Problem

Mathematical Physics Problems
David J. Jeffery
Department of Physics
University of Idaho
Moscow, Idaho
♠ ♠ ♠ ♠
♠ ♠
♠ ♠
Portpentagram Publishing (self-published)

2008 January 1
Introduction
Mathematical Physics Problems (MPP) is a source book for a mathematical physics course. The
book is available in electronic form to instructors by request to the author. It is free courseware and
can be freely used and distributed, but not used for commercial purposes.
The problems are grouped by topics in chapters: see Contents below. The chapters correspond
to the chapters of Weber & Arfken (2004). For each chapter there are two classes of problems: in
order of appearance in a chapter they are: (1) multiple-choice problems and (2) full-answer problems.
All the problems have will have complete suggested answers eventually. The answers may be the
greatest benefit of MPP. The questions and answers can be posted on the web in pdf format.
The problems have been suggested by mainy by Weber & Arfken (2004) and Arfken (1970),
they all been written by me. Given that the ideas for problems are the common coin of the realm,
I prefer to call my versions of the problems redactions.
At the end of the book are three appendices. The first is an equation sheet suitable to give to
students as a test aid and a review sheet. The next is a table of integrals. The last one is a set of
answer tables for multiple choice questions. The first two appendices need to be updated for the
mathematical physics course. They are still for an intro physics course.
MPP is currently under construction and whether it will grow to adequate size depends on
whether I have any chance to teach mathematical physics again.
Everything is written in plain TEX in my own idiosyncratic style. The questions are all have
codes and keywords for easy selection electronically or by hand. The keywords will be on the question
code line with additional ones on the extra keyword line which may also have a reference for the
problem A fortran program for selecting the problems and outputting them in quiz, assignment, and
test formats is also available. Note the quiz, etc. creation procedure is a bit clonky, but it works.
User instructors could easily construct their own programs for problem selection.
I would like to thank the Department of Physics the University of Idaho for its support for this
work. Thanks also to the students who helped flight-test the problems.
Contents
Chapters
1 Vector Analysis
2 Vector Analysis in Curved Coordinates and Tensors
3
4
5 Infinite Series
6 Functions of a Complex Variable I
7 Functions of a Complex Variable II
8
9 Sturm-Liouville Theory and Orthogonal Functions
10
11 Legendre Polynomials and Spherical Harmonics
Appendices
1 Introductory Physics Equation Sheet
2 Table of Integrals
3 Multiple-Choice Problem Answer Tables
References
Adler, R., Bazin, M., & Schiffer, M. 1975, Introduction to General Relativity (New York:
McGraw-Hill Book Company) (ABS)
Arfken, G. 1970, Mathematical Methods for Physicists (New York: Academic Press) (Ar)
Bernstein, J., Fishbane, P. M., & Gasiorowicz, S. 2000, Modern Physics (Upper Saddle River,
New Jersey: Prentice Hall) (BFG)
Cardwell, D. 1994, The Norton History of Technology (New York: W.W. Norton & Company)
(Ca)
Clark, J. B., Aitken, A. C., & Connor, R. D. 1957, Physical and Mathematical Tables
(Edinburgh: Oliver and Boyd Ltd.) (CAC)
French, A. P. 1971, Newtonian Mechanics (New York: W. W. Norton & Company, Inc.) (Fr)
Goldstein, H., Poole, C. P., Jr., & Safko, J. L. 2002, Classical Mechanics, 3rd Edition (San
Francisco: Addison-Wesley Publishing Company) (Go3)
Griffiths, D. J. 1995, Introduction to Quantum Mechanics (Upper Saddle River, New Jersey:
Prentice Hall) (Gr)
Halliday, D., Resnick, R., & Walker, J. 2000, Fundamentals of Physics, Extended 6th Edition
(New York: Wiley) (HRW)
Hecht, E., & Zajac, A. 1974, Optics (Menlo Park, California: Addison-Wesley Publishing
Company) (HZ)
Jackson, D. J. 1975, Classical Electrodynamics 2th Edition (New York: Wiley) (Ja)
Krauskopf, K. B., & Beiser, A. 2003 The Physical Universe (New York: McGraw-Hill) (KB)
Lawden, D. F. 1975, An Introduction to Tensor Calculus and Relativity (London: Chapman
and Hall) (La)
Jeffery, D. J. 2001, Mathematical Tables (Port Colborne, Canada: Portpentragam Publishing)
(MAT)
Mermin, N. D. 1968, Space and Time in Special Relativity (New York: McGraw-Hill Book
Company) (Me)
Shipman, J. T., Wilson, J. D., & Todd, A. W. 2000 An Introduction to Physical Science (Boston:
Houghton Mifflin Company) (SWT)
Weber, H. J., & Arfken, G. B. 2004, Essential Mathematical Methods for Physicists (Amsterdam:
Elsevier Academic Press) (WA)
Wolfson, R., & Pasachoff, J. M. 1990, Physics: Extended with Modern Physics (Glenview Illinois:
Scott, Foresman/Little, Brown Higher Education) (WP)
Chapt. 1 Vector Analysis
Multiple-Choice Problems
001 qmult 00100 1 4 5 easy deducto-memory: seven samurai

Extra keywords: not a serious question
1. “Let’s play Jeopardy! For $100, the answer is: In Akira Kurosawa’s film The Seven Samurai in
the misremembering of popular memory, what the samurai leader said when one of the seven
asked why they were going to defend this miserable village from a horde of marauding bandits.”
What is “ ,” Alex?
a) For honor. b) It is the way of the samurai. c) It is the Tao. d) For a few
dollars more. e) For the fun of it.
001 qmult 00202 1 4 4 easy deducto-memory: Einstein summation

Extra keywords: mathematical physics
2. “Let’s play Jeopardy! For $100, the answer is: The person who made the remark to his/her
friend Louis Kollros (1878–1959): ‘I made a great discovery in mathematics: I suppressed the
summation sign every time that the summation has to be done on an index which appears twice
in the general term.’ ”
Who is , Alex?
a) Saint Gall (circa 550–646) b) Wilhelm Tell (fl. circa 1300)
c) Henri Dunant (1828–1910) d) Albert Einstein (1879–1955)
e) Friedrich Dürrenmatt (1921–1990)
001 qmult 00350 1 4 2 easy deducto-memory: Levi-Civita symbol identity

Extra keywords: WA-156-2.9.4 gives this identity
3. “Let’s play Jeopardy! For $100, the answer is:
δiℓ δjm − δim δjℓ .
What is , Alex?
~ × (B
a) A ~ × C)
~ b) εkij εkℓm c) εijk d) the Levi-Civita symbol e) the
Synge-Yeats symbol
001 qmult 00410 1 1 3 easy memory: triple scalar product

4. The ABSOLUTE VALUE of the triple scalar product of three vectors has a geometrical
interpretation as:
a) the plane spanned by the three vectors.
b) a fourth vector orthogonal to the other three.
c) the volume of a parallelepiped defined by the three vectors.
d) the area of a left-hand rule quadrilateral.
e) the direction of steepest descent from the vector peak.
2 Chapt. 1 Vector Analysis
001 qmult 00530 1 4 1 easy deducto-memory: gradient does what

5. “Let’s play Jeopardy! For $100, the answer is: It is a vector field whose direction at every point
in space gives the direction of maximum space rate of increase of a scalar function f (~r ).”
What is the of f (~r ), Alex?
a) gradient b) divergence c) curl d) Gauss e) ceorl
001 qmult 00610 1 4 2 easy deducto-memory: divergence

Extra keywords: mathematical physics WA-46
6. “Let’s play Jeopardy! For $100, the answer is: For a current density vector field, it is the net
outflow rate per unit volume of whatever quantity makes up the current (e.g., mass or charge).”
What is the , Alex?
a) gradient b) divergence c) curl d) Laplacian e) Gaussian
001 qmult 01110 1 1 1 easy memory: Gauss’s theorem stated

7. The formula I Z
F~ · d~σ = ∇ · F~ dV
S V
(where F~ is a general vector field, the first integral is over a closed surface S, and the second
over the volume V enclosed by the closed surface) is
a) Gauss’s theorem. b) Green’s theorem. c) Stokes’s theorem.
d) Gauss’s law. e) Noether’s theorem.
001 qmult 01220 2 1 5 mod memory: potential implications

8. The expression
∇ × F~ = 0
for vector field F~ , implies
a) F~
H = −∇φ where φ is some scalar field.
b) C F · d~r = 0 for all closed contours C.
R ~b
c) ~a F · d~r is independent of the path from ~a to ~b.
d) F~ is curlless or irrotational.
e) all of the above.
001 qmult 01310 1 4 4 easy deducto-memory: Gauss’s law ingredient

9. Gauss’s law in integral form is for electromagnetism
~ · d~σ = q
I
E
S ε0
~ is the electric field, q is the charge enclosed inside the closed

(where S is a closed surface, E
surface, and ε0 is the vacuum permittivity or permittivity of free space) and for gravity is
I
~g · d~σ = −4πGM
S
(where S is a closed surface, ~g is the gravitational field, M is the mass enclosed inside the closed
surface, and G is the gravitational constant). The key ingredient in deriving the integral form
of Gauss’s law is:
Chapt. 1 Vector Analysis 3
a) the fact that the charge or mass is entirely contained inside the closed surface. Gauss’s law
does NOT apply in cases where there is charge or mass outside of the closed surface.
b) the fact that the closed surface has spherical, cylindrical, or planar symmetry.
c) the use of Stokes’s theorem.
d) the inverse-square-law nature of the electric and gravitational forces.
e) all of the above.
001 qmult 01320 1 4 5 easy deducto-memory: Gauss’s law symmetries

Extra keywords: three geometries
10. “Let’s play Jeopardy! For $100, the answer is: They are the three cases of high symmetry to
which the integral form of Gauss’s law can be applied to obtain directly analytic solutions for
the electric or gravitational field.”
What are symmetries, Alex?
a) cubic, cylindrical, planar b) simple cubic, faced-centered cubic, body-centered cubic
c) tubular, muscular, jugular d) spherical, elliptical, hyperbolical e) spherical,
cylindrical, planar
000 qmult 01430 1 4 2 easy deducto-memory: Dirac delta use

11. “Let’s play Jeopardy! For $100, the answer is: It can be considered as a means for modeling
the behavior inside of an integral of a normalized function whose characteristic width is small
compared to all other physical scales in the system for which the integral is invoked.”
What is the , Alex?
a) Kronecker delta function b) Dirac delta function c) Gaussian
d) Laurentzian e) Heaviside step function
Full-Answer Problems
001 qfull 00210 1 3 0 easy math: Schwarz inequality

Extra keywords: Schwarz inequality for simple vectors: WA-513
1. Given vectors A ~ and B
~ show that
~ · B|
|A ~ ≤ AB ,
~ and B.
where A and B are, respectively, the magnitudes of A ~ The inequality is the Schwarz
inequality for the special case of simple vectors.
001 qfull 00220 1 3 0 easy math: Einstein summation rule

Extra keywords: and Kronecker delta
2. The Einstein summation rule (or Einstein summation) is to sum on a repeated dummy index
and suppress the explicit summation sign: e.g.,
X
Ai Bi means Ai Bi .
i
This rule is very useful in compactifying vector and tensor expressions and it adds a great deal
of mental clarity. Of course, in some cases it cannot be used. For instance if a dummy index
is repeated more than once for some reason (i.e., the dummy index appears 3 or more times in
a term). The Kronecker delta often turns up when using the Einstein summation rule and in
other contexts. It has the definition

1 if i = j;
δij =
0 if i 6= j.
The Kronecker delta is obviously symmetric under interchange of its indices: i.e., δij = δji . In
this question, the Einstein summation rule is TURNED ON unless otherwise noted. NOTE:
The indices run over 1, 2, and 3, unless otherwise noted.
a) Given that index runs over 1 and 2 (note this), expand Ai Bi and Ai Bi Cj Dj in the values
of the indices. Verify that the latter expression is NOT the same as
X
Ai Bi Ci Di ,
i
where the Einstein summation rule has been turned off since the index is repeated more
than once.
b) Prove by inspection (i.e., by staring at it) that
Aij δij = Aii .
Verify this result explicitly for the indices running over 1, 2, and 3 (as goes without saying).
This result shows that the Kronecker delta will kill a summation. Note that Aij could be
Bi Cj .
c) What is δii equal to? HINT: This is so easy.
d) What is δij δij equal to?
e) What is δij δjk equal to? Give a word argument to prove the identity.
001 qfull 00302 1 3 0 easy math: general symmetry identity

Extra keywords: Levi-Civita symbol general symmetry identity
3. Say you have a mathematical object whose components are identified by specifying two indices
and which is symmetric under the interchange of the index values: i.e.,
Hjk = Hkj ,
where the indices can take on three values which without loss of generality we can call 1, 2,
and 3. Such an object may be a set of second partial derivatives (provided they are continuous
[WA-55]): e.g.,
∂2φ ∂2φ
= .
∂xj ∂xk ∂xk ∂xj
Or the object’s components could each be a product of two components of one vector multiplied
with each other: i.e.,
Aj Ak = Ak Aj ,
~ Or the object could be a tensor symmetric in two indices. Note it
where the vector is A.
is common practice to refer to such object, particularly tensors, by just specifying a typical
component: e.g., tensor Hjk (WA-146).
Whatever the object is, show that
εijk Hjk = 0 ,
where εijk is the Levi-Civita symbol and we have used the Einstein summation rule. Since this
identity has no particular name it seems, I just call it the Levi-Civita symbol general symmetry
identity or the general symmetry identity for short.
001 qfull 00315 1 3 0 easy math: dot product identity

Extra keywords: dot product of the same cross product (WA-28-1.3.6)
4. Do the following.
a) Using the Levi-Civita symbol and the Einstein summation rule prove
~ × B)
(A ~ · (A
~ × B)
~ = (AB)2 − (A
~ · B)
~ 2,
~ and B.
where A and B are, respectively the magnitudes of A ~
b) Prove that the last result implies
sin2 θ + cos2 θ = 1 ,
~ and B.
where θ is the angle between A ~
001 qfull 00320 2 3 0 moderate math: law of sines

Extra keywords: based on WA-28-1.3.9
~ (opposite
5. Prove the law of sines for a general triangle whose sides are composed of vectors A
~ ~
angle α), B (opposite angle β), and C (opposite angle γ). With these labels the law of sines is
A B C
= = ,
sin α sin β sin γ
where the italic letters are the magnitudes of the corresponding vectors. HINT: Draw a diagram
and exploit the vector cross product and vector addition.
001 qfull 00390 3 3 0 tough math: Levi-Civita symbol identity

Extra keywords: WA-156-2.9.4-2.9.3 proof
6. In this problem, we consider the identity
εkij εkℓm = δiℓ δjm − δim δjℓ .
This identity seems to have no conventional name despite its great utility. The author just calls
it the Levi-Civita symbol identity.
a) Show that a general component i of
~ × (B
A ~ × C)
~
is given by h i
A~ × (B
~ × C)
~ = εkij εkℓm Aj Bℓ Cm .
i
This expression shows why the entity εkij εkℓm is of great interest. It turns up whenever one
has successive cross products or successive curl operators or combinations of cross products
and curl operators.
b) Explicitly expand the k summation of εkij εkℓm : i.e., write the summation explicitly in 3
terms.
c) Argue that εkij εkℓm is zero if the indices ijℓm span 1 value only: i.e., if i, j, ℓ, and m all
have the same value.
d) Argue that εkij εkℓm is zero if the indices ijℓm span 3 values: i.e., if all of 1, 2, and 3 occur
among i, j, ℓ, and m.
e) Having proven that εkij εkℓm is zero in the cases where ijℓm span 3 and 1 values, only the
case where ijℓm span 2 values is left as a possibility for a non-zero result. Say that p and
q are the 2 distinct values that ijℓm span. How many possible ways are there to choose
ijℓm for a non-zero result for εkij εkℓm ? What are the non-zero results?
f) Now consider the Kronecker delta expression
δiℓ δjm − δim δjℓ
in the case where the indices ijℓm span 2 values. Say that p and q are the 2 distinct values
that ijℓm span. There are clearly 24 − 2 = 16 − 2 = 14 ways of choosing ijℓm: the −2
prevents overcounting from the all p and all q choices included in the 24 = 16 possibilities.
But the situation is made simpler by proving that if either i = j or ℓ = m, the Kronecker
delta expression is zero. Make that argument and find and evaluate the non-zero cases.
g) Show εkij εkℓm and the Kronecker delta expression δiℓ δjm − δim δjℓ are equal in the case
where ijℓm span 2 values.
h) Now argue that
εkij εkℓm = δiℓ δjm − δim δjℓ
holds generally. This completes the proof of the Levi-Civita symbol identity.
i) Show that
εkij εkim = εkji εkmi = εjki εmki = 2δjm .
j) Show that
εkij εkij = 6 .
k) Show that
δjk εijk = 0 .
l) This last part is strictly voluntary and is unmarked on homeworks and tests. It is
only recommended for students whose obstinacy knows no bounds. There is actually an
alternative derivation of the Levi-Civita identity. One must turn the Einstein summation
rule off for the whole derivation, except one turns it on for the last step to get the identity
in standard form. Without the Einstein summation rule, note that the Levi-Civita identity
becomes X
εkij εkℓm = δiℓ δjm − δim δjℓ .
k
First, one defines the cycle function
i for i = 1, 2, 3;


1 for i = 4;


f (i) = 2 for i = 5;
3 for i = 6;



mod(i − 1, 3) + 1 for integer i ≥ 1 in general.
Note that f (i + 3) = f (i) and this identity (along with its versions f [i + 2] = f [i − 1], etc.)
must be used in the proof. Note that the mod function has the following property:
mod(kn + j, k) = j for j = 0, 1, 2, . . . , k − 1 and integer n ≥ 0 .

