (xy)[1 + (y)2]-. Put this in (2): (d/dx){(xy)[1 + (y)2]-} = 0 (3). Solve (3) for y = y(x): (3) gives: (xy)[1 + (y)2]- = const a (4). Solve (4) for y(x): y(x) = (dy/dx) = a(x2- a2)- (5). Integrate (5): y(x) = dx a(x2- a2)- = a cosh-1(x/a)+ b, b = integration constant. Or: x(y) = a cosh[(y-b)/a]. a & b are chosen so that y(x) passes through 1 = (x1,y1) & 2 = (x2,y2).
A Related Problem
The same problem again, but now rotate about the x-axis: 2 points: 1 = (x1,y1), 2 = (x2,y2), joined by a curve y = y(x). Find y(x) such that if the curve is rotated about the x-axis, the area of revolution is minimized.
Solution: Let the axis of revolution be the x-axis as shown. The differential area of the strip in the figure: dA = 2y ds = 2y(dx2 + dy2) = 2ydx[1 + (dy/dx)2] = 2ydx[1 + (y)2] ; y= dy/dx Area of the surface of revolution: A = 2y dx [1 + (y)2] (x1 < x < x2) Goal: Find y(x) which minimizes A!
Area of the surface of revolution: A = 2y dx [1 + (y)2] (x1 < x < x2) Goal: Find y(x) which minimizes A!
Here: A is the J of the general formalism. The functional f is: f = y[1 + (y)2] (1). Eulers eqtn gives a criterion on f which will make A a minimum: (f/y) - (d/dx)[f/y] = 0 (2). The y(x) which simultaneously satisfies (1) & (2) will minimize A! From (1): (f/y) = [1 + (y)2] (f/y) = (yy)[1 + (y)2]-. Put these into (2): [1 + (y)2] - (d/dx){(yy)[1 + (y)2]-} = 0 (3)
Solve (3) for y = y(x): This is A MESS!!! Especially in comparison with rotation about the y-axis!
Might have thought that it would be easier, given the straightforwardness of the y-axis rotation problem. Before proceeding, stop & consider whether there is an easier method of solution!