arXiv:1907.11239v4 [physics.class-ph] 1 May 2020 namoving a on q h w vlto equations The of of evolution consisting two source interacts. fields, the a it these for as which equations properly with Maxwell–Lorentz treated fields but electromagnetic particle” the idealized “test not noto- is a is electron as formula point the this However, when ill-defined vacuum. riously in light of speed o h lcrmgei oc xre ya by field exerted electric force electromagnetic the for oehrwt h w osritequations constraint two the with together mtia orvco oain u otyw ilwork will we mostly but notation, four-vector ometrical a motion. up charge setting to point devoted of defi- is theory paper major classical others this by well-defined the of out rest pointed recall The were we that already. approach then dealing this in and of affairs ciencies problem, of state this current with the summarize first we aiyat larity aei li htMxelLrnz(L ed iha with fields at (ML) source, charge Maxwell–Lorentz point that single plain it make ∗ ( [email protected] t h rcia ucs fteLrnzfrua[1] formula Lorentz the of success practical The ewl cainlyivk aietyc-ain ge- co-variant manifestly a invoke occasionally will We ,advelocity and ), ∂ ∂t f ∂t ∂ Lor em B B B E E E s ( ( ( = t, t, t = ) ti hw htawl-endepeso o h oa elect total the for expression well-defined a that shown is It rpitwt rau mlmne clrdcreto mar correction (colored implemented erratum with Preprint h eoiyadtemmnu fapitcag ih“aeres “bare with charge point a of momentum the and velocity wh importantly, the Most text equations. the field disqualifies Maxwell–Born–Infeld integrability This local Maxwell–Bopp–Land´e–Thomas–Podolsky equa charge. the generic field while point the the is of these neighborhood extract Among be fi electromagnetic can conditions. classical field the whenever electromagnetic conservation classical tum the of source o vrgn ffilsoe egbrodo h on charg point the of wit reno electrodynamics mass neighborhood classical bare a of infinite over foundations No fields microscopic up. of show averaging not wh hoc does position motion the of of derivative equation time third-order infamous the o’ law ton’s oc napitcag oreo h lsia electromagne classical the of source charge point a on Force s s E E etpitelectron point test E q + = ) = ) ( ∇ ( ∇ t, t .INTRODUCTION I. − .I h eane fti introduction this of remainder the In ). s B · E · − e n antcidcinfield induction magnetic and ) B B v E E  c c ( ( ( E E E ∇× ∇× t t, t, swl salse;here, established; well is ) ( d d t, s s t p 0 = ) = ) q B E B E B E = ( ( ( t t, t, )+ )) − f q s s , 4 with ( ) 4 + ) πeδ t , ihcharge with ), 1 c 1 rlnhye d,Psaaa,N 85,USA 08854, NJ Piscataway, Rd., Frelinghuysen 110 v must q 1 f ( πe ( eateto ahmtc,RtesUniversity, Rutgers Mathematics, of Department t t ) ) = × ( v s ( f B B B aesm singu- some have ) t em ) , ( δ t, ie,smooth given, q eoe nitga qainfrtepitpril’ accel particle’s point the for equation integral an becomes ( q − ihe .H Kiessling K.-H. Michael t ) ( e ( t position , s )) )  c , B B B sthe is ( t, (1) (3) (2) (4) (5) s ) B B ttime at where rmgei edslto,frwih( the which for as elec- solution, source-free written continuous field Lipschitz be tromagnetic general their the can linear, of solution sum is sys- equations the distributional Since field general time. Maxwell–Lorentz long of a for tem explicitly acceleration known been continuous has piecewise bounded having and s with | hre h lcrmgei Li´enard–Wiechert electromagnetic fields The charge. es,pu h eaddL´nr–icetfil 2,[3], [2], Li´enard–Wiechert field retarded B the plus sense, of foliation flat points any space in Euclidean Note valid into spacetime are obscured. Minkowski formulas then above even is the seek, co-variance that we Lorentz problem the value formu- initial though the dynamical for a well-suited of splitting lation