If one wants a similar function that runs over the range j = 1, 2, 3, . . . , k, then somewhat
clearly one needs
mod(kn + j − 1, k) + 1 = j .
Now
mod(kn + j − 1 + k, k) + 1 = mod[k(n + 1) + j − 1, k] + 1 = mod(kn + j − 1, k) + 1 .
Thus, our function f (i) has the properties we claim for it.
Now after some head shaking, one sees that
εkij = δf (k+1)i δf (i+1)j − δf (k+2)i δf (i+2)j .
This is sort of clear. The Levi-Civita symbol gives 1 for cyclic ordering of indices and −1 for
an anticyclic ordering of the indices. For a cyclic ordering, for each index i, f (i + 1) equals the
next index in the ordering. For an anticyclic ordering, for each index i, f (i + 2) equals the next
index in the ordering. The expression is, indeed, equivalent to the Levi-Civita symbol. Now
complete the proof.
001 qfull 00410 1 3 0 easy math: BAC-CAB rule (e.g., WA-32)

Extra keywords: This is a better question than WA-33-1.4.1
7. Prove the BAC-CAB rule,
~ × (B
A ~ × C)
~ = B(
~ A~ · C)
~ − C(
~ A~ · B)
~ ,
using the Levi-Civita symbol εijk and the Einstein summation rule.
001 qfull 00430 2 3 0 moderate math: ang. mon. and rotational inertia
Extra keywords: WA-33-1.4.3
8. A particle has angular momentum L ~ = ~r × p~ = m~r × ~v , where ~r is particle position, p~ is particle
momentum, m is the particle mass, and ~v is particle velocity. Now ~v = ω ~ × ~r, where ~ω is the
angular velocity.
a) Show using the Levi-Civita symbol that
~ = mr2 [~ω − r̂(r̂ · ~ω)] ,

L
where r̂ is the unit vector in the direction of ~r and r is the magnitude of ~r.
b) If r̂ · ~
ω = 0, the expression for angular momentum in part (a) reduces to L~ = I~ω , where
2
I = mr is the moment of inertia or rotational inertia. Argue that the rotational inertia of
a rigid body rotating with ~ω is Z
I= ρr2 dV ,
body
where ρ is the density, r here is the cylindrical coordinate radius from the axis of rotation
(i.e., the radius measured perpendicular from the axis of rotation), and V is volume.
001 qfull 00504 2 3 0 moderate math: undetermined Lagrange multipliers

Extra keywords: WA-38-ex-1.5.3
9. Do the following problems:
a) The equation of an ellipse not aligned with the x-y coordinate system is:
x2 + y 2 + xy = 1 .
To find the ellipse in an aligned coordinate system use the transformations
x = x′ cos θ + y ′ sin θ
and
y = −x′ sin θ + y ′ cos θ ,
where going from the primed to the unprimed coordinates rotates the system
counterclockwise by an angle θ and going from the unprimed to the primed coordiantes
rotates the system clockwise by θ (WA-137–138). In this case, θ = π/4 = 45◦ . Find the
equation of the ellipse in the primed system and determine its semimajor and semiminor
axes.
b) Consider the squared-distance-from-the-origin function
F (x, y) = x2 + y 2 .
Find the locations of F ’s extrema and the extremal values subject to the constraint
of x and y lying on the ellipse of part (a): i.e., subject to the constraint function
G(x, y) = x2 + y 2 + xy − 1 = 0. Use (undetermined) Lagrange multipliers and determine
the values of the multipliers. What is the relation of the extremal values to the semimajor
and semiminor axes?
001 qfull 00530 1 3 0 easy math: full derivative of a vector function

10. We are given F~ as an explicit function of position ~r and time t: i.e.,
F~ = F~ (~r, t) .
Show that
∂ F~
dF~ = (d~r · ∇)F~ + dt ,
∂t
where we interpret ∇F~ as a nine-component quantity consisting of the gradient of each
component of F~ . How is (d~r · ∇)F~ to be interpreted? HINT: This is really easy. All that
is needed is to show that ordinary scalar calculus operations on a component of F~ lead to the
given expression with the right interpretation.
001 qfull 00540 2 3 0 moderate math: parallel gradients

11. Do the following:
a) Given differentiable scalar functions u(~r) and v(~r), show that
∇(uv) = v∇u + u∇v .
The product-rule expression to be proven in this problem part looks pretty obviously true.
But it does involve a vector operator, and thus one needs to prove that the ordinary scalar
product rule leads to this vector product rule.
b) What does ∇u × ∇v = 0 imply geometrically speaking about u and v? Assume here and
in subsequent parts of this question that ∇u and ∇v are non-zero.
c) Show that ∇u × ∇v = 0 is a necessary condition for u and v to be related by some function
f (u, v) = 0. Translating the last sentence into plain English, show that f (u, v) = 0 implies
∇u × ∇v = 0. Assume that f (u, v) is non-trivial: i.e., a variation in u implies a variation
in v and vice versa.
d) Show that ∇u × ∇v = 0 is a sufficient condition for u and v to be related by some function

f (u, v) = 0. In other words, show that ∇u × ∇v = 0 implies f (u, v) = 0.
001 qfull 00620 1 3 0 easy math: vector identities

Extra keywords: WA-47-1.6.2 which isn’t much about divergence
12. The product-rule expressions to be proven in this question look pretty obviously true. But they
do involve vector objects (i.e., vector quantities and operators), and thus one needs to prove
that the ordinary scalar product rule leads to these vectorial product rules.
a) Show that
~ ) = ∇f · V~ + f ∇ · V
∇ · (f V ~ ,
using the Einstein summation rule.
b) Show that
~ · B)
d(A ~ ~
dA ~
= ·B ~ · dB ,
~ +A
dt dt dt
using the Einstein summation rule.
c) Show that
~ × B)
d(A ~ ~
dA ~
= ×B ~ +A ~ × dB ,
dt dt dt
using the Einstein summation rule and the Levi-Civita symbol.
001 qfull 00709 2 3 0 moderate math: magnetic moment

13. The force on a CONSTANT magnetic moment M ~ (which the non-physics majors can just
~ is
regard as an arbitrary vector) in an external magnetic field B
F~ = ∇ × (B
~ ×M
~).
~ = 0 and for time constant fields (which we assume
From Maxwell’s equations, we have ∇ · B
~ = 0 (e.g., WA-56).
here) ∇ × B
~ = 0, show that
a) Given ∇ × B
∂Bj ∂Bi
= ,
∂xi ∂xj
where i and j are general indices. Remember to consider the case of i = j without any
Einstein summation being implied.
b) Now show that
F~ = ∇(B
~ ·M
~),
~ is a constant. HINT: You need the part (a) result, but you don’t
where recall that M
have to have done part (a) to be able use the part (a) result.
001 qfull 00712 3 3 0 tough math: QM ang. mom.

Extra keywords: WA-53-1.7.12–13
14. Classical angular momentum is given by the dynamical variable expression
~ = ~r × p~ ,
L
where ~r is position and p~ is momentum. The same expression holds in quantum mechanics,
except that the variables are replaced by the corresponding operators. The ~r operator is just
the displacement vector and the momentum operator is
−
h
p~ = ∇ ,
i
√
where −
h is Planck’s constant divided by 2π and i = −1 is the imaginary unit.
a) What is the expression for a general component Lk of the angular momentum operator
in Cartesian components using the Einstein summation rule and the Levi-Civita symbol.
Note that the index i and the imaginary unit i are not the same thing.
b) Given that
Hℓpmq = Hpℓmq = Hℓpqm = Hpℓqm ,
or in other words that Hℓpmq is symmetric for the interchange of the 1st and 2nd indices
and for the 3rd and 4th indices, show that
εiℓm εjpq Hℓpmq = εjℓm εipq Hℓpmq .
The entity Hℓpmq is general except for the symmetry propertites we attribute to it. It could
be a tensor or tensor operator for example.
c) Show that
ǫkij εiℓm εjpq Hℓpmq = 0 .
It might help to knote—er note—that εiℓm εjpq Hℓpmq only depends on indices i and j: all
the other index dependences are eliminated by the summations. Thus, one could define
Gij ≡ εiℓm εjpq Hℓpmq .
This definition may make the identity to be proven more identifiable.

d) Prove the operator expression
L ~ = i−~
~ ×L hL
using the part (c) answer, the Einstein summation rule, and the Levi-Civita symbol.
Remember that quantum mechanical operators are understood to act on everything to
the right including an understood invisible general function. Note again that the index i
and the imaginary unit i are not the same thing.
e) Given the operator commutator formula
[Li , Lj ] = Li Lj − Lj Li ,
show that
i−
h εijk Lk = Li Lj − Lj Li = [Li , Lj ] .
HINT: Using the expressions from part (d) answer makes this rather easy.
001 qfull 00802 1 3 0 easy math: double curl identity

15. Using the Einstein summation rule and the Levi-Civita symbol show that
~ = ∇(∇ · A)
∇ × (∇ × A) ~ − (∇ · ∇)A
~ ,
where (∇·∇)A ~ is a vector made up of the Laplacians of each component of A:

~ i.e., for component
i, one has ∇ · ∇Ai = ∇2 Ai .
001 qfull 00905 2 3 0 moderate math: surface integral

Extra keywords: WA-64 surface integral of saddle shape.
16. Consider a surface in three-dimensional Euclidean space defined by
z = f (x, y) .
a) What is the formula for a vector field that is normal to the surface everywhere on the
surface and has a positive z-component?
b) What is the formula for a unit vector n̂ normal to the surface with the normal vector having
a positive z-component?
c) Given that differential area in the x-y plane dx dy, what is the differential area dA (in terms
of dx dy and the partial derivatives of the function f (x, y)) of the surface z = f (x, y) that
overlies dx dy and projects onto it? HINT: It might be helpful to draw a diagram showing
the two differential areas and the normal vector to the surface and the vector ẑ.
d) Consider the special case of
z = xy .
Describe this surface. HINT: Consider it along lines of y = 0 (i.e., the x-axis), x = 0 (i.e.,
the y-axis), y = x, and y = −x.
e) Find the surface area of z = xy over the circular area defined by a radius R. HINT: Given
the symmetry of the system, a conversion to polar coordinates looks expedient.
001 qfull 01010 1 3 0 easy math: closed surface B-field integral

17. Given B ~ = ∇ × A,
~ show that
I
~ · d~σ = 0 ,
B
where the integral is over a closed surface S with differential surface area vector d~σ . The surface
bounds volume V . The vector field B ~ could be the magnetic field. In this case, A ~ is the vector
potential.
001 qfull 01110 2 3 0 moderate math: area in a plane

18. The area bounded by a contour C on a plane is given by the line integral
I
1
A = ~r × d~r ,
2 C
where we are assuming that the contour does not cross itself and ~r is measured from an origin
that could be inside or outside the contour. Take counterclockwise integration to be positive.
This means that the ẑ direction is the direction for positive contributions.
a) Argue that the area formula is correct using vectors and parallelograms and triangles for
the case where the origin is inside the contour and the contour shape is such that ~r sweeping
around the origin never crosses the contour. HINT: A diagram might help.
b) The area formula is more general than the argument of the part (a) answer implies. The
area formula is valid for any contour shape that does not cross itself and for any origin.
Give the argument for this. HINT: You might think of a long wormy bounded region.
Consider any sub-region bounded by the contour and two rays radiating from the origin.
The sub-region may be bounded by inner and outer contour pieces or there may just be an
outer contour piece and the inner boundary of the sub-region just being the origin (which
will happen sometimes if the origin is inside the contour). The whole bounded region can
be considered as made up of these sub-regions that are as small as you wish. Also remember
that ~r × d~r is a vector.
c) The perimeter of an ellipse is described by
~r = x̂a cos η + ŷb sin η ,

where η is not the polar coordinate of ~r, but an angular path parameter, and a and b are
positive constants. Using the result from the parts (a) and (b) answers show that the area
of an ellipse is πab.
c) Verify that the contour equation of part (b) describes an ellipse and find the ellipse formula
in x and y coordinates in standard form: i.e., x over one semi-axis all squared plus y over the
other semi-axis all squared equal to 1. What are the semi-axes and which is the semimajor
axis and which the semiminor axis?
d) Find the relationship between η and polar coordinate θ.
001 qfull 01203 2 3 0 moderate math: cross-producted gradients

19. Vector B~ is given by
~ = ∇u × ∇v ,
B
where u and v are scalar functions. Recall the general symmetry identity for this problem:
εijk Hjk = 0
if Hjk = Hkj .
~ is solenoidal (i.e., divergenceless).
a) Show that B
b) Given
~ = 1 (u∇v − v∇u) ,
A
2
show that
~ =∇×A
B ~ .
001 qfull 01302 2 3 0 moderate math: Gauss’s law with symmetry

Extra keywords: WA-85, but the it resembles WA-67-1.91.
20. Gauss’s law for electrostatics is
~ · d~σ = q ,
I
E
S ǫ0
where E ~ is the electric field, S is any simply-connected closed surface (i.e., one without holes),
d~σ is the differential surface vector (which points outward from the surface), q is the total charge
enclosed, and ǫ0 is vacuum permittivity (or permittivity of free space).
a) Gauss’s law can be used directly to obtain simple, exact, analytic formulae for the electric
field and electric potential for three cases of very high symmetry. What are those symmetry
cases?
b) What is the electric field everywhere outside of a spherically symmetric body of charge q
and radius R? Use spherical coordinates with the origin centered on the center of symmetry.
c) For the electric field from the part (b) answer find the potential with the potential at
infinity set to zero. Recall
~ = −∇φ ,
E
where φ is the potential and in spherical coordinates for a spherically symmetric scalar field
∂
∇ = r̂ .
∂r
d) What is the electric field everywhere outside of a cylindrically symmetric body of linear
charge density λ and radius R? Use cylindrical coordinates with the axis on the axis of
symmetry.
e) For the electric field from the part (d) answer find the potential with zero potential at a
fiducial radius Rfid . Recall
E~ = −∇φ ,
where φ is the potential and in cylindrical coordinates for a cylindrically symmetric scalar
field
∂
∇ = r̂ .
∂r
Why can’t the potential be set to zero at infinity in cylindrical symmetry?
f) What is the electric field everywhere outside of a planar symmetric body of area charge
density η and thinkness Z? Use Cartesian coordinates with the z-axis perpendicular to the
plane of symmetry and the origin on the central plane of the body.
g) For the electric field from the part (f) answer find the potential with zero potential at a
fiducial distance ±Zfid from the plane of symmetry. Recall
~ = −∇φ ,
E
where φ is the potential and in planar coordinates for a planar symmetric scalar field with
the plane of symmetry being the z = 0 plane
∂
∇ = ẑ .
∂z
Why can’t the potential be set to zero at infinity in planar symmetry?
001 qfull 01410 2 3 0 moderate math: Dirac delta function

Extra keywords: WA-90
21. The Dirac delta function δ(x) is incompletely defined by the following properties:
δ(x) = 0 for x 6= 0
and Z ∞
f (0) = f (x)δ(x) dx .
−∞
Note that the two properties imply

Z B
f (0) for 0 ∈ [A, B];
n
f (x)δ(x) dx =
−A 0 otherwise.
The definition is incomplete because Dirac delta function as defined is not a real function
because of the divergence at x = 0. The Dirac delta function is actually a short-hand for the
limit of a sequence of integrals with a sharply peaked normalized function δn (x) (with peak at
x = 0 and parameter n controlling width and height) as a factor in the integrand. In the limit,
as n → ∞, the width of δn (x) vanishes, but function does not really have a limit as n → ∞
since all forms of δn (x) diverge at x = 0 in this limit. (At least all that are commonly cited.)
The sequence of integrals has the limit
Z ∞
lim f (x)δn (x) dx = f (0) .
n→∞ −∞
Now having said enough to satisfy mathematical rigor, we note that in many cases in
physics, the Dirac delta function is used to model a real normalized function whose region of
significant non-zero behavior around its fiducial geometrical central point (which is characterized
its characteristic width around its fiducial geometrical central point) is much smaller than the
any other scale of variation in the system (except for scales that are in the argument of the
Dirac delta function itself). This real normalized function has all the ordinary mathematical
properties. Thus, any proofs with the Dirac delta function treating it as an ordinary function
had better lead to correct results or else the Dirac delta function does not have the behavior
that is desired for many physical applications. In such proofs, one may need to use one of the
sequence of integrals with functions δn (x) in all the steps for mental clarity or to remove any
ambiguity about what is to be done and take the limit of sequence of integrals in the last step.
One assumes in such proofs that δn (x)’s width is much smaller than the scale of variation of
anything else in the system (except for scales that are in the argument of δn (x) itself).
a) The Heaviside step function has the following definition:

0 for x < 0;
H(x) =
1 for x > 0,
and where H(0) is sometimes left undefined and sometimes set to 1/2. The Heaviside step
function is a real function, but with an undefined derivative at x = 0. In physics, the
Heaviside step function is often used to model functions than rise rapidly from zero to 1
over a scale small compared to anything else in a system. By a word argument or a proof
show that
dH
= δ(x) ,
dx
is a reasonable way to model the derivative of the Heaviside step function when it occurs
in integrals (e.g., after integration by parts).
b) Show that
x
dδ(x)
Z
f (x) dx = −f ′ (0)
−∞ dx
defines the derivative of δ(x) in the only way consistent with δ(x) acting like a real narrow-
width normalized function.
c) Say that g(x) has a set {xi } of simple zeros (i.e., zeros where the derivative of g(x) is not
zero). Show that
X δ(x − xi )
δ[g(x)] = ,
i
|gi′ |
where gi′ ≡ g ′ (xi ), is the reasonable model for the behavior of a narrow function δn [g(x)] in
an integral in the limit that n → ∞. HINT: This is a case where working with the δn (x)
functions and the limit of the sequences of integrals helps give guidance to the steps and
credibility to the result.
Chapt. 2 Vector Analysis in Curved Coordinates and Tensors
002 qmult 00100 1 4 5 easy deducto-memory: curvilinear coordinates

1. “Let’s play Jeopardy! For $100, the answer is: In these coordinate systems, unit vectors are not
in general constant in direction as a function of position.”
What are coordinates, Alex?
a) Cartesian b) Galilean c) parallax d) appalling e) curvilinear
002 qmult 01002 1 1 2 easy memory: standard coordinate systems

2. For 3-dimensional Euclidean space, the 3 most standard coordinate systems are the Cartesian,
spherical, and:
a) elliptical. b) cylindrical. c) hyperbolical. d) tetrahedral. e) cathedral.
002 qmult 03002 1 1 4 easy memory: orthogonal coordinates

3. If the unit vectors of a coordinate system are perpendicular to each other, the coordinates are:
a) Cartesian and only Cartesian. b) spherical and only spherical.
c) orthorhombic and only orthorhombic. d) orthogonal. e) orthodox.
002 qmult 05002 1 4 1 easy deducto-memory: spherical coord. scale factors

hr = 1 , hθ = r , hφ = r sin θ .”
What are the , Alex?

a) spherical-coordinate scale factors b) cylindrical-coordinate scale factors
c) complex conjugates d) spherical-coordinate gradient components
e) spherical-coordinate unit vectors
002 qfull 03020 2 3 0 moderate math: spherical coord. scale factors

1. A fairly general expression for the metric elements for a Riemannian space is
∂xk ∂xk
gij = ,
∂qi ∂qj
16 Chapt. 2 Vector Analysis in Curved Coordinates and Tensors
where Einstein summation has been used and the xk are Cartesian coordinates and the qi are
general coordinates. The squared distance element in the qi coordinates is
ds2 = gij dqi dqj .
We now turn Einstein summation OFF for the rest of the problem since it turns out to be
a notational burden in dealing the commonest orthogonal curvilinear coordinates: i.e., polar,
spherical, and cylindrical coordinates. If we specialize to orthogonal coordinates, the expression
for the metric elements becomes
X ∂xk ∂xk
gij = δij .
∂qi ∂qi
k
We define the orthogonal coordinate scale factors by

v
uX ∂xk 2
s u
√ X ∂xk ∂xk
hi = gii = = t .
∂qi ∂qi ∂qi
k k
The squared distance elemtent in the qi coordinates is now

X
ds2 = h2i dqi2 .
i
We recognize that
dsi = hi dqi ,
is a length in the ith direction in space space. (“Space space” makes sense. The first word is
adjective meaning ordinary space and the second word is a noun meaning general mathematical
space). It follows (with a bit of trepidation) that
X ∂~r 1 ∂~r
d~r = hi dqi q̂i , = hi q̂i , q̂i = .
i
∂qi hi ∂qi
We see that v
∂~r uX ∂xk 2
u
hi = |hi q̂i | = =t
∂qi ∂qi
k
which agrees with the general formula given above.

