MANUAL OF MATHEMATICAL PHYSICS by PAUL IL. RICHARDS Technical Operations, Ine, Burlington, Mawachuseta, U.S.A PERGAMON PRESS LONDON - NEW YORK - PARIS - LOS ANGELES 1959PERGAMON PRESS LTD. 4&6 Piteroy Square, London W. PERGAMON PRESS INC. 122 Baet 65th Street, Now York 22, N.Y. P.O. Boz 41716, Los Angeles, California PERGAMON PRESS 8.A.R.L. 24 Rue dee Beoles, Paria V+ 1989 Panoawox Pasas Lp. Library of Congress Card Number 68-12659 Print in Northern Ireland at The Univorvtin Prev, BelfastCONTENTS PREFACE PART I, PHYSICS Cuarren 1. MECHANICS Pint mass and rigid bodien Lagrange's equations Continuous media Elastic media. Fuids 2 ‘Non-viacous ude Teentzopie Bow Surface tension. Cuaron 2, ‘THERMODYNAMICS General relations - Alternative formulations Special substances Internal equilibrium Gibbs’ phase rule ‘Mase action laws ‘The third law Examples Irreversible thermodynamics Cuarsmn 3, ELECTROMAGNETIC THEORY Unite Maxwell's equat Foros 10 12 7 20 26 28 aReess 46 o 5 33 60 60 a 6 or B 6 6 eee 102 108 105" contents Cuareen 4. RELATIVITY Special relativity Electromagnetics Mechanic. ‘Thermodynamics General relativity Comology Cuarmee 5. QUANTUM MECHANICS ‘Postulates and formaliam General consequences and methods ‘Angular momenta ‘Matrix mechanios 5 Interactions with radiation Collision theory Specific results ‘One-particle problems ‘Two-particlo Atm. Molecules High-energy quantum mechanics Dirac theory. Electromagnetic fields ‘Regults (eross sections) Second quantization . Renormalization Cuarren 6. STATISTICAL PHYSICS ‘Simple kinetic theory ‘Knudsen flow Solid ate : "Thermodynamic equilibrium Partition function (sum-over-states) Classica! limit ‘Thermal radiation deal gases (quantum) deal gases at low density. Real gases (classical) Solids Liquids oe : Of equilibrium (transport phenomena) ‘Fluctuations , ‘Onsager relations ‘Nyquist theorems Gano ‘Boltemann transport equation no no uz us ne 120 ie 126 126 128 131 137 138 144 a9 9 163 158 11 mm am 16 19 184 185 187 187 189 190 195 197 202 au 215 221 222 222 223 224 235 227contexts Dense gases. Plasmas (ionized gases) ‘Chemical reaction kinotios Solids Solid, electron transport Semi-conduetors Liquids PART II. MATHEMATICS Cuarrer 7, ALGEBRA Roots of polynomials Curvecitting Finite oriea Permutations and combinations ‘Topology (elementary) ‘Nomographs Cuarenn &, DIFFERENTIATION AND INTEGRATION Differentiation . Seale transformations in physi Euler's theorem =. Integration Eliptic integrals. ‘Miscellanoous theorems ‘Methods for definite integrals ‘Numerical integration Modem integration theories Cuavren 9, INFINITE SERIES ‘Summing a series. Orthogonal function series Moment sequences Cuarren 10. VECTOR ANALYSIS Definitions Rigid motions Differentiation gradient operator (V) Integration ‘Tonsors (dyadics) N-dimensionsl spaces Cuarmn 11. DETERMINANTS AND MATRICES . Determinanta . Matrices . 27 239 2, 28 Me 251 253 253 256 257 200 261 262 205 205 265 268 267 212 213 276 218vm contents: Solution of linear equations Proctical calculations Special types Goneral theorems Signal-fow grapha Cnarren 12. FUNCTIONS OF A COMPLEX VARIABLE Analytio functions Singularities Conformal mapping Contour integration General theorems . Entire and meromorphie functions Cuarmn 13, INTEGRAL TRANSFORMS Laplace transforms». Fourier and bilateral Laplace transforma Others Approximate inversion Convolution transforms Curren 14, ORDINARY DIFFERENTIAL EQUATIONS ‘Methods applicable to all types . Linear equations (general order) First-order differential equations ‘Systems of differential equations Second-order linear equations ‘Table of sok : ‘Total differential equations Difference equations (Z-transforms) Cuaprmn 15, PARTIAL DIFFERENTIAL EQUATIONS First-order partial equations ; Quasi-linear ‘Non-linear : Systems of equations Wave-diffusion equation (second-order) Green's functions oe ‘Poiason’s equation ‘Laplace's equation ‘Wave equation . Diffusion equation ‘Linear partial equations (order > 2) Special equations, solutions Numerioal solution; etabilty ‘Non-linear equations 305 308 300 310 312 315, 315 si 318 319 319 320 322 322 328 328 331 332 335 239 aa 34 346 349 352 353 354 354 358 355 357 380 301 Bee 387 370 am 372contents ‘Tho transport equation Fermi's age theory