time & space the with where n otemnfsl l-endsmoi expressions symbolic ill-defined manifestly the “ to ing too. source, charge point the of location the at singularity s E E E n hr “ where and , lw ret − h edsnuaiyascae ihamotion a with associated singularity field The na tep ogv oemteaia mean- mathematical some give to attempt an In ( t, q on charges. point h := E E E q ( H H H t D D D l hoystse adu fregularity of handful a satisfies theory eld lw ret t ret c lge h Abraham–Lorentz–Dirac the plagues ich ( r n nteuulrltvsi eainbetween relation relativistic usual the en ret dfo h otlt fttlmomen- total of postulate the from ed oantcforce romagnetic H H H lw ret lw ret t ftefil oetmdniyoe a over density momentum field the of okMxelLrnzfil equations, field Maxwell–Lorentz book t ) n “ and ))” ( in ulf,adpeual od the do so presumably and qualify, tions denotes ( 1, xii oha both exhibit q lw ret ) ∈ t, ( ( | t, t, , ∗ here, ; mass” t .Teapoc asterigorous the lays approach The e. .Tecretsaeo affairs of state current The A. s s R & s s en endipiil by implicitly defined being ) mlzto sivkd n oad no and invoked, is rmalization = ) . = ) = ) E E E | ret lw ret kings): | − | n − B B B s a s s en ht( that means ” m − − := ( ( e − e steaclrto etro h point the of vector acceleration the is q t, q q b c n | , | 2 s q q D D D spsuae,New- postulated, is 0 6= s ( ( sa is | − ( f q − ) | lw ret t t ∝ s | , pi 1 2020 21, April em ) ) when ))” ret | q s ih(f [4]) (cf. with , v − hyec aeadirectional a have each they ; 1 ) | × | napitcharge point a on normalized 2 × /r 2 q D D D |  2 c lw ret c c c n a and − n n − q ( i field tic t, , ( ( n E E E n q v q s ( ( , eration; , , ( t, q ) s q a s , ∝ , s ) ) , ( = ) 1 s etrfrom vector and ) s ae perfect makes ) − ) − 1 ) · /r · v v v c v q
(

t 3 singularity, × , B 3 B B

v − ret a ( , t t,  a t 7→ s ret s )( ret ∈ are ) q t a q = ) B B B ret (6) (7) R ( ( lw ret to t t 3 ) ) ) 2 a sum of a regular source-free field and the Li´enard– to quantum electrodynamics and, more generally, quan- Wiechert fields (6) and (7), Lorentz and his contempo- tum field theory. However, if electrons are true points raries averaged the fields (t, s) and (t, s) over a neigh- without structure, then Dirac’s construction is logically borhood of the point chargeE at q(t), butB this does not lead incomprehensible: if a point electron has a bare mass, to unambiguous finite vector values for “ (t, q(t))” and then it cannot depend on the radius r of a sphere over “ (t, q(t)),” and when the neighborhood isE shrunk to the which one averages the Maxwell–Lorentz fields. pointB q(t) the infinities are back. The conclusion at the Postponing such logical concerns until later in his life, time (and also more recently in [5]) was that the physical Dirac obtained the Abraham–Lorentz–Dirac equation, electron cannot be assumed to be a point, but must have which in the four-vector notation explained in [9] reads an extended charge distribution and perhaps some other 2 structure, all of which to determine became a goal of d e ext d mobs 2 q = F (q) q (9) what Wiechert and Lorentz called “electron theory” [6], dτ − c · dτ 2 nowadays referred to as “classical electron theory.” It’s 2e 1 d d d3 + g + q q 3 q an interesting dynamical theory in its own right; for more 3c3 c2 dτ ⊗ dτ · dτ recent investigations, see the books [7] and [8], and the   papers [5] and [9]. Since we are interested in the theory where the term in the first line at r.h.s.(9) is an “exter- of point charge motion, we here do not spend much time nally generated” test-particle-type Lorentz Minkowski- with classical electron theory, except that we note that force, and the term in the second line at r.h.s.