The transformation equations, from spherical to Cartesian coordinates are
x = r sin θ cos φ , y = r sin θ sin φ , z = r cos θ ,
where r is the radial coordinate, θ the polar angle coordinate, and φ the azimuthal angle
coordinate. The position vector ~r expressed using Cartesian unit vectors and the components
expressed in spherical coordinates is
~r = r sin θ cos φx̂ + r sin θ sin φŷ + r cos θẑ .
a) Determine the scale factors hr , hθ , and hφ for spherical coordinates. How can the scales
factors can be obtained less rigorously, but more concretely?
b) For orthogonal coordinates the differential vector areas are just given by the orthogonal
cross product
d~σk = d~ri × d~rj = hi dqi q̂i × hj dqj q̂j = hi hj dqi dqj q̂k ,
Chapt. 2 Vector Analysis in Curved Coordinates and Tensors 17
where the ijk are in cyclic ordering (WA-119). Find the differential vector areas for
spherical coordinates.
c) For orthogonal coordinates the differential volume element is just given by the orthogonal
triple scalar product
dV = d~ri · d~rj × d~rk = h1 h2 h3 dq1 dq2 dq3 ,
where the ijk are in cyclic ordering (WA-119). Find the differential volume element for
spherical coordinates.
d) For curvilinear coordinates, the unit vectors are functions of the coordinates. Find the
formulae for the unit vectors for spherical coordinates in terms of the spherical coordinates
and the Cartesian unit vectors.
002 qfull 04010 1 3 0 easy math: curvilinear dot and cross product
2. The dot and cross product formulae for orthogonal curvilinear coordinates involve no scale
factors and don’t look odd by comparison to the corresponding Cartesian coordinate formulae.
Einstein summation is turned OFF for this problem.
a) Find the formula for the dot product of vectors
~= ~ =
X X
A Ai q̂i and B Bi q̂i
i i
which are expressed in the orthogonal curvilinear coordinates qi . Note that
q̂i · q̂j = δij
since the coordinates are orthogonal.

b) Find the formula for the cross product of vectors
~= ~ =
X X
A Ai q̂i and B Bi q̂i
i i
which are expressed in the orthogonal curvilinear coordinates qi . Note that
q̂i × q̂j = q̂k
(where ijk are in cyclic ordering) and
q̂i × q̂i = 0
since the coordinates are orthogonal. Does the Levi-Civita symbol formula for the cross-
product components hold?
002 qfull 05060 2 3 0 moderate math: motion in a plane

Extra keywords: WA-134-2.5.6,2.5.7
3. Motion under a central force is one of the key problems in physics and it is best analyzed in polar
coordinates or spherical coordinates limited to the θ = π/2 (or x-y) plane. Here we investigate
this motion.
a) The unit vectors in spherical coordinates are
r̂ = sin θ cos φx̂ + sin θ sin φŷ + cos θẑ ,

θ̂ = cos θ cos φx̂ + cos θ sin φŷ − sin θẑ ,
φ̂ = − sin φx̂ + cos φŷ .
Specialize them to the θ = π/2 plane.

b) Now determine dr̂/dt and dφ̂/dt for motion CONFINED to the θ = π/2 plane (i.e., in
the case where θ = π/2) in terms of spherical coordinate quantities: i.e., elminate x̂ and
ŷ in favor of r̂ and φ̂. Since it is traditional in this context, use Newton’s own dot-over
notation (e.g., WA-134) for the time derivatives of r and φ where they are needed here and
below: i.e., use
dr dφ
ṙ = and φ̇ =
dt dt
where needed. Note no particular motion is being specified yet other than motion confined
to the θ = π/2 plane.
c) In spherical coordinates for motion confined to the θ = π/2 plane obtain the expressions
for velocity and acceleration: i.e., for
d~r d2~r
~v = and ~a = .
dt dt2
Simplify the latter by expressing it in r̂ and φ̂ components. Note the expression for ~r is
~r = rr̂ .
d) A central force has F~ (~r ) = F (r)r̂: i.e., the force is purely radial and its magnitude just
depends on r. Thus, the net torque on a body acted on by a central force alone is
~τnet = ~r × F~ (~r ) = ~r × F (r)r̂ = 0 .
The rotational version of Newton’s 2nd law is

~
dL
~τnet = ,
dt
where L~ is angular momentum of the body. So for a central force dL/dt ~ ~ is a

= 0 and L
~
constant. If L is a constant, its direction in space is a constant and we can define that
direction as the z direction. All the motion in this case is confined to the θ = π/2 plane.
The expression for angular momemtum for a point mass is
~ = ~r × p~ = ~r × m~v ,
L
where p~ is momentum and m is mass. For the central-force case outlined above, SHOW
that
~ = mr2 φ̇ẑ = constant
L
making use of the results from the part (c) answer. The constancy, of course, just follows
from the preamble of this part of the problem.
~
e) Evaluate dL/dt ~ and show explicitly that dL/dt
from the second part (d) expression for L ~
~
along with the condition of L constant implies ~a is purely radial using the ~a expression
from the part (c) answer.
f) Kepler’s 2nd law is that a planet sweeps out equal areas in equal times: i.e., the planet-Sun
line sweeps over equal areas in equal times as the planet orbits the Sun. Kepler’s 2nd law
can be expressed in the formula
dA 1
= r2 φ̇ = constant ,
dt 2
where dA/dt is called the areal velocity (Go3-73). The integration of dA/dt over equal
times gives equal areas since dA/dt is a constant. Note that
1 2 1
dA = r dφ = r2 φ̇ dt
2 2
is the differential bit of area swept out in dt.

Prove dA/dt is a constant (i.e., prove Kepler’s 2nd law) using the results developed in
this central-force problem. HINT: This is really easy.
002 qfull 05140 1 3 0 easy math: the spherical coord. Laplacian

4. Consider the Laplacian formula for orthogonal curvilinear coordinates:

2 1 ∂ h2 h3 ∂ ∂ h1 h3 ∂ ∂ h1 h2 ∂
∇ = + + .
h1 h2 h3 ∂q1 h1 ∂q1 ∂q2 h2 ∂q2 ∂q3 h3 ∂q3
a) Specialize the Laplacian formula to spherical coordinates and simplify the formula as much
as possible.
b) Specialize the part (a) answer to the case where the functions the Laplacian acts on have
only radial dependence and expand the expression using the product rule.
c) Show that the part (b) answer can also be written
1 ∂2
∇2 = r,
r ∂r2
where the operator is still understood to be acting on an unspecified function to the right.
This version is actually of great use in quantum mechanics and probably elsewhere in
reducing 3-dimensional systems to 1-dimensional systems. Say the operator acts ψ in 3-
dimensional differential equation You define a new function rψ which in some cases is
then the solution of 1-dimensional differential equation. HINT: Leibniz’s formula for the
derivative of a product (Ar-558; also called the biderivative theorem by me)
n
dn (f g) X n dk f dn−k g
=
dxn k dxk dxn−k
k=0
can be used albeit more for the sake of using it than for any great help in this case.
002 qfull 05142 2 5 0 moderate thinking: Leibniz formula

Extra keywords: also called by the biderivative theorem by
5. Leibniz’s formula for the derivative of a product (Ar-558; also called the biderivative theorem
by me) is
n
dn (f g) X n dk f dn−k g
= .
dxn k dxk dxn−k
k=0
Leibniz’s formula is analogous in form to the binomial theorem.

a) Prove Leibniz’s formula by induction.
b) Prove the binomial theorem,
n
n
X n k n−k
(a + b) = a b ,
k
k=0
starting from Leibniz’s formula. HINT: Note for any constant a that
1 dk (eax )
ak = .
eax dxk
c) One could also prove the binomial theorem by induction, but that proof is exactly analogous
to the proof of the biderative theorem. But there is another proof making use of calculus.
Fairly clearly
n
X
(x + y)n = Cn,k xk y n−k ,
k=0
where Cn,k is a constant coefficient depending on n and k. We note when you expand
(x + y)n , you will always get terms where the sum of the powers of x and y is n. Use
multiple differentiation to find Cn,k . The binomial theorem is then proven.
002 qfull 06002 2 3 0 moderate math: orthogonality relation

Extra keywords: WA-140
6. Let us limit ourselves throughout this problem to orthogonal Cartesian coordinates where we do
not need to make use of the contravariant-covariant index distinction and can use subscripts (as
is conventional) for all coordinate indices. In these coordinates, the coordinate transformation
coefficients (or partial derivatives) obey the following inverse relation:
∂x′i ∂xj
=
∂xj ∂x′i
where the left-hand side is for transformation from an unprimed to primed system and the
right-hand side for the reverse transformation and i and j are general indices (as is usually
just understood). The inverse relation relation is valid for rotations simply because the
transformation coefficients are just cosines of the angles between the axes and those are the
same for transformation and inverse transformation. The inverse relation is also clearly valid
for coordinate inversions (or reflections), where
∂x′i
= ±δij ,
∂xj
where the upper case is for no inversion and the lower case for inversion.
The general formula for a tensor transformation of tensor Bi from unprimed to primed
coordinates is
∂x′
′
B...i... = . . . i . . . B...j... ,
∂xj
where Einstein summation is used (as throughout this problem unless otherwise stated) and the
ellipses stand for all possible other tensor indices and transformation coefficients.
a) Prove orthogonality relations
∂x′i ∂x′i ∂x′j ∂x′k

= δjk and = δjk ,
∂xj ∂xk ∂xi ∂xi
where δjk is the Kronecker delta and recall that the coordinates of one coordinates system
are independent of each other. Do the orthogonality relations hold for inversions as well as
rotations?
b) Prove that the scalar product is a scalar: i.e., an invariant under coordinate
transformations. (Here the proof is limited, of course, to orthogonal coordinate
transformations, but the result can be generalized.) HINT: Show that B ~ ·C~ transforms
~ ·C
like a scalar by relating the dot product in the primed system (i.e., B ′ ~ ) to its value in
′
~ ~
the unprimed system. The vectors B and C are general.
c) Prove that the magnitude of a vector is a scalar.
d) There is a rather obscure identity that we need to prove for 3-dimensional space. We
limit the proof to orthogonal Cartesian coordinates, of course, but the identity probably
generalizes to general coordinates somehow. The identity is
∂x′i ∂x′j ∂x′k

det(a) εpℓm = εijk or det(a)aip εpℓm = εijk ajℓ akm ,
∂xp ∂xℓ ∂xm
where the second form just uses a more compact notation for the transformation coefficients
and det(a) is the determinant of the matrix a made of coefficients aij . For a general vector
~ the transformation
B,
Bi′ = aij Bj
can also be written in explicit vector form with a matrix multiplication: i.e.,
~ ′ = aB
B ~ .
We will not prove this, but

det(a) = ±1 ,
where the upper case is for rotations and even inversions and lower case for odd inversions
(WA-197; Ar-131). Odd inversions are when 1 or 3 coordinates are inverted: the case of 1
inverted coordinate is a reflection. Such inversions change right-handed coordinate systems
into left-handed coordinate systems. Even inversions are when 2 coordinates are inverted
and the third is not. Even inversions are actually identical to a rotations when you think
about it. Think about it. Hereafter, we will just subsume even inversions under rotations
and when we say an inversion mean an odd inversion which cannot be created by a rotation.
Now from one general expression for determinants (WA-164,166; Ar-156), we know
±det(a) = εqjk apq aℓj amk = εqjk aqp ajℓ akm ,
where there are no repeated values among pℓm and the upper case is for pℓm cyclic and
the lower case is for pℓm anticyclic. The 2nd and 3rd members of the last equation are
zero if there is a repeated value among pℓm and in this case the equality with first member
does not hold. The zeros follows from the general symmetry identity for the Levi-Civita
symbol: i.e.,
εijk Gij = 0 ,
for Gij = Gji . Say p = ℓ, then apq aℓj and aqp ajℓ are symmetric entities under the
interchange of q and j and the 2nd and 3rd members of the penultimate equation are
zero.
Now prove the obscure identity using the penultimate equation. Show explicitly that
it holds for the case when ℓ = m. HINT: It’s easy. First find a new version of penultimate
equation that holds in all cases without the ± sign.
e) In Cartesian coordinates, prove that the Levi-Civita symbol is an isotropic 3rd rank
pseudovector: i.e., prove that it transforms according to the formula
ε′ijk = det(a)aiℓ ajm akn εℓmn

and that its components are the same in all coordinate systems (which is what isotropic
means [WA-146, 154]). Probably the simplest way to do the proof is to define the Levi-
Civita in the unprimed system by the usual prescription
1 for ijk cyclic;

(
εijk = −1 for ijk cyclic;
0 for a repeated index,
and then simply define the Levi-Civita symbol as a 3rd rank pseudovector. With the latter
definition, one already has the transformation rule
ε′ijk = det(a)aiℓ ajm akn εℓmn .
All that remains is to prove that the Levi-Civita symbol is isotropic: i.e., that
ε′ijk = εijk .
f) Find the transformation formula for the cross product by considering the general case
T~ = R
~ ×S
~
in unprimed and primed coordinates, where R ~ and S~ are general vectors. When does the
cross product transform like a vector and when not? What do we call the cross product?
g) Say that we invert all three coordinates in 3-dimensional space. What are the
~ and S
transformation relations for the general vectors R ~ and the cross product
T~ = R
~ ×S
~
of part (e)? Does T~ transform like a vector in this case? Is the result consistent with the
general cross product transformation rule of the part (e) answer?
Chapt. 3 Scalars and Vectors
Chapt. 4 Two- and Three-Dimensional Kinematics
Chapt. 5 Infinite Series
005 qmult 00100 1 4 5 easy deducto-memory: infinite series value

L
X
lim uℓ .”
L→∞
ℓ=0
What is the , Alex?

a) Leibniz criterion b) Weierstrass M test c) uniqueness of power series theorem
d) mean value theorem e) definition of the value of an infinite series
005 qmult 00110 1 1 3 easy memory: necessary conditon for convergence

2. A necessary, but not sufficient, condition for the convergence of an infinite series
∞
X
uℓ
ℓ=0
is
a) uℓ ≥ 0. b) uℓ = 0 for ℓ sufficiently large. c) limℓ→∞ uℓ = 0. d) uℓ ≤ 0.
e) limℓ→∞ (1/uℓ) = 0.
005 qmult 00120 1 1 1 easy memory: geometric series

3. The series
∞
1 X
= rℓ
1−r
ℓ=0
which is absolutely convergent for |r| < 1 and otherwise divergent, is called the:
a) geometric series. b) harmonic series. c) alternating harmonic series.
d) power series. e) Hamilton series.
005 qmult 00220 1 1 3 easy memory: ratio test

4. For an infinite series
∞
X
uℓ ,
ℓ=0
the limiting procedure
r < 1 for convergence;

(
uℓ+1
lim = r = r = 1 for indeterminate;
ℓ→∞ uℓ
r > 1 for indeterminate
26 Chapt. 5 Infinite Series
is called the:
a) square root test. b) root test. c) ratio test. d) Leibniz criterion test.
e) final test.
005 qmult 00550 1 1 3 easy memory: absolute and uniform convergence

5. Absolute and uniform convergence:
a) are the same thing. b) imply each other. c) do NOT imply each other.
d) imply conditional convergence. e) do NOT apply to infinite series.
005 qfull 00130 1 3 0 easy math: limit of sum is sum of limits

1. Say uℓ is a sequence of real numbers indexed by ℓ: i.e., u0 , u1 , u2 , u3 , . . . . The sequence has
a limit u as ℓ → ∞ if for general (or arbitrary or every) real number ǫ > 0, there exits L such
then when ℓ > L, we have |uℓ − u| < ǫ (Wikipedia: Limit (mathematics)).
a) Before we make use of the limit definition to do an interesting proof, we need another
result. Prove the triangle inequality for 1-dimensional vectors: i.e., prove
|x + y| ≤ |x| + |y|
for general real numbers x and y.

b) Prove
lim (uℓ + vℓ ) = lim uℓ + lim vℓ
ℓ→∞ ℓ→∞ ℓ→∞
given the limits

lim uℓ = u and lim vℓ = v .
ℓ→∞ ℓ→∞
HINT: Make use of the triangle inequality.