Cuareen 18, INTEGRAL EQUATIONS ‘Particular equations, solutions Wiener-Hopf equations Fredholm and Volterra equations ‘Symmetric, normal modes ‘Equation of the “Bret kind” Cuarnm 17, VARIATIONAL PROBLEMS; LINEAR PROGRAMDING Basic methods Lagrange’s problem General forms Numerical solution Tinear programming Elgenvectors and eigenvalues Generalizations : Computation of eigenvalues and eigenvectors (Cuarrmn 19. EIGENVALUE PROBLEMS; PERTURBATION THEORY Second-order differential problems Infinite regions General formalism of eigenvalue problems With symmetric (Hermitian) operators With unit metric (p(F) = 1) Perturbation theory Ordinary. ‘Time-dependent Cuarmmn 20, PROBABILITY AND GAME THEORY Probability ‘Random variables ‘Limit theorems Chain reactions : Recurrent composite events Markoff chains =. Continuous random walks Game theory am 376 318 378 380 381 386 386 388 388 389 201 301 407 an a2 a3 as 418 ais 420 422 422 423 425 47 428 420 43 433x contents: Cuarran 21. TENSOR ANALYSIS Elementary relations Veotors, tensors Covariant derivatives Riemannian geometry Curvature tensor Non-Riemannian goometry Cuarren 22. GROUP THEORY “Lie groups: Formal group theory Representation theory Character systems Infinite groups Applications 5 ‘Determining aymmetry-adapted functions 436 «a7 438 440 43 443 45 “7 ar 450 452 452 455 450PREFACE ‘Tax goal of this book is to condense established theoretical physics, its applications and its mathematical equipment into a single reference volume of reasonable size ‘without sacrificing either logical continuity or fundamentals, In this way, each for- mula appears in its deductive context and its origin, as well as any approximations or aasumptiona which it may entail, can readily be determined. To render this ideal more approachable the Physics section has been limited to ‘theories which have been well established by experiment, and their deductive rami. fieations have been terminated while the results atill maintain a wide utility in applications. Likewise, the Mathematics section has been economized by omitting ‘much material which can be found in tables of integrals or in compilations of the properties of the classical functions of analysis, ‘To facilitate rapid extraction of information, an attempt was made either to define or to cross-reference every special concept and every symbol within at least a few preceding its every appearance, although some exceptions necessarily occur with the more standard symbols. ‘Three guides for finding information have been provided: (1) The table of eontenta diaplays the overall organization of the material and lists the major subjects within each chapter. (The chapter number forms the first half of each equation number.) (2) The left-hand page headings designate the major subject area, while the right-hand page headings mention individual items which appear on the corresponding pair of pages. (3) The index is as completo as the author could make it and should suffice to locate any item eontained in the book. The index has also been designed for use as a dictionary of terms and concepts, by including even items whieh are merely mentioned or parenthetioally defined in the text. Any suggestions or comments which users may feel would add to the general utility of the book, either as a reference or as a study guide, will be gratefully received.vant 1 PHYSICS PHILOSOPHY Untree mathematics, physics has as its prime purpose the description of the real world Tho mathematician need only assure himself that his theorems follow from his axioms, but the physicist must continually ask Nature to pass judgement on the smallest facet of his theories. ‘The facts of the world alone are the final judges of the usefulness and truth of any theoretical model. Ideally, there should be no disagree- ment whatever, but we must frequently make do with imperfect theories, being fully cognizant of their limitations. ‘Yet physics (especially of all the sciences) is not # mere listing of data. Rather physios is mainly a set of mathematical models of Nature. These models serve the dual ‘purpose of describing the logical interrelations of various facts and at the same time of summarizing great masses of data which can then be discarded. Unlike