(9) is von some insights gained in its pursuit made their way into Laue’s radiation-reaction Minkowski-force. the prevailing classical theory of point electron motion While (9) is free of infinities if Fext(q) is smooth, the which was put together by Dirac and Landau & Lifshitz. third proper time derivative in the von Laue Minkowski- In 1938 Dirac [10] invented negative infinite bare mass force means that (9) is a third-order ODE for the position renormalization to avoid the infinities which occur when of the particle as a function of (proper) time. The perti- the averaging surface about q(t) is shrunk to q(t). Fol- nent initial value problem therefore requires vector initial lowing Fermi’s contribution to classical electron theory data for position, velocity, and acceleration. Yet a clas- [11], Dirac averaged the fields over a sphere of radius r sical initial value problem of point particle motion may only involve initial data for position and velocity. centered at the electron in its instantaneous rest-frame. ... Also, he worked with linear combinations of the retarded Landau & Lifshitz [12] handled this q problem in the following perturbative manner. They argued that von and advanced representations of the fields. With mobs ... denoting the electron’s “observable rest mass,” Dirac as- Laue’s q Minkowski-force term must be a small pertur- bation of f ext whenever test-particle theory works well. signed an averaging radius-dependent bare mass mb(r) ... to the point electron, defined by In such situations one may compute q perturbatively by taking the proper time derivative of the test-particle e2 1 equation of motion, i.e. mobs = limr↓0 mb(r)+ 2c2 r ; (8) d3 e d ext d   3 q F (q) q . (10) dτ ≈− mobsc dτ · dτ evidently, mb(r) as r 0. As is well-known, Dirac’s mass-renormalization↓ −∞ computations↓ became the R.h.s.(10) depends only on q, q˙ , q¨. If we substitute
it for template for the modern renormalization group approach d3 dτ 3 q at r.h.s.(9), equation (9) becomes

2 3 d e ext d 2e 1 d d d ext d mobs 2 q = F (q) q 4 g + 2 q q F (q) q , (11) dτ − c · dτ − 3mobsc c dτ ⊗ dτ · dτ · dτ

an implicit second-order ODE for the position of the elec- owe their longevity to their reputation as practically ef- tron, compatible with the available initial data. Equa- fective equations of motion for the computation of (first- tion (11) is called the Eliezer–Ford–O’Connell equation order) radiation-reaction-corrected test-particle dynam- in [14]. The equation presented by Landau–Lifshitz [12] ics of point charges in smooth “external” fields. When differs from (11) by an additional approximation: noting point charges are replaced by extended charged parti- 2 d ext d d ext d ext d cles the Landau–Lifshitz equation can be derived rig- that dτ F (q) dτ q = dτ F (q) dτ q+F (q) dτ 2 q, ·−e ext d d·2 · orously from the Abraham–Lorentz model with nonzero they substitute F (q) q for 2 q in the last term. mbc
·dτ d
τ bare mass, using center-manifold theory [8], and also in a As recently as in [13] the Eliezer–Ford–O’Connell equa- “vanishing-particle limit” [5]. Its solutions are expected tion (11), resp. its Landau–Lifshitz approximation, was to agree reasonably well with empirical electron motion still presented as the state of affairs in the classical theory in the “classical regime” of weak and slowly varying “ex- of point charge motion in flat spacetime. These equations ternal” fields, and demands for higher precision can be 3 met with improved effective equations, obtained either charge. The N-body situation will be discussed in sec- by adding higher-order classical radiation-reaction cor- tion VI. The motion in a constant applied electric field rection terms, or by invoking QED. However, (11) has is revisited in section VII. major shortcomings even as an effective equation of mo- tion for true point charges! II. BASIC DEFINITION OF THE ELECTROMAGNETIC FORCE