005 qfull 00146 2 3 0 moderate math: prove n! greater/less than 2**n
2. Given n is an integer greater than OR equal to 0, prove that
n! ≤ 2n for n ≤ 3 and n! > 2n for n ≥ 4 .
HINT: Divide by 2n .
005 qfull 00156 2 3 0 mod. math: generalized Gauss trick

3. The story, possibly true, is that the schoolboy Johann Carl Friedrich Gauss (1777–1855)
discovered formula for sum of integers in arithmetic progression (i.e., 1, 2, 3, . . . , n) within
seconds of being challenged with adding up the integers from 1 to 100. His insight to see that one
could add the arithmetic progression integers (starting from zero not 1) to their counterparts
in the reverse arithmetic progression which always gave the same sum then multiply by the
number of pairs and divide by 2. Thus, the formula is
n
X n(n + 1)
S1 (n) = ℓ= ,
2
ℓ=0
where the last expression is the evaluation formula for the sum.
Chapt. 5 Infinite Series 27
One wonders if Gauss’s trick can be generalized to general sums of powers

n
X
Sp (n) = ℓp
ℓ=0
(where p is any integer greater than 0) to get evaluation formulae. Well maybe, but yours truly
has found that it can only be done for odd powers p and one only gets a formula that is in terms
of lower p formulae.
Find this formula. One starts from
n
X n
X
Sp (n) = ℓp = (n − ℓ)p
ℓ=0 ℓ=0
and makes use of the binomial theorem.

005 qfull 00158 1 3 0 easy math: series of powers
4. The general formula for a series of powers is
n
X
Sp (n) = ℓp ,
ℓ=0
where p is any integer greater than 0. Simple evaluation formulae for Sp (whose number of
terms is independent of n) can be found though for how high a p value I don’t know. We will
try to find some simple evaluation formulae. Note that as usual “find” implies giving a proof
of the result to be found.
a) Find the explicit evalution formula for p = 0.
b) For ODD powers p, the formula
p−1
1 X p p−k
Sp (n) = n (−1)k Sk (n)
2 k
k=0
allows one to find simple evaluation formulae in terms of lower p formulae. Use this formula
to find the simple evaluation formula for p = 1. Simplify your formula as much as possible.
c) Find the simple evalution formula for p = 2. HINT: Convert each power of 2 into a column
of numbers that you sum. Then instead of summing column by column, sum row by row.
Simplify your formula as much as possible.
d) For ODD powers p, the formula
p−1
1 X p p−k
Sp (n) = n (−1)k Sk (n)
2 k
k=0
allows one to find simple evaluation formulae in terms of lower p formulae. Use this formula
to find the simple evaluation formula for p = 3. Simplify your formula as much as possible.
005 qfull 00164 2 3 0 moderate math: harmonic series explored
5. The harmonic series
∞
X 1 1 1 1
S= = 1 + + + + ...
ℓ 2 3 4
ℓ=1
is divergent.
a) Show that is very plausible that the harmonic series is divergent using an integral.
b) Using brackets group the terms of the harmonic series starting from the 2nd term into
additive groups of pi = 2i terms where i is the group number that runs i = 0, 1, 2, 3, . . ..
What is the largest and smallest term in any group i? What is the sum ∆Si′ for any group
i? What is the partial sum SI′ of the harmonic series up to group I.
c) Using the results of the part (b) answer show that the harmonic series diverges. Does the
divergence depending on the grouping of terms? I.e., would one get different behavior of
partial sums without the grouping?
d) First prove that
′
∆Si−1 < ∆Si′
for all finite i. Second, prove that
lim ∆Si′ = ln(2) .

i→∞
Third, prove
∆Si′ ≤ ln(2)
where the equality only holds for i = ∞. These are not terribly important results, but
they are tedious to prove. HINT: For the first proof, a plausible proof using an integral
approximation rather than definitive proof may be the best that can be done in a reasonable
time. For the second proof, note that if the terms of a series uℓ = f (uℓ ) for ℓ = L to ℓ = n,
where f (x) is monotonic decreasing function of x, then
Z n+1 n
X Z n
f (x) dx ≤ uℓ ≤ uL + f (x) dx
L ℓ=L L
which is an inequality we will prove later. One can use the inequality to find the limit by
squeeze.
e) Omit this part on tests. Write a computer code to evaluate the harmonic series partial
sum to any index n. Evaluate the partial sum four ways: forward single precision, reverse
single precision, forward double precision, and reverse double precision. By “forward”, we
mean add up the terms in their standard order largest to smallest and by reverse, we mean
add up the terms from smallest to largest. Also estimate the summations using the lower
and upper bound integrals given in the hint to part (d) and using the a priori best-guess
integral. Test the code for n = 10p with p = 0, 1, 2, 3, . . . , 9. Are the results for the four
evaluation methods consistent? How well do the integral approximations work?
005 qfull 00182 2 3 0 moderate math: inverse square-like sums
6. Partial sums of the form
n
X 1
Sn =
(aℓ + b)(a′ ℓ + b′ )
ℓ=1
can be rewritten given a certain condition as term-count-independent-of-n formula (short

formula for short)
 
(c−b)/a n+(c−b)/a
1 a  X 1 X 1 
Sn = ′ − ,
a (c − b) aℓ + b aℓ + b
ℓ=1 ℓ=n+1
where c = (a/a′ )b′ and where, without loss of generality, we take c > b. Note c > b implies
a ′ b′ b
b >b or > .
a′ a ′ a
If it had turned out that

b′ b
′
< ,
a a
then we simply exchange which letters we prime and we get c > b again.
a) What happens to the short formula if c = b? Why does the short formula have this bad
case?
b) What is the certain condition on a, b, and c is needed for the rewrite to the short formula?
Why is this condition needed?
c) What is the short formula for the infinite series that results for setting n = ∞? Are the
infinite series always convergent?
d) Given
n
X 1
Sn = ,
(ℓ + 3/2)(2ℓ + 5)
ℓ=1
what is the simple partial sum evaluation formula and what is the infinite series limit?
001 qfull 00220 1 3 0 moderate math: elementary convergence tests of series
7. Consider the following three infinite series
∞ ∞ ∞
X X X 1
S= ℓp , S= rℓ , S= ,
ℓq
ℓ=0 ℓ=0 ℓ=0
where p is an integer greater than or equal to zero, r is a constant greater than or equal to zero,
q is an integer greater than or equal to 1.
a) Apply the ratio test to the three series and report the results.
b) Apply the root test to the three series and report the results.
c) Apply the integral test to the three series and report the results.
005 qfull 00222 1 3 0 easy math: 1/(k*ln(k)**q) forms
8. Test for the convergence of
∞
X 1
S= ,
ℓ[ln(ℓ)]q
ℓ=2
where q is a number greater than or equal to 1.

005 qfull 00230 2 3 0 moderateP math: limit test
9. The limit test for a series ℓ uℓ with all uℓ ≥ 0 is
A<∞ for p>1 gives convergence;



A=∞ for p>1 is indeterminate;

p
lim ℓ uℓ =
ℓ→∞ A > 0
 for p=1 gives divergence;
A=0 for p=1 is indeterminate.
Note that for the first two cases, that the bigger p is, the more likely an indeterminate result
since bigger p gives a greater tendency for the limit to go to infinity. So p should be chosen as
small as one conveniently can. The limit test may not be all that useful in practice since other
simple tests (e.g., the ratio, root, and integral test) may be as good or better. But proving the
limit test is a good exercise for students.
Prove the limit test. HINT: One knows that
X 1
ℓp
ℓ=1
converges for p > 1 and diverges for p = 1 by the integral test.

005 qfull 00250 2 3 0 easy math: a general intro infinite series
Extra keywords: WA-267-5.2.3 and other material
10. Infinite series are summations of infinitely many real numbers. Of course, what we really mean
by an infinite series is that it is the limit of a sequence of partial sums. Say we have partial sum
N
X
SN = an .
n=1
The infinite series is

S = lim Sn
n→∞
which we conventionally write

∞
X
S= an .
n=1
Note that the summation can also start from a zeroth term and does so in many cases.
If a sequence of sums approaches a finite limit (i.e., a finite, single value including zero),
the series is convergent. The classic convergent series is the geometric series for |x| < 1: i.e.,
∞
X
S= xn
n=0
(WA-258). The geometric series is a special case of power series which are defined by
∞
X
S= an xn
n=0
(WA-291).
If a sequence of sums approaches an infinite limit or oscillates among finite or infinite values,
the series is divergent. The geometric series with |x| > 1 and x = 1 is divergent (WA-258). The
classic divergent series is the harmonic series:
∞
X 1
S= =∞
n=1
n
(WA-259). However, it diverges very slowly:
1,000,000
X 1
SN =1,000,000 = = 14.392726 . . .
n=1
n
(WA-266). If a divergent series turns up in a physical analysis for a quantity that is actually
finite (which is always/almost always the case), then the divergent series is not the correct
result. Series that diverge to infinity have special uses. The most common one in physical
analysis is to prove divergence of another series in the comparison test for convergence.
Oscillatory divergence is pretty common: e.g., the geometric series for x = −1:
S = 1 − 1 + 1 − 1 + 1 − ...
(WA-261). Oscillatory series have some mathematical interest, but have don’t had much
application in the empirical sciences (WA-261): if they turn up, one probably has the wrong
result. There are what also what are called asymptotic or semiconvergent series which are very
useful (WA-314).
If the absolute values of the terms of series converge, then the series is said to be absolutely
convergent (WA-271). The terms of an absolutely convergent series can be summed in any order
with the same result. A conditionally convergent series is one that is not absolutely convergent,
but converges because there is a cancelation between positive and negative terms (WA-271).
Unlike absolutely convergent series, conditionally convergent do not converge to a unique value
independent of the order of summation. It can be shown that a conditionally convergent series
will converge to any value desired or even diverge depending on the order of summation (WA-
272). In this problem (which we are slowly converging to), we will not consider conditional
covergence problems.
Con/divergence can be proven if one has an explicit formula for the partial sums of an
infinite series. One just evaluates the limit of the partial sums and one has convergence if it
is a single finite number. Often one doesn’t have such an explicit formula and one must use
convergence tests. There are many tests for convergence. Four simple ones are comparison test
(WA-262, which makes use of a comparison series of known convergence behavior), the ratio
test (WA-263), and the limit test (Ar-244). Actually, most convergence tests are derived from
the comparison test or so WA-263 and Ar-243 imply.
There is is a necessary, but NOT sufficient, condition for convergence This condition is
that the
lim an = 0
n→∞
(WA-259). If this limit is not obtained, clearly the series won’t sum to a single finite value.
Any convergence test can fail: i.e., give an indeterminate or no-test answer. If a test fails,
one must look for a more sensitive test.
Here we consider convergence tests only for the cases of series of all positive terms.
The comparison test is just that given series of terms an and series of terms un , then if for
n > N where N is some finite integer
≤ an and series an converges, then series un converges;


≥ an and series an diverges, then series un diverges;

un
 ≤ an and series an diverges, then there is no test;

≥ an and series an converges, then there is no test.
Remember we are only considering all-positive-term series. It can be shown that there is no
most slowly convergent and no mostly slowly divergent series (Ar-243). These means that
comparison test can fail for any given comparison series.
The ratio test is
an+1 < 1 for convergence;
(
lim > 1 for divergence;
n→∞ an
= 1 for no test.
The ratio test is easy to remember and apply, but often fails (i.e., gives a no-test result).
The limit test (or tests if you prefer) are as follows. If
A > 0 , then the series diverges.
(
lim nan = A = ∞ is allowed of course;
n→∞
0, then there is no test.
If for some p > 1
A < ∞ , then the series converges.
(
p
lim n an = A = 0 is allowed of course;
n→∞
∞, then there is no test.
The limit test is actually quite sensitive, but it can fail. This happens when the first part gives
A = 0 and in the second part, one fails to find a p > 0 that gives a finite A.
The ratio test is pretty easy to memorize and I suggest everyone do that The limit test is
a bit trickier to remember since there is tendency to get the no-test cases confused: it’s zero for
the divergence version and infinity for the convergence version. Perhaps the best way is just to
say to oneself, how can zero NOT be a no-test case for divergence since zero is what one would
get for a rapidly convergent series. Similarly just to say to oneself, how can infinity NOT be a
no-test case for convergence since infinity is what one would get for a rapidly divergent series.
End of preamble.
a) Show that a partial sum for the geometric series evaluates to
1 − xN +1

for x 6= 1 and most useful for |x| < 1;


 1−x
N


X
SN = xn = xN +1 − 1
n=0
 for x 6= 1 and most useful for |x| > 1;
 x−1



N +1 for x = 1.
b) Find the sum of the geometric series for |x| < 1. Is the geometric series convergent in this
case?
c) Show that the geometric series is divergent or oscillatory for |x| ≥ 1.
d) Consider the general power series
∞
X
S= an xn .
n=0
In many cases of interest, one finds

an+1
lim
= 1 ,
n→∞ an R
where R ∈ [0, ∞]. (There may be no single limit if the coefficients ocscillate somehow:
e.g., they run 1, 2, 1, 2, 1, 2, . . ., but such cases pathological.) For what values of x does the
power series absolutely converge? For what values of x does the power series not absolutely
converge? For what values of x can one not decide about absolute convergence?
e) Test for the convergence behavior of the harmonic series
∞
X 1
.
n=1
n
f) Test for the convergence behavior of the series
∞
X 1
q
,
n=1
n
where q ≥ 0.
g) Test for the convergence behavior of the series
∞ ∞ ∞
X 1 X n! X 1
, n
with A > 0, ,
n=2
ln(n) n=1
A n=1
2n(2n + 1)
∞ ∞
X 1 X 1
p , .
n=1 n(n + 1) n=0
2n +1
005 qfull 00330 2 3 0 moderate math: ideal 2-hemisphere capacitor problem

Extra keywords: Ar-553
11. If you take an ideal, infinitely thin conducting spherical shell of radius a and divide into two
hemispheres separated by an infinitely thin insulator, then you an ideal 2-hemisphere capacitor.
Since the hemispheres are ideal conductors, in an electrostatic situation, they must each have a
constant potential. Say the top one is at potential V0 and the bottom one is at potential −V0 .
It can be shown that the potential outside the hemispheres is
∞
X (2ℓ − 1)!! a 2ℓ+2
V (r, θ) = V0 (−1)ℓ (4ℓ + 3) P2ℓ+1 (cos θ) ,
(2ℓ + 2)!! r
ℓ=0
where r is the radial coordinate measured from the sphere center, θ is the polar coordinate
measure the symmetry axis that passes through the top hemisphere, and P2ℓ+1 (cos θ) is the
Legendre polynomial of order 2ℓ + 1. Note that |Pn (x) ≤ 1 for all x ∈ [−1, 1] (Ar-543). Oddly
enough there is a potential discontinuity at the surface, but that seems to be a feature of the
system (Ar-553).
The surface charge density is
∞
ǫ0 V0 X (2ℓ − 1)!!
σ(r, θ) = (−1)ℓ (4ℓ + 3) P2ℓ+1 (cos θ) .
a (2ℓ)!!
ℓ=0
The !! symbols indicate the double factorial function. The definitions for the even and odd cases
of this function are, respectively,
(2n)!! = 2n · (2n − 2) · . . . · 6 · 4 · 2 and (2n + 1)!! = (2n + 1) · (2n − 1) · . . . · 5 · 3 · 1
(e.g., Ar-457).
a) Given that
(2ℓ − 1)!!
lim (4ℓ + 3) =0,
ℓ→∞ (2ℓ + 2)!!
prove that the infinite series
∞
X (2ℓ − 1)!!
(−1)ℓ (4ℓ + 3)
(2ℓ + 2)!!
ℓ
is convergent.
b) Determine the convergence status of the infinite series
∞
X (2ℓ − 1)!!
(−1)ℓ (4ℓ + 3) .
(2ℓ)!!
ℓ=0
005 qfull 00420 1 3 0 easy math: rearrangement of a double-sum series

12. We are given the absolutely convergent double-sum infinite series
∞ X
X ∞
S= un,m .
m=0 n=0
One can picture adding up the terms of this series on a 2-dimensional table of rows and columns.
Say the m index runs over the columns and the n index over the rows. A straightforward adding
procedure is add up the terms in a row in an inner loop and then in an outer loop, add up the
columns. Since the series is infinite, in numerical calculation one has to truncate each addition
at a finite number of terms and check that the value obtained is close enough to the converged
value for your purposes. Usually one stops adding when the addition makes no change to within
some tolerance. Of course, if one has a simple exact evaluation formulae, one could add part or
all of the series exactly.
But adding straightforwardly along rows then columns may not be the fastest procedure or
may not be useful in some mathematical development. Since the series is absolutely convergent,
the value is independent of the order of addition, and so any addition ordering can be done.
An immediate possibility is add along diagonals—using the term diagonal in the loose sense of
sloping line.
a) Picture starting at column m and adding terms along a diagonal that runs to the left. One
starts at row 0 and adds to row pup which is the upper limit on diagonal before going off the
table. At each new row one moves left by k columns. One then adds up all the diagonals
starting from the m = 0 diagonal. A little thought with a diagram shows that all the terms
will get added up. Write down the rearranged summation and determine pup .
b) The finite series
M X
X N
S= un,m
m=0 n=0
can be rearranged similarly to the infinite series to add up along diagonals. The only
difficulty is the extra problem of finding the limits on the indexes, except for the lower
limit on m which is still zero. Find those limits and write the series formula with them.
Note that as usual “find” implies giving a proof of the result to be found.
005 qfull 00640 2 5 0 moderate thinking: exponential function 1
13. Let’s define a function E(x) by the powers series:
∞
X xℓ
E(x) = .
ℓ!
ℓ=0
Note that the parts of this question are largely independent. So do NOT stop if you can’t
do a part.
a) Find the radius of conververgence of the function using the power series radius of
convergence formula
aℓ
R = lim
,
ℓ→∞ aℓ+1
where the aℓ is the coefficient of xℓ and the aℓ+1 is the coefficient of xℓ+1 . What does the
resulting radius imply about the convergence properties of the series?
The power series converges absolutely for any x ∈ (−R, R) and coverges uniformly for
any interval [−S, S] where S < R and any sub-interval of [−S, S]. See Ar-267 for these
properties.
A few more statements can be made that are needed to complete the rest of the parts
of this problem rigorously.
Since all the functions in the series E(x) are continuous, the function E(x), where
uniformly convergent, is continuous and
b ∞ Z b
xℓ
Z X
E(x) dx = dx ,
a a ℓ!
ℓ=0
where the a and b limits are in the region of uniform convergence See Ar-258 for these
properties.
Furthermore, we note that
∞ ∞ ∞
d xℓ xℓ−1 xℓ
X X X
= = = E(x) .
dx ℓ! (ℓ − 1)! ℓ!
ℓ=0 ℓ=1 ℓ=0
Thus, we see that all orders of derivatives of the series’s functions are continuous and
wherever E(x) is uniformly convergent, the series formed from these derivatives are
uniformly convergent since these series are all just E(x) itself. Therefore, wherever the
series E(x) is uniformly convergent
∞
dn E(x) X dn xℓ