the mathematician, the physicist must stand ready at all times to abandon previously “established” theories. However many and varied may be the experiments which have verified” a theory, it has never been proved; a new fact of Nature can Iways arise to contradict it. Yet “‘well established” theories are never wholly wrong; if they properly describe & wide range of data, they must be good approximations and, indeed, must appear as special eases of any more general theory. Thus Newtonian ‘Mechanics, while now known to be wrong in a philosophic sense, remains a respected ‘and indispensable tool of physics. Under appropriate conditions, it gives to high ‘accuracy the same answers as the more ponderous and inconvenient theories of Quantum Mechanios or of Relativity, of each of which it is @ special case. Not to use it would be mere pedantry.onarren 1 MECHANICS A. POINT MASSES AND RIGID BODIES Point Particles; Fundamental Concepts ‘Drxore the position ofa “emall” particle (mass point) by the vector, x, relative to any’ convenient origin ‘tis found experimentally that the acceleration, dtr/d!, of « mass point is, in many cireumstances, independent of ita previous motion and depends only on its position (and sometimes on the time). Moreover, different mass points are usually found to suffer proportional accelerations under the same cireumstances, Tei to introduce concepts of “masa” and “force” defined by ar Fame ay ‘whore m i constant characteristic of each particle and called the mass ofthe particle (chosen as unity for some standard particle) and is the force which is then defined by (1-1) (as applied to the standard particle or one whose mass has been determined by comparison with the standard under identical conditions). Newton's Third Law* states that if a body A exerts a foree, F, on body B then, conversely, body B exerta an equal and opposite foree, —F, on body . ("To every ‘ction, there is an equal and opposite reaction.”) ‘Unlike the definition (1-1), this is a statement of fact which withstood 200 years experimental investigation (but is now mown to have exeeptions—at least with the simple definition, (1-1). The term “body” used above ultimately refers to mass-points but, by summation or integration, may also beinterpreted as a physical ystem of any degree of complexity ‘Note finaly the implication that the forces ean be regarded as “caused” by other agenciea:t ‘This, too, i question of fact and haa been well bore out by experience. Tn greater detail, experiment indicates that forees “from” different agencies, which act on the same particle, may be added veetorally Frater 4 = BF ee dst From Newtons Third Law t flown that i aysem of paris “alate” in th te tht alae pode within th tent (no xe tdareindopndent of extemal condition, thn 0 = SF, = 5 (52m) oF v= def) he race ae ‘The fret wo are contained in (1-1) and are really « definition of “force.” 4 Philonophioally better: certain forces are presont when and only when the auociated agencies fare prosont(cortelation, not causation) More precisely, ifthe Lagrangian function of the aystem is independent of translations of the coordinates describing the system. 34 MECHANICS OF Mass POINTS (The law of conservation of “momentum”, P.) Correspondingly, if we define the ‘momentum of a particle as py =m,¥, then S p, = constant if no external foroes act. ‘The “work” done between times t, and f, by the force F acting on a particle which ‘undergoes displacements dr (whether “due to” the force or not) is defined by we ['P-ar 0-3) ‘From (1-1) * we ino -7, (4) A where Taint 5) is mown as the “kinetic energy” of the particle, ‘The work done on a particle is always equal to the change in ita kinetio energy, according to (1-4). ‘The change in momentum p = mv of a particle is always equal to x= [Pa == on oy which is called the “impulse”. ‘Moving Coordinates, Coriolis Forces All of tho above relations are valid in general (since, indeed, most of them are definitions) but in moving coordinates one must remember that the “basis vectors”, e, in r =e, + 2,e, + 2,¢, aro also moving so that their time derivatives enter along with those of the “componenta", x. Sinee (I-1) involves a double time deriva- tive, some of the extra terms will appear multiplied