B. Critique Mechanically the point charge is a point particle with a mechanical momentum To convey a first feeling for the limitations of the Eliezer–Ford–O’Connell equation (11) and its Landau– v(t) p(t)= mb , (12) Lifshitz approximation, we recall the well-known fact 1 2 1 2 v(t) that the last term in (11) vanishes for electron mo- − c | | tion along a constant applied electric field, and so does q v d q its Landau–Lifshitz approximation; cf. [15]. Thus in with (t) := dt (t) its velocity and mb = 0 its bare rest nd 6 this textbook situation (11) fails to take the energy- mass. By Newton’s 2 law the rate of change with time momentum loss due to radiation by the electron into of the particle momentum equals the force acting on it, account, i.e. its solution is identical to the familiar test- d particle motion. To radiation-reaction-correct these re- p(t)= f(t). (13) quires a non-vanishing higher-order term. dt A much more serious limitation of the Eliezer–Ford– The force f depends on the non-kinematical qualities of O’Connell and Landau–Lifshitz equations was discovered the point particle, in this case its electric charge which recently [16], by considering the many-body version. In couples the point particle to the electromagnetic field. this case each point charge satisfies its own equation (11), Electrodynamically the moving point charge is a source ext indexed by a subscript k (say) at ek, mk, qk, and at Fk , / sink for a classical electromagnetic field with field mo- ext field where Fk is now the Faraday tensor of the Maxwell– mentum (vector-)density Π (t, s). Suppose the fields Lorentz field given by a sum of the Li´enard–Wiechert decay sufficiently rapidly as s so that Πfield(t, s) is fields (6) and (7) of all the other particles but the k- integrable w.r.t. d3s “at spatial| | → infinity.” ∞ Suppose also th, plus the source-free field. Since nobody knows the that the field singularity caused by the point charge is past histories of all the particles which enter the Li´enard– mild enough so that Πfield(t, s) is locally integrable over Wiechert formulas (6) and (7), one has to stipulate some any neighborhood of q(t), so that past motions. But whatever one stipulates, as shown in [16], typically a singularity in the Maxwell–Lorentz pfield(t)= Πfield(t, s)d3s (14) fields will propagate along the initial forward light-cone R3 Z of each and every point charge, so that this system of equations of motion coupled with the Maxwell–Lorentz is a well-defined total field momentum vector. Finally, suppose that the motion of the charge is sufficiently reg- field equations is typically well-defined only until a point field charge meets the forward initial light cone of another ular so that p (t) is differentiable with respect to time. point charge. This is much too short a time span to Now, following Poincar´e(cf. [17]), we postulate that be relevant to, e.g., plasma physics. This problem can- in the absence of non-electromagnetic forces only those not be overcome perturbatively by adding a higher-order motions are permissible which satisfy the balance law radiation-reaction correction term at r.h.s.(11). More- d d over, it’s not just the radiation-reaction term in (11) p(t)= pfield(t); (15) dt − dt which causes trouble — the expression of the Lorentz force of one particle on another is typically not well- i.e. any momentum gain by the particle is compensated defined on the initial forward light-cones. Lorentz elec- through a corresponding momentum loss by the field, and trodynamics for N > 1 point charges is in serious trouble! vice versa. It then follows from (15) in concert with (13) The above discussion leaves no room for reasonable that the electrodynamical force on a point charge source doubts that ingenious extraction of effective equations of the classical electromagnetic field with a single source of point charge motion from the mathematically ill- is to be defined by defined, merely symbolic “Lorentz electrodynamics of d point charges,” is not a winning strategy to arrive at a f em(t) := Πfield(t, s)d3s. (16) − dt R3 mathematically well-defined and physically accurate rel- Z ativistic theory of point charge motion in the classical realm. In the remainder of this paper we explain how to Remark II.1 formulate such a theory in a manner which preserves the Postulating the balance law (15) is equiv- spirit of Lorentz electrodynamics as much as possible. alent to postulating conservation of total momentum, To keep matters as simple as possible we first consider d P (t)= 0, (17) an electrodynamical system featuring only a single point dt 4 for a total momentum defined as the sum of particle and charges into a well-defined theory: there is an even larger field momenta, i.e. set of Maxwell–Lorentz field initial data satisfying the constraint equations (4) & (5) for which (16) is not well- P (t) := p(t)+ pfield(t). (18) defined already at the initial time. Also, even with favor- able special Maxwell–Lorentz field initial data the elec- We pause for a moment to comment on (16) in the tromagnetic initial force typically cannot be continued context of Lorentz electrodynamics with point charges. into the future; i.e. typically the expression (16) is not well-defined for continuous stretches of time, the static A. Connection with Lorentz electrodynamics special case being an exception. Moreover, the initial en- ergy density of an ML field with a point charge source is If we assume the electromagnetic fields satisfy the never locally integrable, i.e. for such data the field en- Maxwell–Lorentz field equations with a point charge ergy in any finite volume containing the point charge is source (2) & (3), and (4) & (5), in which case infinite even if their field momentum (14) exists. Πfield 1 × (t, s) = 4πc , then r.h.s.(14) is generally ill- defined (generallyE Bin magnitude), and then r.h.s.(16) III.THE ROLE OF THE ELECTROMAGNETIC has no well-defined∞ meaning either. However, pretend- VACUUM LAW ing that r.h.s.(14) was well-defined, and that so was r.h.s.(16), and furthermore pretending that all the ensu- In the following we will discuss (16) for some electro- ing (advanced) multi-variable calculus and analysis steps magnetic field theories which (are expected to) yield a were justified, r.h.s.(16) would turn precisely into the ex- typically differentiable (14). Incidentally, (16) was also pression of the Lorentz force (1). For later convenience the starting point of Abraham [18] and Lorentz [6] for we recall those steps: computing the electromagnetic force on a charged par- ticle in their classical theories with extended electron (a) pull the time derivative into the integral; models; cf. [4]. While Abraham and Lorentz and their (b) apply the Leibniz rule to get peers chose to replace point electrons by extended struc- ∂ ( × )= ∂ × + × ∂ ; tures but otherwise continued to work with the Maxwell– ∂t E B ∂tE B E ∂tB Lorentz field equations, we instead continue to work with (c) now use (2) to express
the partial time derivative point electrons but replace the Maxwell–Lorentz fields of in terms of ∇× , and (3) to express the partial with solutions to the pre-metric Maxwell equations [19] B E time derivatives of in terms of ∇× and vδq; which satisfy different electromagnetic vacuum laws — (d) use an advanced vectorE calculus identityB in concert which furnish locally integrable energy-momentum den- with (4) & (5) to write the so manipulated r.h.s.(16) sities of fields with point charge sources. (“Pre-metric” as a volume integral over the sum of the divergence means, the spacetime metric plays no role; see [19]. It of Maxwell’s stress tensor, plus the Lorentz force enters through the vacuum law.) Since any classical elec- 1 × tromagnetic field theory will be about some distinguished vector density e + c v δq; (e) use Gauss’ theorem− E to concludeB that the contribu- subset of solutions of the pre-metric Maxwell field equa- tion from the stress tensor vanishes;