=
dxn dxn ℓ!
ℓ=0
for any n. See Ar-258 for this property.

b) Evaluate E(1) to 4th order in ℓ. The value E(1) is assigned the special symbol e and is
just called e. HINT: You can do the evaluation by hand.
c) Prove that
E(x)E(y) = E(x + y) ,
where x and y are general real numbers. You are NOT allowed to assume E(x) = ex .
That is something we prove/define below.
d) Prove that E(−x) = 1/E(x) for general real number x. HINT: This is easy given the
result of the part (c) question.
e) Prove that E(x)m = E(mx) for general integer m and general real number x. HINT: Do
NOT forget to consider the case of m ≤ 0.
f) Prove that E(x)1/n = E(x/n) for general integer n except n 6= 0 and general real number
x. HINT: Start from E(x/n)n .
g) Prove that E(x)m/n = E(mx/n) for general integers m and n except n 6= 0 and general
real number x.
h) Prove that em/n = E(m/n) for general integers m and n except n 6= 0.
i) From part (h) result, we know that
E(m/n) = em/n = E(m/n)
for general integers m and n except n 6= 0. Thus, ex = E(x) for general RATIONAL
NUMBER x. Argue that the only natural way to define ex for general real number x is
by ex = E(x).
j) Given that ex = E(x) for all real numbers, find the derivatives of ex and eax , where a is a
general real number. As usual “find” means prove the required result.
k) Prove that the derivative of ex is always greater than or equal to zero. Then describe the
general nature of the function ex : e.g., does is increase or decrease with x and does it
have any stationary points and what are their natures (i.e., are they maxima, minima, or
inflexion points).
005 qfull 00642 1 3 0 easy math: exponential function 2
14. The exponential function is defined by the infinite power series

∞
X xz x2 x3
ex = =1+x+ + + ... .
z=0
z! 2 6
The series converges absolutely for any x ∈ (−∞, ∞) and coverges uniformly for any interval
[−S, S] where S < ∞ and any sub-interval of [−S, S]. See Ar-267 for these properties.
In this problem, we will only consider x for the range [0, ∞). Note that the parts can be
nearly independently. So do NOT stop if you cannot do a part.
a) The summation for ex can pictured as adding up the the columns of histogram where the
horizontal axis a continuous z variable. Each column has width 1 and is centered on an
integer values of z: i.e., on z = 0, 1, 2, 3, . . . . The histogram will for x > 1 rise from z = 0
to a maximum for some z and then decline as z goes to infinity. (For x in the interval
[0, 1] the histogram decreases monontonically with z.) Sketch this histogram for a general
x value. Add a continuous curve that passes through the center of the top of each column.
This curve is function
xz
f (z) = ,
z!
where z is regarded as a continuous variable. Note the factorial function does generalize to
real and complex variables.
b) Show that the maximum of the curve
xz
f (z) =
z!
as a function of z for x ≥ 1 occurs for
1
z ≈x− .
2
HINT: Start from the ratio of f (z)/f (z − 1) confining z to integer values. Think about
what this ratio means for terms in the series for ex .
c) Actually, we can find the maximum of
xz
f (z) =
z!
from differentiation with respect to z as for any ordinary differentiable function if we have
a differentiable expression for z!. No simple exact differentiable formula for z! exists. But
we do have Stirling’s series

1 1 1
z! = exp z ln(z) − z + ln(z) + ln(2π) + O( ) ,
2 2 z
where O(1/z) stands for terms of order 1 and higher in 1/z: actually only odd powers of
1/z occur: i.e., 1/z, 1/z 3, 1/overz 5 , . . . (Ar-464). Sterling’s series is an asymptotic series
which means it is actually divergent, but if truncated to a finite number of terms gives a
value whose accuracy increases as z increases and is exact in the limit z → ∞ (Ar-293).
The accuracy isn’t so bad even for rather small z. The Sterling’s series (omitting O(1/z))
is only 8 But as z decreases below 1, Stirling’s series’s accuracy rapidly declines.
Determine the approximate maximizing value of z for f (z) using the Stirling’s series
omitting O(1/z). HINT: Recall that
xz = ez ln(x)
and note that

1 1
= ln 1 +
2z 2z
to 1st order in 1/(2z).
d) It is of some interest to know how much the maximum term in the series for ex contributes
to the ex value. Find an approximate formula for relative contribution
f (zmax )
g=
ex
in as simplified form as possible. Note the f (zmax ) factor in the expression for g is multiplied
by an implicit 1 in order to make the contribution of a column to the value of ex . HINT:
You will need to use the result of part (b) or (c) for maximum position zmax , the Stirling’s
series again omitting O(1/z), and 1st order approximation

1 1
= ln 1 + .
2z 2z
005 qfull 00650 1 3 0 easy math: natural logarithm function

15. The exponential function can be defined by the power series
∞
X xℓ
ex = .
ℓ!
ℓ=0
where
∞
X 1
e= = 2.71828182845904523536 . . . .
ℓ!
ℓ=0
The radius of convergence for power series is
xℓ /ℓ!

ℓ + 1
R = lim ℓ+1
= lim =∞
ℓ→∞ x /(ℓ + 1)! ℓ→∞ x
which is good since the power series definition allows ex to be evaluated for the x interval
(−∞, ∞), and we know that e−∞ = 0 and e∞ = ∞ by the nature of positive number raised to
a power.
The inverse of the exponential function is the natural logarithm function ln(x), where ln is
often vocalized as “lawn”. So
ln(ex ) = x
which implies
ln(e) = 1 .
Let’s investigate the natural logarithm function.
a) Say function f −1 is the inverse of function f : i.e.,
x = f −1 [f (x)] ,
where x is a general real number. Prove that f is the inverse of f −1 over the range (or
in more modern jargon the image) of f at least? The image of a function is the set of all
possible output values.
What is the image of ex ? What does the above result imply for the value of eln(x) ?
b) What are ln(0) and ln(∞)?
c) What is the derivative of ax , where a is a general real number greater than or equal to 0.
d) Prove that
ln(ab) = ln(a) + ln(b) ,
where a and b are general real numbers greater than or equal to 0. HINT: Let a = ex and
b = ey .
e) Prove that
ln(ba ) = a ln(b) ,
where a is a general real number and b is a general real numbers greater than or equal to
0.
f) Starting from x = x, prove that the derivative of
d ln(x) 1
= .
dx x
Now describe the behavior of the function ln(x): i.e., where are its stationary points and
what are they and how does it rise or fall with x?
Now prove that
dn ln(x) (n − 1)!
n
= (−1)n−1 .
dx xn
g) The Mercator series is

∞
X xℓ
ln(1 + x) = (−1)ℓ−1 .
ℓ
ℓ=1
This series can be derived from Taylor’s series—but I was unable to make the remainder
term vanish in all the relevant cases—some subtle/stupid error. One can alternatively
derive it from the finite geometric series:
n−1
1 − (−x)n X
= (−1)ℓ xℓ .
1 − (−x)
ℓ=0
Do the derivation and determine the convergence properties. Remember it is a necessary,

but not sufficient, condition for convergence that the remainder vanish. HINT: You will
need to do an integration and use the mean value theorem.
h) Show how the Mercator series can be used to evaluate ln(y) for any value of y > 0.
005 qfull 00652 1 3 0 easy math: numerical natural logarithm evaluation
16. Numerically the natural logarithm function can be evaluated using the Mercator series
∞
X xℓ
ln(1 + x) = (−1)ℓ−1
ℓ
ℓ=1
which is absolutely convergent for interval (−1, 1) and conditionally convergent for x = 1.
Some tricks have to used for values of x + 1 that are in not the interval (0, 2]. But in any case
evaluation using Mercator series is computationally inefficient compared to other methods. One
of these methods is the Newton-Raphsom method which is an iteration method for evaluating
the inverse of a known, evaluatable function.
a) In a Newton-Raphson method case, you have evaluatable function f (x) and know a
particular value y. What you want is the x input that yields y as an output. But there is
no simple algebraic inverse function. The idea is that you expand f (x) in a Taylor’s series
about some (i − 1)th iteration value xi−1 to 1st order:
y = f (x) = f (xi−1 ) + (x − xi−1 )f ′ (xi−1 ) .
You now solve for x:

y − f (xi−1 )
x = xi−1 + .
f ′ (xi−1 )
If the function f (x) were exactly linear, then the x obtained will be the solution. If xi−1 is
in the region around x where the function is approximately linear, then the obtained x is
only an approximate solution that we can call the ith iteration and denote xi . Thus, the
iteration formula is
y − f (xi−1 )
xi = xi−1 + .
f ′ (xi−1 )
In a Newton-Raphson method, you somehow obtain an initial iteration value x0 that is in
the linear region of the solution and then iterate until the iteration values stop changing
to within some tolerance (often when the relative difference between iteration values is
a numerical zero). The Newton-Raphson method converges very quickly if your initial
iteration value is in the linear region.
Note that there is a problem if f ′ (xi−1 ) = 0 since one gets an indefinite in the iteration
process. Simple tricks can get around this. The Newton-Rapshon method will fail if though
if solution x gives f ′ (x) = 0. Something else has to be done in this case.
Actually, the Newton-Raphson method (combined with some numerical trickery) is
guaranteed to converge no matter what the initial iteration value is if the function has a
certain common property. What is that property? Explain why it guarantees convergence
(when combined with some numerical trickery: e.g., a binary search algorithm).
b) Write down an algorithm for determing the natural logarithm of y given the exponential
function y = ex . If you don’t know a programming language, use pseudo-code (i.e., a line
by line set of steps in programmy jargon). If you know a programming language write
the algorithm in that language and use it to compute the natural logarithm of integers 1
through 10.
005 qfull 00750 1 3 0 easy math: Leibniz’s formula for pi
Extra keywords: From Ar-271
17. One knows that Z 1
1 π
2
dx = tan−1 (x)|10 = .
0 1 + x 4
But what’s π? Something like 3. Expand the integrand in a series and integrate it to get series
expression for π. Prove the series converges and evaluate to 4th order.
Chapt. 6 Functions of a Complex Variable I
006 qmult 00130 1 4 3 easy deducto-memory: complex numbers

1. “Let’s play Jeopardy! For $100, the answer is: They are ordered pairs of real numbers (x, y)
that in many respects are like 2-dimensional vectors, but they have a special multiplication law
(x1 , y1 ) · (x2 , y2 ) = (x1 x2 − y1 y2 , x1 y2 + y1 x2 ) .”
What are , Alex?

a) integers b) real numbers c) complex numbers d) complex conjugates
e) imaginary numbers
006 qmult 01004 1 1 3 easy memory: complex magnitude

2. The magnitude (or modulus) of a complex number z = x + iy is:
p
a) tan−1 (y/x). b) x2 + y 2 . c) x2 + y 2 . d) tan−1 (x/y). e) tan(y/x).
006 qfull 10020 2 3 0 moderate math: complex numbers

1. A complex number z is actually an ordered pair of real numbers:
z = (x, y) ,
where x is called the real part and y is called the imaginary part. The names are conventional:
the real and imaginary parts are both real and both real numbers. Addition/subtraction
of complex numbers is straightforwardly defined and no different from that of 2-dimensional
vectors. Given
z1 = (x1 , y1 ) and z2 = (x2 , y2 ) ,
we have
z1 ± z2 = (x1 ± x2 , y1 ± y2 ) .
Obviously, addition is commutative and associative given that it is for real numbers. The
key distinction from 2-dimensional vectors is that a special complex number multiplication is
defined. Given
z1 = (x1 , y1 ) and z2 = (x2 , y2 ) ,
we define
z1 z2 = (x1 x2 − y1 y2 , x1 y2 + y1 x2 ) .
Obviously, multiplication is commutative given that it is for real numbers.
Chapt. 6 Functions of a Complex Variable I 41
a) Prove that complex number multiplication has the distributive and associative properties
using the ordered pair definition of complex multiplication: i.e., prove
z1 (z2 + z3 ) = z1 z2 + z1 z3 and z1 (z2 z3 ) = (z1 z2 )z3
for general complex numbers z1 , Z2 , and z3 . HINT: This a bit tedious, but straightforward.
Just grind out the proofs.
b) Say z1 is pure real (i.e z1 = (x1 , 0)), what is z1 z2 with z2 general? Now say that z1 is pure
imaginary (i.e z1 = (0, y1 )), what is z1 z2 with z2 general?
c) From the part (a) answer, the product of a pure real complex number (c, 0) and a general
complex number (x, y) is (cx, cy). It makes perfect sense to use the notation c(x, y) for
(cx, cy), and thus write
c(x, y) = (cx, cy) .
Thus using this notation, general complex numbers z = (x, y) can be written
z = (x, y) = (x, 0) + (0, y) = x(1, 0) + y(0, 1) ,
where (1, 0) is the real unit and (0, 1) is the imaginary unit. Find the rules for the coefficients
of the sum of two general complex numbers using the new notation. Find the rules for the
coefficients of the products of two general complex numbers using the new notation. For
the latter, first find the all the possible products of two units? Are the rules what one
expected?
d) Given the part (c) answer, one can write
z = rx + iy ,
where r = (1, 0) and i = (0, 1). For sums with this representation, one can just sum
the coefficients of r and i like real numbers. The products of complex numbers in this
representation are obtained by real-number-like multiplication and the terms collected into
coefficients of r and i. The products of r and i with themselves and each other follow from
the rules established in the part (c) answer:
r2 = r , ri = i , , i2 = −r .
It makes sense to just replace r by an invisible 1 everywhere since r acts just like 1 in a real-
number-like multiplication and since the i alone suffices to distinguish real and imaginary
parts of the complex number. Thus, we write
z = x + iy
which is, in fact, conventional representation of a complex number. Show that z1 z2 using
the conventional representation agrees with our earlier multiplication rule for complex
numbers.
√
e) i = (0, 1) is evidently the square root of (−1, 0) which in the conventional notation is −1.
But there is another square root for −1. What is it?
f) For several reasons, it turns out to be useful to be useful to define the complex conjugate
of a complex number. The complex conjugate of z is symbolized z ∗ and is defined by
z ∗ = (x + iy)∗ = x − iy .
The ∗ symbol actually means a function that outputs the complex conjugate of the imput
z. Now we define the magnitude (or modulus) of z (symbolized by |z| where the vertical
42 Chapt. 6 Functions of a Complex Variable I
lines are a generalization of the absolute value sign of real numbers) to be the positive real
number given by √
|z| = zz ∗
Find the expression for |z| in terms of the general x and y.
g) What about complex number division you ask? First, we define division of a general
complex number z1 by a general pure real number x2 to be given by
z1 x1 y2
= +i .
x2 x2 x1
No other definition would make much sense: i.e., lead to useful developments as far as one
can see. Second, the only sensible definition for z/z is
z
=1.
z
This is consistent with the real number definition which is certainly a sensible consistency
to maintain. Third, it also seems seems sensible to define
z1 z3 z1 z3
= .
z2 z4 z2 z4
(Remember as Leopold Kronecker [1823–1891] said “God made integers, all else is the work
of man.”) Given the above definitions, determine what z1 /z2 equals in standard form: i.e.,
in the form which clearly consists of a real part plus an imaginary part.
h) What is 1/i in standard format?
006 qfull 01030 1 3 0 easy math: complex conjugation proofs

2. Using general complex numbers z = x + iy, z1 = x1 + iy1 , and z2 = x2 + iy2 , do the following:
a) Prove (z ∗ )∗ = z. This result shows that the complex conjugation function is its own inverse.
Give an example of another function that is its own inverse and an example of one that is
not.
b) Prove (z1 ± z2 )∗ = z1∗ ± z2∗ .
c) Prove (z1 z2 )∗ = z1∗ z2∗ .
d) Prove (z1 z2∗ )∗ = z1∗ z2 .
e) Prove (1/z)∗ = 1/z ∗ and then as a corollary that (z1 /z2 )∗ = z1∗ /z2∗ .
006 qfull 01020 3 3 0 hard math: triangle inequalities

3. The triangle inequalities for complex numbers are

|z1 | − |z2 | ≤ |z1 + z2 | ≤ |z1 | + |z2 | ,
where z1 and z2 are general complex numbers.

a) Prove the inequalities. HINT: In their addition and magnitude properties, complex
numbers are just like 2-dimensional vectors. Let ~c stand for z1 + z2 and ~a and ~b for,
respectively, z1 and z2 . Find the dot product of ~c with itself and proceed as seems fit. There
are at least two other ways to do the proof. The 2nd proof is to assume the inequalities
are true and work toward obviously true statements and the proof is completed by just
following in the steps in reverse (and this reversal of the steps can be left implicit). The
3rd proof is to start from
|z1 + z2 |2 = (z1 + z2 )(z1 + z2 )∗
and use explicitly complex number formalism.

b) Interpret the complex number triangle inequalities in terms of 2-dimensional vectors.
006 qfull 01210 1 3 0 moderate math: RLC loop