by the (local, apparent) velocity of the particle, dd. These extra “forces”, which have the property that they appear only when the particle is in motion (as seen from the moving coordinates), are known as “Coriolis foroes”. Potential Fields, Conservative Forces In many circumstances F oan be expressed as the gradient of a simple scalar fanotion: F=—V9(e, 0) (7) (evaluated at the position of the particle). Here g(r, t) is known aa the “potential enorgy”. If ¢ is actually independent oft, then from (1-4), (1-1) and (1-1), Sr 4 $=; + $=F,ecomtant (-8) Hore H is called the total energy of the particle. Forces which keep 2 constant are called “conservative forces”. Conversely, if the total energy is conserved, the forces must be derivable from some ¢ via (1-7). (Thus, if |v] depends only on r—for any orbit—then 4 defined by (1-8) satisfies (1-1). TBxanrtz: (Nowton's law of grovity). The gravitational free which each of two “particles” exerts on the other is experimentally found to be sama =" — oo 2 awPOTENTIAL FIELDS: CONSTRAINTS 5 Here @ is a universal constant of Nature. Both forces are simply obtained from 155 we take mG 19) in—"d ae Orbit-Tracing in Potential Fields Lot U = (constant — 4) s0 that F = +L VU and adjust the constant so that, for the orbit desl, me? —'U, hen if the fadapenent variable, fin (1) replaced bythe pothlongths, the nul dee a sv [oe] wa Tats Curate seco Hj ot (5) ppt cit ‘This relation ean be made the basis of numerical or graphical method for tracing the path of a particle under the influence of conservative stati foroes. Constraints In some cases, a particle is required to satisfy such conditions as g(r, 0 (Exampie: motions of a particle over a surface.) Such cases are often most easily treated by substituting these conditions directly in the equations of motion (or by using the Lagrange equations; see later), Occasionally, however, Lagrange’s method of undetermined multipliers is more convenient: We have Vg. = 0; multiply these by unspecified functions, Ag(t, t) and add to (1-1), ar enh + TAVa (1) These equations may often conveniently be solved by algebraically eliminating the 4, and picking out solutions of the resulting relations which satisfy ga(t, f) =0. (If the constraining foroes, 2,Wya, are desired, they may then be found by returning to (1-12) ‘Systems of Mass Points As already hinted, a system of many mass points, m,, is described by a set of equations (1-1), one for each particle, along with a presoription, perhaps of the form (1), for the forces. ‘The total momentum and total kinetic energy of the system aro defined as the sum of those of the individual particles @ P=Smyj T= ime? aig) and if F, = —V,¢ (that i, if (F), Bde te sory ete) then +++; Sn equations) (1-13) 7 + $=, a constant, the energy of the system. Note that the concept of potential of « partile of the system need not have meaning.6 MECHANICS OF SYSTEMS OF PARTICLES Center of Mass If we define 1 Ragin (i= 5m) (1-15) 1as the position of the “conter of mass” of the system, then P = M(dr,/d!) and from (1-18) and (1-14), ap er, GaRRs oo: Fane (ate) where F is the total ezlernal force acting on all the particles. (By Newton's third law, the overall sum of internal forves is zero.) ‘Thus the center of mass of a system moves as if all the mass and all external forces were concentrated there. "Thus macroscopic bodies, whether rigid or not, ean be treated as point partiols, if we wish to know only the motion of the oonter of mass. Kinetic Energy From (1-14) and (1-16) it follows immediately that 7 = tty. + 3 amivs — ve (-17) (where v, = dr.Jat). That is, the kinetio energy of the aystem can be expressed as that of a mass point If = ¥'m, concentrated at the center of mass plus the kinetio ‘energy due to motions of the particles relative to the center of mass.