tions, the issue is indeed to identify the physically correct (f) carry out the volume integral of the force vector classical electromagnetic vacuum law! We refrain from density and thus obtain (1). trying to make such a definitive identification but instead Remarkably, not a single one of these six steps is gen- consider two well-known proposals: the nonlinear system erally justified within the symbolic system of equations proposed by Born & Infeld [20] (see also [21]), and the known as “Lorentz electrodynamics with point charges.” linear higher-order derivative system of Bopp [22, 23], Nevertheless this “pseudo derivation” of (1) from (16) Land´e& Thomas [24, 25], and Podolsky [26] (see also does suggest that by replacing (1) with (16) one may [27]). accomplish what was intended by Lorentz and his con- temporaries. A.The pre-metric Maxwell field equations As a first encouraging observation we register that The pre-metric Maxwell field equations are a four- there is a large set of field initial data satisfying the con- dimensional (complex) analog of the familiar three- straint equations (4) & (5) for which the formula (16) dimensional ∇ =0 = = ∇× . Explicitly, of the electromagnetic force is initially well-defined — the continuity equation·BB ⇒ B A even for the Maxwell–Lorentz field theory! In the special case of an electrostatic field of a point charge at rest, the ∂ ρ(t, s)+ ∇ j(t, s)=0 (19) electromagnetic force (16) is actually well-defined for all ∂t · times and consistently equals 0, which it should in this for the charge density ρ and the current vector-density case; cf. section 3.4. This already demonstrates the su- j is a four-dimensional analog of ∇ = 0. It im- periority of the formula (16) over Lorentz’ (1), which is plies that ρ and j can be expressed·B asB linear combi- ill-defined even in this simplest non-test charge situation. nation of first-order space and time derivatives of two Unfortunately, replacing Lorentz’ formula (1) with (16) complex three-dimensional fields, +i and i ; viz. does not convert Lorentz electrodynamics with point 4πρ = ∇ ( +i ) and 4πj = c∇×D( Bi ) HH−∂ ( E+i ). · D B HH− E − ∂t D B 5

The “4π” factor occurs for historical reasons, and the v(t) satisfying the speed limit v < c, the general solu- speed “c” at this point is just a conversion factor. Sorted tion to the system (22), (23) then| | is into real and imaginary parts these are precisely the pre- ret metric Maxwell equations, which we write as one pair of (t, s)= lw(t, s)+ ∇× (t, s), (26) D D Z homogeneous equations for and , B E ret 1 ∂ ∂ (t, s)= lw(t, s)+ ∇Z(t, s)+ (t, s). (27) (t, s)+ c∇× (t, s)= 0 , (20) H H c ∂tZ ∂tB E ∇ ret ret (t, s)=0 , (21) Here, lw and lw are the Li´enard–Wiechert fields (6) ·BB and (7D), and theH vector potential field (t, s) R3 and and one pair of inhomogeneous equations for and , a scalar potential field Z(t, s) R generateZ the∈ general D H solution to the associated homogeneous∈ system. Note the ∂ ∇× ∂t (t, s)+ c (t, s)=4πj(t, s) , (22) sign difference between the homogeneous ( , ) system − D H B E ∇ (t, s)=4πρ(t, s) . (23) and the homogeneous system associated with the ( , ) · D field pair. D H ret Note that the constraint equations (21) & (23) only im- We remark that +i = lw +∇×( +i ) and that ret ∇D B D 1 ∂ Z A pose on the initial data (0, s) and (0, s) which need to i = lw + (Z + iA)+ c ∂t ( + i ). be supplied when viewingB (as we willD do) (20) & (22) as HH− E H Z A initial value problems for and , respectively. To see 2. Gauge invariance this for (21), take the divergenceB D of (20); for (23), take the divergence of (22) and the time derivative of (23), As is well-known, the r.h.s.s of (24), (25) are invariant and recall (19). under the gauge transformation The pre-metric Maxwell equations are familiar from Maxwell’s theory of electromagnetic fields in material (t, s) (t, s)+ ∇Υ(t, s), (28) media, though here they are used for fields sourced by A 7→ A A(t, s) A(t, s) 1 ∂ Υ(t, s). (29) point charges in an otherwise empty space. 7→ − c ∂t Similarly, the r.h.s.s of (26), (27) are invariant under the 1. Their general solution for point charge sources gauge transformation