4. In treating electrical circuits with potentials that are sinusoidal with time (i.e., AC circuits), it
is expedient to use complex numbers. Say the potential drop across a single current loop is
V = V0 cos(ωt) ,
where V0 is a constant, ω is the angular frequency of the voltage, and t is time. What one does
is imagine an imaginary dual world in which Kirchoff’s voltage and current laws also hold and
where
Vim = V0 sin(ωt) .
The original voltage expression is now
Vre = V0 cos(ωt) ,
where the subscript “re” stands for real. The complex potential is
V = Vre + Vim = V0 eiωt .
The advantage of using this complex potential along with the corresponding complex current
in the current loop is just that it is easier to deal with the function eiωt in solving the current
loop differential equation. One solves for a complex current and the real part is the solution to
the original problem.
Kirchoff’s voltage law states that the sum of potential changes going around a current loop
is zero. This is law is just a special case of rule that applies when a potential can be defined: the
change in potential around a close path is zero. In simple circuit theory, we idealize a current
loop as consisting of elements across which potential drops occurs and ideal wires across which
they don’t. Real wires actually have some potential drop across them. In our case, V stands
for a potential rise that is due to the whole circuit external to the current loop in question: it
could be immensely complex in general, but all we know is V . The potential drops are in the
current loop are due to the elements. The simplest are resistor (with resistance R), inductor
(with inductance L), and capacitor (with capacitance C) which together in the current loop
(which means they are in series in circuit jargon) form an RLC loop as it is called. The drops
across them just add linearly as is consistent with Kirchhoff’s voltage law. For a simple RLC
loop with one of each of resistor, inductor, and capacitor, one has
Q
LI ′ + IR + = V = V0 eiωt ,
C
where Q is the charge on the capacitor and is given by
Z t
Q= I(t′ ) dt′ + Q0 ,
−∞
where Q0 is value at time −∞. For a periodic sinusoidal solution, the charge on the capacitor
is periodic and averages to zero over a period.
44 Chapt. 6 Functions of a Complex Variable I
a) For the given differential equation, solve for the complex current I and define impedance
Z such that ZI = V .
b) What are the magnitude and argument (phase) θ of Z? What are the limits on the phase
θ? Write Z in the polar representation and use that to simplify the expression for I.
c) Determine the real current Ire which is really real in this case and not just mathematically
real.
d) For what ω is the amplitude of the current a maximum? A minimum?
e) How would you characterize the maximum with respect to ω? HINT: Set R and the
driver term V to zero in the complex differential equation and solve for the current. At
what angular frequency does it oscillate?
f) What happens physically when C = 0 and C = ∞.
g) If you have multiple impedances Zi in series (i.e., on the same loop) what is the net
impedance Z of the loop? HINT: Remember Kirchhoff’s voltage law. If you have multiple
impedances Zi in parallel (i.e., on parallel loops with the same V across the loops), what is
the net impedance of the parallel loops collectively? HINT: Remember Kirchhoff’s current
law: the sum currents into a node, equals the sum of currents out a node in a steady state
situation. What are the special cases for series and parallel for resistances, inductances,
and capacitances alone?
006 qfull 02004 2 3 0 moderate math: complex derivative rules

5. The derivative of a complex function (of a complex variable) f (z) is defined analogously to the
case of a real function (of real variable): i.e.,
df f (z + δz) − f (z)
= lim .
dz δz→0 δz
This being the case, there are many differentiation rules for complex functions that are analogous
and analogously proven to those for real function. But there are two special points. First, one
has assume the theorem that the limit of a product of functions equals a product of the limits
of complex functions. Even for real functions, proving this rigorously is tricky I think and needs
compact sets and the like: I could be wrong. Someone has to proven it for complex numbers
too. We will assume that has been done. Second, to complete the proofs one always has to say
something like “since the factors and terms on the right-hand side are continuous, the left-hand
side which is the derivative is continuous and exists.”
a) Prove the chain rule for complex functions of a complex variable: i.e., prove
df [g(z)] df (g) dg(z)
= .
dz dg dz
Assume f and g are continuous and analytic: i.e., they have existing or continuous
derivatives.
b) Prove the product rule for complex function of a complex variable: i.e.,
d(f g) df dg
= g+f .
dz dz dz
Assume f and g are continuous and analytic: i.e., they have existing or continuous
derivatives.
c) The Cauchy-Riemann conditions for a general complex function
f (z) = u(x, y) + iv(x, y)

are
∂u ∂v ∂u ∂v
= and =− .
∂x ∂y ∂y ∂x
If these conditions hold, the partial are themselves continuous, and the function itself exists
(e.g., is not infinite), then df (z)/dz exists (i.e., is continuous) and f (x) is said to be analytic.
Conversely if df (z)/dz is continuous and f (x) exists, then the Cauchy-Riemann conditions
hold. These last two statements are proven by WA-332–333 (with some awkwardness I
might add). Of course, a function is generally only analytic in some places and those are
also the places the Cauchy-Riemann conditions hold.
There three rather special function of the complex variable: z ∗ , Re(z) = x, and
Im(z) = y. Show that these three functions are NOT analytic anywhere. It follows
from the chain rule that functions of these functions will not be analytic either except in
special (but maybe not especially interesting) cases: e.g.,
f (z) = Re(z) + iIm(z) = z .
d) Actually real functions f (x) generalize straightforwardly to being complex functions f (z).
However, in many cases it is non-trivial to separate f (z) into real and imaginary parts:
e.g.,
∞
X zk
ez = ,
k!
k=0
where we take WA-334’s assurance that the series converges. Thus, applying the Cauchy-
Riemann conditions to test for analyticity is NOT straightforward in many cases. On the
other hand, a real function whose derivative exists at least somewhere on the real axis when
generalized to a complex function will (at least in many interesting cases) have a derivative
at least somewhere in the complex plane. Argue why this is so.
Note a generalized function might not exist at some points in the complex plane: e.g.,
1
,
x2 + 1
which is defined everywhere on the real axis, generalizes to
1
z2 + 1
which is undefined, and so non-analytic, at z = ±i.

Chapt. 7 Sturm-Liouville Theory and Orthogonal Functions
009 qmult 01004 1 4 5 easy deducto-memory: Hermitian operator

1. “Let’s play Jeopardy! For $100, the answer is: A mathematical operator A defined by the
PROPERTY A = A† , where the A† in turn is defined by
hα|A|βi = hβ|A† |αi∗ ,
where |αi and |βi are general vectors of a Hilbert (vector) space (which has, of course, an inner
product defined for it).”
What is a/an , Alex?
a) self-adjoint operator b) adjoint operator c) Henrician operator
d) Hermitian conjugate operator e) Hermitian operator
009 qmult 02002 1 1 5 easy memory: Hermitian operator properties

2. A Sturm-Liouville self-adjoint Hermitian operator has the following property/properties:
a) real eigenvalues.
b) orthogonal eigenfunctions, except for those eigenfunctions with degenerate eigenvalues. The
degenerate eigenfunctions (as they are called), however, can always be orthogonalized.
c) a complete set of eigenfunctions.
d) a continuum of eigenvalues.
e) all of the above, except (d).
009 qmult 03002 1 4 2 easy deducto-memory: Gram-Schmidt procedure

3. “Let’s play Jeopardy! For $100, the answer is: It is a PROCEDURE for orthonormalizing
a set of linearly independent vectors where the vectors are of very general sort, but with the
inner-product property among other things. More exactly one can say that the procedure takes
a linearly independent, but not orthonormal, set of vectors and constructs a new orthonormal
set of vectors by linear combinations of the old set.”
What is the procedure, Alex?
a) Sturm-Liouville b) Gram-Schmidt c) Hartree-Fock d) Euler-Lagrange
e) Heimlich
009 qmult 04002 1 4 3 easy deducto-memory: completeness
4. “Let’s play Jeopardy! For $100, the answer is: This PROPERTY possessed by a set of vectors
{|ni} means that any vector |ψi in the space of the set (which is called a Hilbert space which
must have the inner-product property among other things) can be expanded in the set thusly
X
|ψi = |nihn|ψi ,
n
Chapt. 7 Sturm-Liouville Theory and Orthogonal Functions 47
where hn|ψi is the inner product of |ni and |ψi and we have also assumed the set {|ni} is
orthonormalized (which can always be arranged).”
What is , Alex?
a) orthonormality b) cleanness c) completeness d) degeneracy
e) depravity
009 qfull 00520 1 3 0 easy math: normalizable

1. Say we have two general complex functions u and v that are normalizable (or square-integrable)
over the interval [a, b]. This means that inner product of each function with itself exists (i.e.,
does not diverge). Expanding the bra-ket notation for the inner products for our function space,
the self inner products are
Z b Z b Z b Z b
hu|ui = u∗ u dx = |u|2 dx and hv|vi = v ∗ v dx = |v|2 dx .
a a a a
Prove that the inner product

Z b
hv|ui = v ∗ u dx
a
exists. HINT: Write the functions in complex number polar form
u = ru eφu and u = rv eφv ,
where the r’s stand for the magnitudes and the φ’s for the phases of the functions. Both
magnitudes and phases are functions of x in general.
009 qfull 01005 2 3 0 moderate math: Sturm-Liouville weight function
Extra keywords: WA-484–485
2. Recall the 2nd order linear operator form
d2 d
L = p0 + p1 + p2
dx2 dx
(where p0 , p1 , and p0 are general functions of x, except that p0 cannot be zero except at the
boundaries of the x interval of interest) can be transformed to the self-adjoint form

d d
Lself−adjoint = p + q(x)
dx dx
by multiplying by a weight function

R x
exp p1 (t)/p0 (t) dt
w= ,
p0
where
p = wp0 and q = wp2 ,
and where the lower boundary of the integral for w can be left undefined since it just gives a
constant scale factor and w can have any constant scale factor (e.g., 1 or some other value) one
wants for whatever purpose (WA-484–485).
48 Chapt. 7 Sturm-Liouville Theory and Orthogonal Functions
a) What differential equation must w satisfy in order for it to give the transformation? HINT:
Expand Lself−adjoint, compare to wL, and equate what needs to be equated for the two
expressions consistent. Remember both operators are understood to be operating on an
unspecified function to the right. And do NOT assume you know what w is since that is
what you solve for in part (b).
b) Solve the differential equation from the part (a) answer for w. HINT: Get w and its
derivative alone on one side of the differential equation.
c) Show that the expression for w yields 1 if L is itself self-adjoint. HINT: What is p1 if L
is self-adjoint.
009 qfull 01010 3 3 0 tough math: Laguerre equation

3. The Laguerre equation
xy ′′ + (1 − x)y ′ + αy = 0
and is a special case of its big brother the associated Laguerre equation
xy ′′ + (ν + 1 + −x)y ′ + αy = 0 .
(where α and ν are constants) The associated Laguerre equation is immensely important since
it turns up as a transformed version of the radial Schrödinger equation for the hydrogen atom
(and other hydrogenic atoms too) in quantum mechanics. Oddly enough the Laguerre equation
doesn’t turn up in the hydrogen atom solution since ν ≥ 1 in the transformed radial Schrödinger
equation. Nevertheless it is interesting to study the Laguerre equation as simpler warm-up for
the associated Laguerre equation.
We will restrict the x interval of interest to [0, ∞], since this is the interval orver which
the self-adjoint form of the Laguerre operator is a Hermitian operator for a set of its solutions.
This set is the complete set for the Hilbert space of functions normalizable over [0, ∞].
a) Find the asymptotic solution of the Laguerre equation equation for large x. HINT:
Approximate the equation for very large x, solve, and then show that asymptotic solution
validates the approximations made.
b) The asymptotic solution of the Laguerre equation for large x has the nasty property that it
grows exponentially with x in magnitude. Solutions of the Laguerre equation that have this
asymptotic form cannot be normalized with the weight function that turns the Laguerre
operator into a self-adjoint operator, and so can’t form part of the complete set of solution
of solutions of the Laguerre equation for the Hilbert space defined by [0, ∞]. Remember
functions in this Hilbert space are normalizable on this interval. But there are solutions
that don’t grow exponentially in magnitude for certain values of α. These solutions are
polynomial solutions (i.e., finite power series solutions).
Find the α values that yield polynomial solutions by substituting the general power
series
∞
X
y= cℓ xℓ
ℓ=0
into the Laguerre equation and finding the recurrence relation for the coefficients cℓ . Is our
asymptotic solution wrong for these values of α?
c) The polynomial solutions obtainable from the recurrence relation of the part (b) answer
with c0 = 1 are, in fact, the Laguerre polynomials. Calculate the first three Laguerre
polynomials. There are easier ways to generate the Laguerre polynomials (Ar-616, WA-
653).
Chapt. 7 Sturm-Liouville Theory and Orthogonal Functions 49
d) Recall the 2nd order linear operator form
d2 d
L = p0 + p1 + p2
dx2 dx
(where p0 , p1 , and p0 are general functions of x, except that p0 cannot be zero except at
the boundaries of the x interval of interest) can be transformed to the self-adjoint form

d d
Lself−adjoint = p + q(x)
dx dx
by multiplying by a weight function

R x
p1 (x′ )/p0 (x′ ) dx′

exp
w= ,
p0
where
p = wp0 and q = wp2 ,
and where the lower boundary of the integral for w can be left undefined since it just gives
a constant scale factor and w can have any constant scale factor (e.g., 1 or some other
value) one wants for whatever purpose (Ar-425, WA-484–485). Find the explicit w for the
Laguerre (equation) operator and put the Laguerre operator in self-adjoint form.
e) A self-adjoint operator
d d
L= p +q
dx dx
(with p and q being functions in general) is a Hermitian operator for a Hilbert space of
normalizable function vectors defined on interval [a, b] for its set of eigenfunctions {ui } that
satisfy the boundary conditions
x=b
du k

pu∗j =0,

dx

x=a
where uj and uk are general eigenfunctions of the set {ui } (Ar-430). The eigenfunctions
are solutions of the eigen equation
Lu = λwu ,
where lambda is an eigenvalue and w is a weight function. We also require that all Hilbert-
space functions including those eigenfunctions in the set {ui } be normalizable over the
interval where the normalization rule will in general include a weight function w: i.e., we
require for general Hilbert space function f that
Z b
|f |2 w dx
a
exist (i.e., be non-divergent). Show that only the polynomial solutions of Laguerre equation
(and also the self-adjoint Laguerre equation) satisfy the boundary conditions for the interval
[0, ∞].
009 qfull 01012 3 3 0 tough math: associated Laguerre equation
4. The associated Laguerre equation
xy ′′ + (ν + 1 + −x)y ′ + αy = 0
50 Chapt. 7 Sturm-Liouville Theory and Orthogonal Functions
(where α and ν are constants) is immensely important in the solution of the radial part of the
wave function of the hydrogen atom (and other hydrogenic atoms too) in quantum mechanics.
The associated Laguerre equation is, in fact, a transformed version of the radial part of the
Schrödinger equation with x being dimensionless scaled radial coordinate. The hydrogen atom
solution is a solution for the behavior of an electron in the spherical symmetric potential well
provided by the proton nucleus. Spherical polar coordinates are the natural coordinates for the
solution. Since x is a radial coordinate the interval of interest for it is [0, ∞],
a) Find the asymptotic solution of the associated Laguerre equation equation for large x.
HINT: Approximate the equation for very large x, solve, and then show that asymptotic
solution validates the approximations made.
b) The asymptotic solution of the associated Laguerre equation for large x has the nasty
property that it grow exponentially with x in magnitude. The quantum mechanical wave
function of which the associated Laguerre equation solution must be a factor must go to
zero as x goes to infinity as a necessary, but not sufficient, condition to be normalizable.
It turns out that the asymptotic solution grows too strongly with x in magnitude to be
allowed (Gr-152). But there are solutions that don’t grow exponentially in magnitude
for certain values of α. These solutions are polynomial solutions (i.e., finite power series
solutions).
Find the α values that yield polynomial solutions by substituting the general power
series ∞
X
y= cj xj
j=0
into the assciatedLaguerre equation and finding the recurrence relation for the coefficients
cℓ . Is our asymptotic solution wrong for these values of α?
Chapt. 8 Legendre Polynomials and Spherical Harmonics
011 qmult 05531 1 4 2 easy deducto-memory: generating function

1. “Let’s play Jeopardy! For $100, the answer is: It is a function that in many important cases can
be used for relatively easily determining general properties of special sets of functions (which
are often complete sets of solutions of Sturm-Liouville Hermitian operator eigenvalue problems).
There seems to be a bit of black magic though in finding the function for particular cases.”
What is a , Alex?
a) general function b) generating function c) generic function
d) genitive function e) genuine function
011 qmult 05551 1 1 3 easy memory: uniqueness theorem and Leg. poly.
2. It using the generating function to determine properties of the Legendre polynomials, one
frequently makes use of the uniqueness theorem of
a) the harmonic series. b) the geometric series. c) power series.
d) the Legendre series. e) the world series.
011 qmult 05571 1 4 5 easy deducto-memory: [n/2] function
Extra keywords: Really one could just int(n/2) or floor(n).
3. “Let’s play Jeopardy! For $100, the answer is: It is an integer function used among other things
in the general power series formula for the Legendre polynomials.”
What is , Alex?
a) (n) which equals (n − 1)/2. b) (n) which equals n/2. c) ceiling(n/2).
d) [n/2] which equals n/2 for n even and (n + 1) for n odd. e) [n/2] which equals
n/2 for n even and (n − 1)/2 for n odd.
011 qmult 05631 1 3 1 easy math: Leg. poly. recurrence relation
4. Given the Legendre polynomial recurrence relation
(n + 1)Pn+1 (x) = (2n + 1)xPn (x) − nPn−1 (x)
and P0 = 1, find P1 (x) and P2 (x).
a) P1 = x and P2 (x) = (1/2)(3x2 − 1). b) P1 = 2x and P2 (x) = 4x2 − 2.
c) P1 = −x + 1 and P2 (x) = (1/2!)(x2 − 4x + 2). d) P1 = x and P2 (x) = x2 .
e) P1 = 1/x and P2 (x) = 1/x2 .
011 qmult 05651 1 1 5 easy memory: Leg. poly. properties
5. As the non-degenerate solutions of the Sturm-Liouville Hermitian operator eigenvalue problem
for the interval [−1, 1], the Legendre polynomials:
a) form a complete set for [−1, 1]. b) are orthogonal for [−1, 1]. c) have real
associated eigenvalues. d) are normalized as obtained from the generating function.
e) all of the above, except (d).
52 Chapt. 8 Legendre Polynomials and Spherical Harmonics
011 qmult 05791 1 4 4 easy deducto-memory: Rodriques’s formula