* Angular Momentum; Torque To indioate gross features ofthe additional motions of the system around its center of mass, definet TaImyxy Larnxk (1-18) called respectively the total “angular momentum” and total “torque” (or “moment”) on the system. From (1-13), it follows that a a 4s a “gross” equation for the motion about the center of mas, L (1-19) Equation (1-19) is the antisymmetric part of the more general dyadic relation or each particle) a Gpmey =F + mv 0-20) Which ia alao an immediate consequence of (1-1). Tho symmetric part of thie relation is a wre ear —er-p) * In view of (1-16), it might appoar that tho extra term in (1-17) violates tho genoral princi (1-4), edocs not, of course, and the reason is that, dospito (1-10), the external foreas do not fctuaily act on r, and this murt be taken into account in computing the work done by them {Note thet the ordar of the factors in (I-18) i purely a matter of convention; the correspond. ing sign-ambiguity indicates that J and L aro “axial” vectors.COLLECTIVE RELATIONS 1 26m ayn i ah hat ond, san aking tne avrg 4) ue] ve the “ira theorem’; summing over all tho particle ofthe petem: Xue = (BAO R), (oan) If both the positions and the velocities are taken relative to the center of mase, one obtains the same equation, Thus if J, = Tmley— 8) % (Ve — Vd) xP (1-22) and if L=S(y—r) eR Ln xP (23) then from (1-16) and (1-19) Ee ‘e (1-24) a quito independent of the motion of r,. In partioular, if no external forces act and the mutual (equal and opposite) forees between two particles lio on a line joining them,* then L, = 0 so that J, is constant. Rigid Bodies Fora rigid systom of mass points, equations (1-16) and (1-24) completly determine the entire motion: First, define the following tensor (“moment of inertia tensor”—relative to the center of mass), 9.= Emde, — est = (ey ne, A) (1-25) (the last term isnot a dot product) where 1 is the unit tensor and r, the conter of mass. Noto that, juntas fin dtined independent ofthe condinte mite, 0 the tensor 3, ia completely efiaad by (1-20) quite indopendently of wether the body iin notion er 08 Irn m moving, the component of 5, ine Gxed scone stern may change (ord, theme ‘of but the fonoe changer” lyin the sre snus that F changes For a rigid body, it follows from (10-26) that ‘the position of Vim V 0X (Hy 1) (1-26) where e (which can be time-varying) is known as the instantaneous angular velocity of the body. From (1-26), (1-26), (1-22) and (1-11) it then follows that 3.=3,-0=0-3, 27) and that} 1 =iMyP + jw-3,-0 (1-28) ‘The main advantage of these relations appears upon transforming to a coordinate aystem fized in the bly. ‘Then the components of 3, become eonstants and, moreover,8 MECHANIOS OF RIGID BODIES because these components form a symmetric (and thus Hermitian) matrix, there is one set of (orthogonal) coordinates, z, y, 2 in which 3, is diagonal: Boa =1y5 (constanta) (1-29) ‘The three constants, I,, are known as the ‘‘principal” moments of inertia. It is easily shown [by taking the origin at r, and writing out explicitly the “zz” component of (1-25)] that those oonstanta are given by Tae = Tm? + #2), ete. (1-80) ‘Thus the principal moments of inertia are necessarily positive. ‘From (10-27) it follows that, even though the coordinates are fixed in the rotating body, the equations of motion (1-24) take, with the help of (1-29) and (1-27), the form a, _ 3, deo 1a a ox a tex B-0] (3D or, in components: do, Tan Get Uae — Tne (a-s2) ‘These equations completaly determine the motion of the rigid body about its center of ‘mass, Note that tho simple form (1-82) is obtained only if principal axes fized to the body are used. Note also that if L — 0, ois still not constant unless two of ita three ‘components (in thie system) vanish; that is, the only stable axes of rotation for a free body are the three principal axes. Rigid Body with One Fixed Point If a body is “anchored” at one point but is otherwise free, it becomes useful to take that point as the coordinate origin, r =O, and then define a moment of inertia tensor: S=Ymfral — ee) (1-33) Unlike 3, this tensor does depend on the position of the coordinate origin but, of course, is otherwise independent of the coordinates or motion of the body. Since v= 0 at r =0, it follows from (10-26) that v, =w x F, for some (time- varying) , the instantaneous angular velocity. ‘Then, as above, J=0-3=3-0 39) T= 3-0 (1-85) ‘Note that (1-85) includes the term, $v, of (1-28) and that both (I-34) and (1-35) fare special consequences of the fact that v = 0 when r = 0. ‘The point need not be actualy a part of the body: in nome epecial cass (gyro op in a cage, {or example) an imaginary extension of