The pre-metric Maxwell equations are easily solved if (t, s) (t, s)+ ∇℧(t, s), (30) Z 7→ Z the charge density ρ(t, s) = eδq(t)(s) and the current 1 ∂ ℧ − Z(t, s) Z(t, s) c ∂t (t, s). (31) vector-density j(t, s) = eδq(t)(s)v(t), provided that 7→ − t q(t) is continuously differentiable− so that the conti- 7→ The gauge transformations can be merged in complex nuity equation (19) is automatically satisfied in the sense notation: the addition of a four-dimensional “pseudo gra- of distributions, and provided that v(t)

As shown by Mie [28], [29], in a Lorentz co-variant Since mathematically the pre-metric Maxwell field electrodynamics the vacuum law follows from a Lorentz- equations are quite naturally interpreted as a pair of evo- scalar Lagrangian (density) . The notion of Lorentz lutionary equations (20) & (22) for the fields and , invariance makes it obviousL that the spacetime metric with initial data which are constrained by (21)B & (23),D it enters at this point. In the orthodox version depends is desirable to rather express the field pair ( , ) in terms only on the Lorentz invariants 2 2 andL ( )2, of the pair ( , ). Happily (36), (37) canE beH converted but Lagrangians which in addition|EEE| depend− |BBB| on the LorentzEE·BB into B D invariant (∇ )2 ∇× 1 ∂ 2 have also been consid- ·EE −| BB− c ∂tEE| 1 ered in the literature (see below). The fields and are b2 ( ) D H = BB− DD× DD×BB , (38) in either case obtained by functional differentiation from H 1 2 2 1 2 3 1+ b2 ( + )+ b4 the action = d sdt, viz. = δE and = δB . |BBB| |DDD| |BB × DD| A L D 2A 2 H − A2 q 1 If depends only on the invariants and ( ) , 2 ( ) L R |EEE| −|BBB| EE·BB = DD− b BB× B × D . (39) this is equivalent to conventional partial differentiation of E 1 2 2 1 2 1+ 2 ( + )+ 4 the Lagrangian density, viz. = ∂E and = ∂B . b |BBB| |DDD| b |BB × DD| We next list the field LagrangiansD andL theH implied− elec-L q tromagnetic vacuum laws for the ML, the MBI, and the Complemented with (38) & (39) the pre-metric Maxwell MBLTP field equations, in historical order. field equations (20) & (22) turn into the Maxwell–Born– Infeld (MBI) evolution equations for the fields and , their initial data being constrained by (21) & (B23). D 1. Schwarzschild’s field Lagrangian and Maxwell’s vacuum In the absence of any sources the initial value problem law for the MBI field equations is globally well-posed for clas- sical initial data with sufficiently small energy [31]. How- Schwarzschild’s [30] Lagrangian, given by ever, it is also known that certain smooth plain-wave data 1 2 2 can lead to a singularity after a finite time, see [32, 33]. 4π S = , (34) L 2 |EEE| − |BBB| Speck in his thesis showed that this can be extended to yields Maxwell’s “law of the pure ether,”
(32) & (33), finite energy data which coincide with such plane wave obeyed by the Maxwell–Lorentz fields. We already re- data on a sufficiently large bounded domain in space. It viewed the Maxwell–Lorentz field equations in the intro- is not known whether such blow-up in finite time hap- duction. pens for all finite-energy data in an open neighborhood Even though (34) is not an admissible field Lagrangian of these finite-energy “local plane-wave” type data. for a classical electrodynamics with point charges, the For the MBI field equations with fixed point charge success of the Maxwell–Lorentz field equations in the