6. “Let’s play Jeopardy! For $100, the answer is: It is an altnerative definition of the Legendre
polynomials.”
What is , Alex?
a) Borracho’s formula b) Gomez’s formula c) Morales’s formula
d) Rodrigues’s formula e) Ruiz’s formula
011 qmult 05841 1 4 2 easy deducto-memory: spherical harmonics
7. “Let’s play Jeopardy! For $100, the answer is: They form the standard complete orthonormal
set for the 2-dimensional spherical surface subspace of the 3-dimensional Euclidean space.”
What are the , Alex?
a) harmonies of the spheres b) spherical harmonics c) Harmonices Mundi
d) associated Legendre functions e) Legendre polynomials
011 qmult 05851 1 1 1 easy memory: memorable spherical harmonic
8. The one spherical harmonic that everyone can remember is
√
a) Y0,0 = 1/ 4π. b) Y0,0 = 4π. c) Y1,0 = cos θ. d) Y1,0 = sin θ.
e) Y0,m = eimφ .
011 qfull 00110 2 3 0 moderate math: Legendre generating function

1. The Legendre polynomial generating function is
∞
X
2 −1/2
g(x, t) = (1 − 2xt + t ) = Pn (x)tn ,
n=0
where the Legendre polynomials Pn (x) come from arranging the infinite sum as a power series
in t (WA-553). The series is absolutely convergent for |t| < 1 provided the |Pn (x)| have a finite
upper bound which they do. This follows because the geometric series
∞
X 1
xn =
n=0
1−x
is absolutely convergent for |x| < 1 (WA-258). First, we note that if the terms un of a series
obey |un | ≤ |an |, then the un series absolutely converges if the an series does (WA-262,271). If
U is the upper bound of the |Pn (x)|, then |Pn tn | ≤ U tn and the generating function absolutely
converges for |t| < 1 since the U tn is just the geometric series times a constant U . (Actually
U = 1 [WA-566].) The absolute convergence of the generating function for |t| < 1 is useful
because it means the series converges for any ordering of the terms (WA-271) which is a property
often needed in making use of the generating function.
We will make use of the generating function to prove a number of general results about the
Legendre polynomials.
a) Show that
∞
X (2n)! n
(1 + x)−1/2 = (−1)n x .
n=0
22n (n!)2
Chapt. 8 Legendre Polynomials and Spherical Harmonics 53
HINT: It helps to make use of the double factorials:


 2n · (2n − 2) · (2n − 4) · . . . · 4 · 2 for n ≥ 1;
(2n)!! ≡ 1 for n = 0;
 n
2 n! in general,
(2n − 1)!! ≡ (2n − 1) · (2n − 3) · (2n − 5) · . . . · 3 · 1 for n ≥ 1;

n
1 for n = 0.
b) Use the generating function to prove
Pn (±1) = (±1)n
or
Pn (1) = 1 and Pn (−1) = (−1)n .
HINT: Use the uniqueness of power series (WA-292).
c) Prove
Pn (−x) = (−1)n Pn (x) :
i.e., prove that the even order Pn are even functions and the odd order ones are odd
functions. HINT: Consider g(−x, −t) and use the uniqueness of power series (WA-292).
d) Find the expressions for Pn (0). HINT: Use g(0, t), the part (a) identity, and the uniqueness
of power series (WA-292).
011 qfull 00240 1 3 0 easy math: satisfying the Legendre equation

2. Legendre’s differential equation is
(1 − x2 )y ′′ − 2xy ′ + n(n + 1)y = 0 .
a) Prove that the Legendre operator
d2 d
L = (1 − x2 ) − 2x
dx2 dx
is a Sturm-Liouville self-adjoint operator for the interval [−1, 1].
b)
011 qfull 00250 2 3 0 moderate math: Legendre orthogonality
Extra keywords: Legendre inner product
3. Because the Legendre polynomials are the non-degenerate eigensolutions of a Sturm-Liouville
Hermitian operator eigenvalue problem for interval [−1, 1], they are guaranteed to have real
eigenvalues, be orthogonal for [−1, 1], and form a complete set for [−1, 1] (WA-496). But
Legendre polynomials as obtained from the generating function
∞
X
g(x, t) = (1 − 2xt + t2 )−1/2 = Pn (x)tn
n=0
are not normalized for [−1, 1]. In expanding functions in the Legendre polynomials, normalized
versions are needed. Here we show how to use the generating function to find the general
normalization constant.
a) Integrate the square of the generating function
g(x, t) = (1 − 2xt + t2 )−1/2

54 Chapt. 8 Legendre Polynomials and Spherical Harmonics
over [−1, 1].

b) Expand
1 1+t
ln
t 1−t
in a Taylor’s series about t = 0. For what values of t is the series absolutely convergent?
HINT: It is easier to expand ln(1 + t) and then obtain the required series.
c) Integrate the square of the generating function series
∞
X
g(x, t) = Pn (x)tn
n=0
over [−1, 1] to obtain a series with coefficients that are the inner product hPn |Pn i.
d) Using the uniqueness of power series theorem (WA-292), determine the values of hPn |Pn i
and thus the general normalization constant of the Legendre polynomials.
011 qfull 00650 1 3 0 easy math: sph. har. expansion

4. Recall that the spherical harmonics form a complete set for 2-dimensional angular subspace of
the Euclidean 3-dimensional space. This means that any piecewise continuous non-divergent
angular function can be expanded in the spherical harmonics. Consider following function
expanded in spherical harmonics:
ℓ
∞ X
X
f (r, θ, φ) = aℓ,m rℓ Yℓm .
ℓ=0 m=−ℓ
Recall the definition of Yℓm :

s
2ℓ + 1 (ℓ − m)! m
Yℓm = (−1)m P (cos θ)eimφ ,
4π (ℓ + m)! ℓ
where Pℓm (cos θ) is an associated Legendre function and eimφ is an azimuthal eigenfunction
(WA-584).
a) Find the azimuthal-angle averaged value of f (r, θ, φ): i.e., hf (r, θ, φ)iφ . HINT: Make use
of orthogonality.
b) Now find the full angle averaged value of f (r, θ, φ) i.e., hf (r, θ, φ)iθ,φ . HINT: Make use of
orthogonality. Recall the inner product relation for pairs of Legendre polynomials:
1
2δmn
Z
hPm |Pn i = Pm (x)Pn (x) dx = .
−1 2n + 1
c) What is f (0, θ, φ)?

Appendix 1 Introductory Physics Equation Sheet
Note: This equation sheet is intended for students writing tests or reviewing material. Therefore it
is neither intended to be complete nor completely explicit. There are fewer symbols than variables,
and so some symbols must be used for different things: context must distinguish.
56 Appendix 1 Introductory Physics Equation Sheet
Equation Sheet for Physical Sciences Courses

The equations are mnemonic. Students are expected to understand how to interpret and use them.
Usually, non-vector forms have been presented: i.e., forms suitable for one-dimensional calculations.
1 Geometry
4 3
Ccir = 2πr Acir = πr2 Asph = 4πr2 Vsph = πr
3
c2 = a 2 + b 2 Pyth.Thm.
2 Kinematics
dfinal − dinitial vfinal − vinitial
d = vt vave = v = at aave =
t t
Amount v2
Amount = Constant Rate × time time = acentripetal =
Constant Rate r
3 Dynamics
mv 2
Fnet = ma 1 N ≈ 0.225 lb Fcentripetal = p = mv
r
4 Gravity
r r
Gm1 m2 GM 2GM
F = Fg = mg vcircular = vescape =
r2 r r
G = 6.6742 × 10−11 MKS units (circa 2002)

g = 9.80 m/s2 (latitude range ∼ 9.78030–9.8322 m/s2 [CAC-72])
5 Energy and Work
W 1
W = Fd 1J = 1N·m P = KE = mv 2 P Egravity = mgy
t 2
c = 2.99792458 × 108 m/s ≈ 2.998 × 108 m/s ≈ 3 × 108 m/s

p
E = mc2 Erest = mrest c2 ∆tproper = ∆t 1 − (v/c)2
6 Thermodynamics and Buoyancy
9
Tabsolute = TCelsius + 273.15 TFahrenheit = TCelsius + 32 ∆Q = Cspecificm∆T
5
m N F
ρ= n= p= p = psurface + ρgy Fbuoyant = mdis g = ρfluidVdis g
V V A
ρfluid Vdis = mfloating P V = N kT k = 1.3806505 × 10−23 J/K
Wdone Tcold
ε= εupperlimit = 1 −
Qabsorbed Thot
Appendix 1 Introductory Physics Equation Sheet 57
7 Electricity and Magnetism

kQ1 Q2
F = k = 8.99 × 109 N m2 /C2 e = 1.60217733 × 10−19 C
r12
coulomb (C)
1 ampere (A) = 1
second (s)
joule (J) volt (V)
1 volt (V) = 1 1 ohm (Ω) = 1
coulomb (C) ampere (A)
X X
∆Vrise = ∆Vdrop V = IR P =VI
8 Waves
λn 2L v
v = fλ p = 1/f n =L λn = fn = n
2 n 2L
vsound 20◦ C 1 atm = 343 m/s vsound 0◦ C 1 atm = 331 m/s
9 Nuclear Physics
A N0
ZX n(t) = t/t1/2
1 amu = 931.494043 MeV
2
10 Quantum Mechanics
h h
h = 6.6260693 × 10−34 J s me = 9.1093826 × 10−31 kg E = hf λ= =
p mv
ih ∂Ψ
KEphotoelectron = hf − w HΨ =
2π ∂t
11 Astronomy
km/s
v = Hd H = 71+4
−3 (circa 2004)
Mpc
12 Geometrical Formulae
4 3
Ccir = 2πr Acir = πr2 Asph = 4πr2 Vsph = πr
3
13 Trigonometry Formulae
1 1
x = h cos θ y = h sin θ cos2 θ = [1 + cos(2θ)] sin2 θ = [1 − cos(2θ)]
2 2
sin(2θ) = 2 sin(θ) cos(θ)
sin(a + b) = sin(a) cos(b) + cos(a) sin(b) cos(a + b) = cos(a) cos(b) − sin(a) sin(b)
14 Approximation Formulae
∆f df 1
≈ ≈ 1 + x : (x << 1)
∆x dx 1−x
1
sin θ ≈ θ tan θ ≈ θ cos θ ≈ 1 − θ2 all for θ << 1
2
15 Quadratic Formula
√ s
2
−b ± b2 − 4ac b b c
If 0 = ax2 + bx + c , then x= =− ± −
2a 2a 2a a
16 Vector Formulae
~a + ~b = (ax + bx , ay + by , az + bz ) ~a · ~b = ab cos θ = ax bx + ay by + az bz
~c = ~a × ~b = ab sin(θ)ĉ = (ay bz − by az , az bx − bz ax , ax by − bx ay )
17 Differentiation and Integration Formulae
d(xn ) d(x0 ) d[ln(x)] 1

= nxn−1 except for n = 0; =0 =
dx dx dx x
b
dF (x)
Z
f (x) dx = F (x)|ba = F (b) − F (a) where = f (x)
a dx
xn+1 1
Z Z
xn dx = except for n = −1; dx = ln(x)
n+1 x
18 One-Dimensional Kinematics
∆x dx ∆v dv d2 x
vavg = v= aavg = a= = 2
∆t dt ∆t dt dt
1
v = v0 + at x = x0 + v0 t + at2
2
1
x = x0 + (v0 + v)t v 2 = v02 + 2a(x − x0 ) g = 9.8 m/s2
2
x′ = x − vframe t v ′ = v − vframe a′ = a
19 Two- and Three-Dimensional Kinematics: General
∆~r d~r ∆~v d~v d2~r

~vavg = ~v = ~aavg = ~a = = 2
∆t dt ∆t dt dt
v2 v2
~acentripetal = (−r̂) acentripetal =
r r
20 Projectile Motion
1
x = vx,0 t y = y0 + vy,0 t − gt2 vx,0 = v0 cos θ vy,0 = v0 sin θ
2
x x x2 g
t= = y = y0 + x tan θ −
vx,0 v0 cos θ 2v02 cos2 θ
v02 sin θ cos θ v02 sin2 θ

xfor max = ymax = y0 +
g 2g
p " s #
tan θ ± tan2 θ − 2g(y − y0 )/(v02 cos2 θ) v02 sin θ cos θ 2g(y − y0 )
xgen = = 1± 1− 2 2
g/(v02 cos2 θ) g v0 sin θ
2v02 sin θ cos θ v 2 sin(2θ) π v02

x(y = y0 ) = = 0 θfor max/min = ± xmax (y = y0 ) =
g g 4 g
s s
2(y0 − y) 2(y0 − y)
x(θ = 0) = ±v0 t(θ = 0) =
g g
21 Very Basic Classical Mechanics
F~net = m~a F~opp = −F~ Fg = mg g = 9.8 m/s2 F = −kx
Ff static = min[Fapplied , Ff static max ] Ff static max = µstatic FN Ff kinetic = µkinetic FN
dθ dω d2 θ
vtangential = rω = r atangential = rα = r =r 2
dt dt dt
v2 v 2
~acentripetal = (−r̂) F~centripetal = m (−r̂)
r r
22 Work and Energy
1
Z
dW = F~ · d~r W = F~ · d~r KE = mv 2 Emechanical = KE + U
2
dW W
P = Pavg = P = F~ · ~v
dt ∆t
KEf = KEi + Wnet ∆Uof a conservative force = −Wby a conservative force Ef = Ei + Wnonconservative
dU 1 2
F =− F~ = −∇U U= kx U = mgy
dx 2
23 Systems of Particles
R
ρ(~r )~r dV
P P
mi~ri msub~rcm sub
~rcm = i
= sub
~rcm = V
F~net ext = m~acm
mtotal mtotal mtotal
d~
p d~
ptotal
p~ = m~v F~net = m F~net ext = m
dt dt
dm dm
m~a = F~net non-transferred + (~vtransferred − ~v ) = F~net non-transferred + ~vrel
dt dt
m
i
v = vi + vthrust ln rocket in free space
m
24 Collisions
tf
J~ p~1 + ~p2
Z
J~ = F~ (t) dt F~avg = ~p1i + ~p2i = ~p1f + p~2f ~vcm =
ti ∆t mtotal

m1 − m2 2m2
KEtotal f = KEtotal i v1f = v1i + v2i vrel f = −vrel i
m1 + m2 m1 + m2
1-d Elastic Expressions
∞
1
Z
Pperfect gas pressure = pvn(v) dp Pideal gas pressure = nkT
3 0
25 Rotation
s dθ v d2 θ dω a ω 2π
θ= ω= = α= = = f= P = f −1 =
r dt r dt2 dt r 2π ω
1 1
ω = ω0 + αt θ = θ0 + ω0 t + αt2 θ = θ0 + (ω0 + ω)t ω 2 = ω02 + 2α(θ − θ0 )
2 2
~
dL
~τ = ~r × F~ ~τnet = ~ = ~r × p~
L
dt
τ = rF sin θ τnet = Iα L = Iω
X Z
2 2 2
I= mi rxy,i I= ρrxy dV Ipar-axis = mrxy,cm + Icm
i
g sin θ
a=
1 + I/(mr2 )
1 2 1 1
KErot = Iω KEtotal = KEtrans + KErot = mv 2 + Iω 2 dW = τ dθ P = τω
2 2 2
Z Z
∆KErot = Wnet = τnet dθ ∆U = −W = − τ dθ Ef = Ei + Wnonconservative
26 Static Equilibrium in Two Dimensions

X X X
0 = Fnet x = Fx 0 = Fnet y = Fy 0 = τnet = τ
27 Gravity
Gm1 m2
G = 6.67407 × 10−11 m3 kg−1 s−2 (2002 result) F~1 on 2 = 2 (−r̂12 )
r12
GM
I
f~g = 2 (−r̂) f~g · dA
~ = −4πGM
r
r r
Gm1 m2 GM 2GM GM
U =− V =− vescape = vorbit =
r12 r r r
4π 2

2 3 2π dA 1 L
P = r P = √ r3/2 = r2 ω = = Constant
GM GM dt 2 2m
28 Fluids
∆m F
ρ= p= p = p0 + ρgddepth
∆V A
Pascal’s principle p = pext − ρg(y − yext ) ∆p = ∆pext

Archimedes principle Fbuoy = mfluid dis g = Vfluid dis ρfluid g
equation of continuity for ideal fluid RV = Av = Constant
1
Bernoulli’s equation p + ρv 2 + ρgy = Constant
2
29 Simple Harmonic Oscillator

1 2 k
P = f −1 ω = 2πf F = −kx U= kx a(t) = − x(t) = −ω 2 x(t)
2 m
r r
k m
ω= P = 2π x(t) = A cos(ωt) + sin(ωt)
m k
1 1 1 1
Emec total = mv 2 = kx2max = mv 2 + kx2
2 max 2 2 2
s s
I ℓ
P = 2π P = 2π
mgℓ g
30 Waves
s
d2 y 1 d2 y FT
= v= y = f (x ∓ vt) y = ymax sin[k(x ∓ vt)] = ymax sin(kx ∓ ωt)
dx2 v 2 dt2 µ
1 2π ω 2
Period = k= v = fλ = P ∝ ymax
f λ k
L λ 2L v
y = 2ymax sin(kx) cos(ωt) n= L=n λ= f =n
λ/2 2 n 2L
s
∂P 1
v= nλ = d sin(θ) n+ λ = d sin(θ)
∂ρ S 2