the body may be used.ROTATION ABOUT A POINT OR AXIS 9 Again, the relations are especially simple in coordinate axes fixed to the body. In such coordinates, the components of SJ become constants and as before (compare (1-81) the equations of motion take the form,* 388 pax Ow) =, (1-36) which also takes a very simple explicit form, essentially identical to (1-32), if'“prineipal axes", in which 3 is diagonal, are employed. ‘These will in general be different from the coordinates which diagonalize 3. Rotation About a Fixed Axis If a body is constrained to rotate about a fixed axis then « = constant unit veotor. If we also choose the coordinate origin on the azis then ¥, aX r, and from (1-18) aS=o 3m, |ax rye (37) whence by (1-19) the equation of motion may be taken simply as sa=t, (1-38) ‘where the constant, Seman nt (1-89) is called the moment of inertia about the axia, a. Note that r, must be measured from a point on the axis; |a X 1,| is then the distance of m, from the axis. Note also that the constraining forces (at the “axle bearings”) do not contribute to the ‘component, This simple numerical “moment of inertia” constant, Z,, is obtained only when the axis of rotation is constrained to be constant in both direction and points of passage through the body. Otherwise, the full (six-component) concept of an inertia tensor rust be used and even ite components are functions of time when coordinates fixed in space are employed. Parallel Axes If Tyis the moment of inertia about » second axis, parallel to a but passing through the center of mass, r,, then (recalling that the origin must lie on the axis in question in each ease) we have, in the same coordinates used for (1-39) I, = Sm, |a x(t, — 0)? =1, + Mla x 1? — ax 4.) - Smilax rv) =1,— Max? (the last step by definition of r,). Hence 1, (1-40) or: ‘The moment of inertia about any axis is equal to the moment of inertia about a parallel axis through the center of mass plus the moment of the total maas (considered concentrated at the conter of mass) about the chosen axis. Ty may, of course, be ‘computed from either the analogue of (1-89) or by [y= a*3,-'8 = (Saxe * We havo written L, in (1-86) to emphasize thet, in ovalusting L, the coordinato origin must bbe chosen at the fied on,10 LAGRANGIAN MECHANICS Lacuanox’s Equations Notation ‘To simplify non-veotor expressions, it is common to label the components of the various position vectors consecutively (and they are commonly denoted by q,); thus ho a= Yo dy ~ 4 ty — ty ote. Furthor, time differentiation is denoted by « dot; thus dy =d, = ge the y-component of the velocity of particle #2. Equations (1-13) and (1-14) then read mae = Fy (a4) T= Simae (1-49) (Here m, takes the same value for three consecutive values of i, and the have been labeled in a manner paralleling that of the 4's.) Conditions ‘Lagrange's equations are less general than (1-1) but cover the great majority of cases of interest and have the advantage of being easily written down in whatever coordi- nates may seem desirable. If the foroes F, in (I-41) can be expressed by means of a “generalized potential”, ‘¥(Ge dy 8, in the form* an 8 (ae ra Stal) ae Define (with 7 given by (1-42), Ty dot) =T —y (1-44) ‘Then (1-41) and (1-48) are equivalent to (4a) Note, howover, that L is not unique; (1-45) remains equivalent to (1-41) and (1-43) if we add to L any (total) time derivative of an arbitrary function of q,¢ (but not d,); that is, x Li also satisfies (145). ‘Equation (146) has been derived for Cartesian coordinates but it is actually quite general. Namely, (1-45) is also equivalent [for fixed q,(,) and 9,()] to Aff H20 409 a4] but, in this form, it is evidently immaterial what coordinates are used to deseribo the positions, Moreover (1-45) will then also hold as a direet consequence of (1-46). It is this feature which makes the Lagrangian formulation so powerful. ‘The functions 7’ and y ean usually be written down immediately, whereas acceleration al a ras sae Tt Flo) (44a) o (1-48) * Tho notation implice that qu.-**, due dy °°, dust aro to bo considered independent in forming ‘tho partial derivatives in (1-43). ‘Tho dingo deivativo in (1-43) i takon aftr evaluating the partials, ‘These conventions held in ll that follows.