P I
I= β = (10 dB) × log
4πr2 I0
v f0
f =n : n = 1, 3, 5, . . . fmedium =
4L 1 − v0 /vmedium
31 Thermodynamics
TK = TC + 273.15 K TF = 1.8TC + 32◦ ∆L = Lα∆T ∆V = V β∆T β = 3α
X
Q = mc∆T = mc(Tf − Ti ) Q= mk ck (Tf − Ti,k )
k
Z Vf
W = p dV dE = dQ − dW dE = T dS − p dV
Vi
dT 4 4
Fcond = −k Fsurface = ε1 σTsurface Fenv = ε2 σTenv σ = 5.67 × 10−8 W m−2 K−4
dx
pV = nRT = N kT R = 8.31 J mol−1 K−1 k = 1.38 × 10−23 J/K
f f
Wisothermal = nRT ln(Vf /Vi ) E= nRT = nCV T CV = R Cp = CV + R
2 2
Cp CV + R 2
P V γ = Constant γ= = =1+
CV CV f
W Qc Qc W
0 = ∆E = (Qh − Qc ) − W ε= =1− η= =1−
Qh Qh Qh Qh
Tc Tc
εCarnot = 1 − ηCarnot = = 1 − εCarnot εCarnot + ηCarnot = 1
Th Th
" f /2 #
f f
dQ Tf Vf
Z Z
∆S = dS = ∆S = nR ln
i i T Ti Vi
32 Electrostatics
e = 1.602 × 10−19 C me = 9.109 × 10−31 kg
1
ε0 = 8.854 × 10−12 C2 /(N-m2 ) ≈ 10−11 k= = 8.99 × 109 N-m2 /C2 ≈ 1010
4πε0
kq1 q2 ~ r ) = kq r̂ = q r̂ ~ = qenc
I
F~1 on 2 = 2 r̂1,2 F~ = q E
~ E(~ ~ · dA
E
r1,2 r2 4πεr2 ε0
~ = σ n̂
E ~
~τ = ~p × E P E = −~ ~
p·E
ε0
f
kq X qi dq
Z Z
∆U = q∆V ∆V = − ~ s
E·d~ V = V =k V =k ~ = −∇V
E
i r i
ri r
q ε0 A X 1 X 1
C= C= Cparallel = Ci =
V d i
Cseries i
Ci
q2 1 ~ = qenc
I
UC = = CV 2 Cdielectric = κCvacuum ~ macroscopic charge = κE
E ~ net ~ · dA
κE
2C 2 ε0
33 Current and Circuits
dq I
Z
I= I= J~ · dA
~ J= J~ = nq~vdrift J~ = σ E
~ V = IR
dt A
nq 2 τ 1 L V2
σ= ρ= V = IR R=ρ P = IV P = I 2R =
m σ A R
node loop
X X X 1 X 1
0= Ii 0= ∆Vi Rseries = Ri =
i i i
Rparallel i
Ri
∆Vinitial −t/τ
τ = RC I= e ∆V = ∆Vfinal 1 − e−t/τ
R
34 Magnetic Fields and Forces
µ0 = 4π × 10−7 T-m/A 1 tesla (T) = 104 gauss (G)
F~ = q~v × B
~ ~ = I d~ℓ × B
dF ~
mv qB qB 2πm
~rcyclotron = ωcyclotron = fcyclotron = P =
qB m 2πm qB
µ
~ = N IAÂ ~
τ = ~µ × B ~
U = −~µ · B
~ = µ0 I d~s × r̂ µ0 I µ0 Iθ µ0 ℓI1 I2
dB Bwire = Barc = F~1 on 2 = r̂2 to 1
4π r2 2πr 4πr 2πr
µ0 N I ~ circ. loop = µ0 IR2 ~ dipole = µ0 µ

~
Bsolenoid = µ0 nI Btoroid = B ẑ B
2πr 2(R2 + z 2 )3/2 2πz 3
~ ·d~s = − dΦB = − d
I Z I Z
~ ·d~s = µ0 I
B ΦB = ~ A
B·d ~ Vemf = E ~ ·dA
B ~
linked area dt dt linked area
ΦB dI 1 2 B2
L= Lsolenoid = µ0 n2 Vvol Vemf drop = L UL = LI uB =
I dt 2 2µ0
35 LR, LC, LRC Circuits and Alternating Current
L Vemf
τL = I= 1 − e−t/τL I = I0 e−t/τL
R R
1 Vmax
ω=√ Imax = q
LC R2 + [wd L − 1/(wd C)]
2
36 Electromagnetic Waves and Radiation
1 ω 2π 1
c = 2.99792458 × 108 ≈ 3.00 × 108 m/s Period = c = fλ = k= c= √
f k λ µ0 ε0
~ =E
~ 0 sin(kx − wt) ~ =B
~ 0 sin(kx − wt) B 1
E B =
E c
1 2 2 L
Imonochr = εE = εErms Ipoint =
2 0 4πr2
Ωsphere = 4π dΩ = sin θ dθ dφ
1
Itrans. polarized = Iinc cos2 θ Itrans. non-polarized = Iinc
2
37 Geometrical Optics

−1 n2
θinc = θrefl n1 sin θ1 = n2 sin θ2 θ1,critical = sin
n1
c λvacuum vn c
n= λn = f= =
vn n λn λvacuum
38 Interference and Diffraction
4I0 d
2-slit nλ = d sin θmax I= cos2 (α) α=π sin θ
r2 λ
I0 sin2 (α) a
single-slit nλ = a sin θzero I= α=π sin θ
r2 α2 λ
grating nλ = d sin θmax X-ray nλ = 2d sin θmax
39 Special Relativity
c = 2.99792458 × 108 m/s ≈ 2.998 × 108 m/s ≈ 3 × 108 m/s ≈ 1 lyr/yr ≈ 1 ft/ns
v 1 1
β= γ=p γ(β << 1) = 1 + β 2 τ = ct
c 1 − β2 2
Galilean Transformations Lorentz Transformations

′
x = x − βτ x′ = γ(x − βτ )
y′ = y y′ = y
z ′′ = z z′ = z
τ =τ τ ′ = γ(τ − βx)
′ ′ βobj − β
βobj = βobj − β βobj =
1 − ββobj
p p
ℓ = ℓproper 1 − β2 ∆τproper = ∆τ 1 − β2
s s
xintersection 1 − β2 τintersection 1 − β2
x′ = = x′x scale τ′ = = ττ′ scale θMink = tan−1 (β)
γ 1 + β2 γ 1 + β2
m = γm0 p = mv = γm0 cβ E0 = m0 c2 E = γE0 = γm0 c2 = mc2
E = mc2 (pc)2 + (m0 c2 )2 − m0 c2 = (γ − 1)m0 c2

p p
E= (pc)2 + (m0 c2 )2 KE = E − E0 =
s
1−β 1 2
f = fproper for source and detector separating f (β << 1) = fproper 1 − β + β
1+β 2

p 1 2
ftrans = fproper 1 − β 2 ftrans (β << 1) = fproper 1 − β
2
40 Quantum Mechanics
− h h − h
h= E = hf p= = hk ∆λ = (1 − cos θ)
2π λ mc
− 2 −2
h ∂2Ψ ∂Ψ h ∂2ψ −
− 2
+ V Ψ = i−
h − + V ψ = Eψ Ψ(x, t) = ψ(x)e−iEt/ h
2m ∂x ∂t 2m ∂x2
−
h
Z ∞
1 Z2 Z2
∆x∆p ≥ |Ψ|2 dx = 1 En = − me c2 α2 2 = −Eryd 2
2 −∞ 2 n n
e2 1
Eryd = 13.6056981 eV α= =
4πǫ0 −
hc 137.0359895
aBohr = 0.529177249 × 10−10 = 0.529177249 Å
mamu = 1.6605402 × 10−27 kg = 931.49432 MeV me = 9.1093897 × 10−31 kg = 0.510999906 MeV
mp = 1.6726231 × 10−27 kg = 938.27231 MeV mn = 1.6749286 × 10−27 kg = 939.56563 MeV

Appendix 2 Table of Integrals, Etc.
Note: There are no guarantees of accuracy with these integrals, etc. I include some derivatives at
the start, but they are such a minor component that they do not merit a change in title.
1. Derivatives
d 1 du h π πi
(1) sin−1 u = √ for sin−1 ∈ − ,
dx 1 − u dx
2 2 2
d −1 du
(2) cos−1 u = √ for cos−1 ∈ [0, π]
dx 1 − u2 dx
2. Functions containing ax2 + b
 r
1 −1 a

 √ tan x a > 0 and b > 0;
b


 ab

 √ √
1 x a − −b


 √ √

 ln √ a > 0 and b < 0;
 2 −ab x a + −b



Z
dx  √ √ !
(1) = 1 b + x −a
ax2 + b  √ ln √ √ a < 0 and b > 0;



 2 −ab b − x −a


 1
− b = 0;


 ax


x


a=0

b
r
dx 1 x 1 1 a
Z
−1
(2) = + √ tan x a > 0 and b > 0
(ax2 + b)2 2b (ax2 + b) 2b ab b
r r
dx 1 a 1 a
Z
−1 −1
(3) √ = √ sin x − = −√ cos x − a<0
ax2 + b −a b −a b
3. Functions containing ax2 + bx + c
dx 2(2ax + b)
Z
(1) 3/2
=− √
(ax2 + bx + c) (b2 − 4ac) ax2 + bx + c
4. Functions containing sin(ax)

68 Appendix 2 Table of Integrals, Etc.
1 1
Z
(1) sin3 (ax) dx = − cos(ax) + cos3 (ax)
a 3a
1 4 4
Z
(2) sin5 (ax) dx = − sin4 (ax) cos(ax) + cos3 (ax) − cos(ax)
5a 15a 5a
1 n−1
Z Z
(3) sinn (ax) dx = − sinn−1 (ax) cos(ax) sinn−2 (ax) dx
na n
5. Inverse Trigonometric Functions
1p
Z
(1) sin−1 (ax) dx = x sin−1 (ax) + 1 − a2 x2
a
1p
Z
(2) cos−1 (ax) dx = x cos−1 (ax) − 1 − a2 x2
a
6. Algebraic and Trigonometric Functions
1 x
Z
(1) x sin(ax) dx = 2
sin(ax) − cos(ax)
a a
x2 2x 2
Z
(2) x2 sin(ax) dx = − cos(ax) + 2 sin(ax) + 3 cos(ax)
a a a
1 x
Z
(3) x cos(ax) dx = cos(ax) − sin(ax)
a2 a
x2 2x 2
Z
(4) x2 sin(ax) dx = sin(ax) + 2 cos(ax) − 3 sin(ax)
a a a
7. Gaussian Function and Factorial Integral
(x − µ)2

1
(1) G(x) = √ exp − standard form
σ 2π 2σ 2
∞ r
π
Z
2
(2) e−(a+ib)x dx =
−∞ a + ib
Appendix 2 Table of Integrals, Etc. 69
∞ √
π 1
Z
2
(3) x2 e−(a+ib)x dx =
−∞ 2 (a + ib)3/2
Z ∞
(4) z! = e−t tz dt
0
√

1 z!
(5) 0! = 1 1! = 1 − != π (z − 1)! =
2 z
n
!
x
xℓ
Z X
−t n −x
(6) g(n, x) = e t dt = n! 1 − e
0 ℓ!
ℓ=0
(7) g(0, x) = 1 − e−x
(8) g(1, x) = 1 − e−x (1 + x)

1
(9) g(2, x) = 2 1 − e−x 1 + x + x2
2
Appendix 3 Multiple-Choice Problem Answer Tables
Note: For those who find scantrons frequently inaccurate and prefer to have their own table and
marking template, the following are provided. I got the template trick from Neil Huffacker at
University of Oklahoma. One just punches out the right answer places on an answer table and
overlays it on student answer tables and quickly identifies and marks the wrong answers
Answer Table for the Multiple-Choice Questions

a b c d e a b c d e
1. O O O O O 6. O O O O O
5. O O O O O 10. O O O O O
Appendix 3 Multiple-Choice Problem Answer Tables 71

a b c d e a b c d e
1. O O O O O 11. O O O O O
2. O O O O O 12. O O O O O
3. O O O O O 13. O O O O O
4. O O O O O 14. O O O O O
5. O O O O O 15. O O O O O
6. O O O O O 16. O O O O O
7. O O O O O 17. O O O O O
8. O O O O O 18. O O O O O
9. O O O O O 19. O O O O O
10. O O O O O 20. O O O O O
72 Appendix 3 Multiple-Choice Problem Answer Tables

a b c d e a b c d e
1. O O O O O 16. O O O O O
2. O O O O O 17. O O O O O
3. O O O O O 18. O O O O O
4. O O O O O 19. O O O O O
5. O O O O O 20. O O O O O
6. O O O O O 21. O O O O O
7. O O O O O 22. O O O O O
8. O O O O O 23. O O O O O
9. O O O O O 24. O O O O O
10. O O O O O 25. O O O O O
11. O O O O O 26. O O O O O
12. O O O O O 27. O O O O O
13. O O O O O 28. O O O O O
14. O O O O O 29. O O O O O
15. O O O O O 30. O O O O O
Appendix 3 Multiple-Choice Problem Answer Tables 73
NAME:
a b c d e a b c d e
1. O O O O O 26. O O O O O
2. O O O O O 27. O O O O O
3. O O O O O 28. O O O O O
4. O O O O O 29. O O O O O
5. O O O O O 30. O O O O O
6. O O O O O 31. O O O O O
7. O O O O O 32. O O O O O
8. O O O O O 33. O O O O O
9. O O O O O 34. O O O O O
10. O O O O O 35. O O O O O
11. O O O O O 36. O O O O O
12. O O O O O 37. O O O O O
13. O O O O O 38. O O O O O
14. O O O O O 39. O O O O O
15. O O O O O 40. O O O O O
16. O O O O O 41. O O O O O
17. O O O O O 42. O O O O O
18. O O O O O 43. O O O O O
19. O O O O O 44. O O O O O
20. O O O O O 45. O O O O O
21. O O O O O 46. O O O O O
22. O O O O O 47. O O O O O
23. O O O O O 48. O O O O O
24. O O O O O 49. O O O O O
25. O O O O O 50. O O O O O
74 Appendix 3 Multiple-Choice Problem Answer Tables
Answer Table Name:

a b c d e a b c d e
1. O O O O O 31. O O O O O
2. O O O O O 32. O O O O O
3. O O O O O 33. O O O O O
4. O O O O O 34. O O O O O
5. O O O O O 35. O O O O O
6. O O O O O 36. O O O O O
7. O O O O O 37. O O O O O
8. O O O O O 38. O O O O O
9. O O O O O 39. O O O O O
10. O O O O O 40. O O O O O
11. O O O O O 41. O O O O O
12. O O O O O 42. O O O O O
13. O O O O O 43. O O O O O
14. O O O O O 44. O O O O O
15. O O O O O 45. O O O O O
16. O O O O O 46. O O O O O
17. O O O O O 47. O O O O O
18. O O O O O 48. O O O O O
19. O O O O O 49. O O O O O
20. O O O O O 50. O O O O O
21. O O O O O 51. O O O O O
22. O O O O O 52. O O O O O
23. O O O O O 53. O O O O O
24. O O O O O 54. O O O O O
25. O O O O O 55. O O O O O
26. O O O O O 56. O O O O O
27. O O O O O 57. O O O O O
28. O O O O O 58. O O O O O
29. O O O O O 59. O O O O O
30. O O O O O 60. O O O O O

Problem

Uploaded by

Copyright:

Available Formats

Problem

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Problem

Uploaded by

Copyright:

Available Formats

Mathematical Physics Problems

Portpentagram Publishing (self-published)

001 qmult 00100 1 4 5 easy deducto-memory: seven samurai

001 qmult 00202 1 4 4 easy deducto-memory: Einstein summation

001 qmult 00350 1 4 2 easy deducto-memory: Levi-Civita symbol identity

δiℓ δjm − δim δjℓ .

001 qmult 00410 1 1 3 easy memory: triple scalar product

001 qmult 00530 1 4 1 easy deducto-memory: gradient does what

001 qmult 00610 1 4 2 easy deducto-memory: divergence

001 qmult 01110 1 1 1 easy memory: Gauss’s theorem stated

001 qmult 01220 2 1 5 mod memory: potential implications

001 qmult 01310 1 4 4 easy deducto-memory: Gauss’s law ingredient

~ is the electric field, q is the charge enclosed inside the closed

001 qmult 01320 1 4 5 easy deducto-memory: Gauss’s law symmetries

000 qmult 01430 1 4 2 easy deducto-memory: Dirac delta use

001 qfull 00210 1 3 0 easy math: Schwarz inequality

001 qfull 00220 1 3 0 easy math: Einstein summation rule

Aij δij = Aii .

001 qfull 00302 1 3 0 easy math: general symmetry identity

001 qfull 00315 1 3 0 easy math: dot product identity

b) Prove that the last result implies

001 qfull 00320 2 3 0 moderate math: law of sines

001 qfull 00390 3 3 0 tough math: Levi-Civita symbol identity

εkij εkℓm = δiℓ δjm − δim δjℓ .

δiℓ δjm − δim δjℓ

First, one defines the cycle function

mod(kn + j, k) = j for j = 0, 1, 2, . . . , k − 1 and integer n ≥ 0 .

mod(kn + j − 1 + k, k) + 1 = mod[k(n + 1) + j − 1, k] + 1 = mod(kn + j − 1, k) + 1 .

εkij = δf (k+1)i δf (i+1)j − δf (k+2)i δf (i+2)j .

001 qfull 00410 1 3 0 easy math: BAC-CAB rule (e.g., WA-32)

~ = mr2 [~ω − r̂(r̂ · ~ω)] ,

001 qfull 00504 2 3 0 moderate math: undetermined Lagrange multipliers

To find the ellipse in an aligned coordinate system use the transformations

001 qfull 00530 1 3 0 easy math: full derivative of a vector function

001 qfull 00540 2 3 0 moderate math: parallel gradients

∇(uv) = v∇u + u∇v .

d) Show that ∇u × ∇v = 0 is a sufficient condition for u and v to be related by some function

001 qfull 00620 1 3 0 easy math: vector identities

001 qfull 00709 2 3 0 moderate math: magnetic moment

001 qfull 00712 3 3 0 tough math: QM ang. mom.

εiℓm εjpq Hℓpmq = εjℓm εipq Hℓpmq .

Gij ≡ εiℓm εjpq Hℓpmq .

This definition may make the identity to be proven more identifiable.

e) Given the operator commutator formula

001 qfull 00802 1 3 0 easy math: double curl identity

where (∇·∇)A ~ is a vector made up of the Laplacians of each component of A:

001 qfull 00905 2 3 0 moderate math: surface integral

001 qfull 01010 1 3 0 easy math: closed surface B-field integral

001 qfull 01110 2 3 0 moderate math: area in a plane

~r = x̂a cos η + ŷb sin η ,

001 qfull 01203 2 3 0 moderate math: cross-producted gradients

001 qfull 01302 2 3 0 moderate math: Gauss’s law with symmetry

001 qfull 01410 2 3 0 moderate math: Dirac delta function

Note that the two properties imply

002 qmult 00100 1 4 5 easy deducto-memory: curvilinear coordinates

002 qmult 01002 1 1 2 easy memory: standard coordinate systems

002 qmult 03002 1 1 4 easy memory: orthogonal coordinates

002 qmult 05002 1 4 1 easy deducto-memory: spherical coord. scale factors

What are the , Alex?

and note that