sources it has been shown [34] that a unique finite en- realm of weak-field phenomena (i.e. far away from the ergy electrostatic solution exists which is real analytic hypothetical point charge sources) and the low-frequency except at the locations of the point charges; this holds / long wavelength regime (i.e. visible light, infra-red, for any finite number N of point charges with arbitrary radio waves and such) suggests that every admissible La- signs, magnitudes, and placements. However, it is not grangian must reduce to it in the weak-field and low- yet known whether the nonlinear BI law leads to an at frequency / long wavelength regime. least locally well-posed initial value problem for a physi- cally interesting class of Maxwell–Born–Infeld fields with point charge sources. 2. The Born–Infeld field Lagrangian and vacuum law

The Born–Infeld field Lagrangian [20], given by 3. Bopp’s field Lagrangian and vacuum law In the 1940s Bopp, Land´e& Thomas, and Podolsky ar- 2 4 2 2 2 2 4π BI = b b b ( ) ( ) (35) L − − |EEE| − |BBB| − EE·BB gued that a more accessible linear vacuum law is available q if one is willing to admit higher-order derivative electro- yields the Born–Infeld (BI) law of the electromagnetic magnetic field equations. Bopp [22] obtained the equa- vacuum, tions from a Lagrangian given by 1 b2 ( ) 1 2 2 = BB− BB·EE E , (36) 4π BLTP = 2 (40) H 1 2 2 1 2 L |EEE| − |BBB| 1 2 ( ) 4 ( ) b b 1 1 ∇ 2 ∇× 1 ∂ 2 − |EEE| − |BBB| − EE·BB + κ2 (
) q 1 2 ·EE − | BB− c ∂tEE| + 2 ( ) = E b BB·EE B , (37) h i D 1 2 2 1 2 which yields the Bopp–Land´e–Thomas–Podolsky 1 2 ( ) 4 ( ) − b |EEE| − |BBB| − b EE·BB (BLTP) electromagnetic vacuum law q expressing the pair ( , ) in terms of the pair ( , ). (t, s)= 1+ κ−2 (t, s) , (41) The parameter b is Born’sD H field strength constant. InB theE H B κ−2 limit b the BI law converges to Maxwell’s law. (t, s)= 1+ 
(t, s) ; (42) → ∞ D E
7

 −2 2 ∂ ∂2 here, c ∂t ∆ is the classical wave operator. The ( ∂t )(0, s) and ( ∂t2 )(0, s). Thus, (41), while still not parameter≡ κ is “Bopp’s− reciprocal length” [22]; see [35] anB evolution equationB for , is now an elliptic vector for empirical constraints on κ. The singular limit κ Helmholtz equation for (0B, s), which has a unique so- of the BLTP law yields Maxwell’s law. → ∞ lution that vanishes at spatialB infinity, thus determining The pre-metric Maxwell field equations (20) & (22) the initial (0, s) completely. B and (21) & (23), when supplemented by the BLTP law We next collect the pertinent formulas for the field mo- of the vacuum (41) & (42), become the Maxwell–Bopp– mentum vector-densities, then show that there are field Land´e–Thomas–Podolsky (MBLTP) field equations. Dif- initial data, satisfying the Maxwell constraint equations, ferent from the Maxwell–Lorentz and Maxwell–Born– for which (16) is initially well-defined. Lastly we address Infeld field equations, they are higher-order derivative the electrodynamical admissibility of the vacuum laws. field equations, requiring initial data not only for and , but in addition also for and ∂ =: ˙ . B D E tE E We pause for another moment and comment on the C. The electromagnetic field energy-momentum asymmetrical role played by the pair of equations (41)& density (42), despite their symmetric appearance. When judged A Lorentz invariant field Lagrangian also determines in their own right, (42) is a second-order evolution equa- the field energy-momentum density. For the Maxwell– tion for the electric field , given , and (41) is a second- ML E D Lorentz (ML) field theory, the field energy density ε order evolution equation for , given . However, since ΠML B H and field momentum vector-density are of course (41) and (42) are coupled with the pre-metric Maxwell well known and given by evolution equations (20) & (22) for the fields and B D (constrained by (21) & (23)), appearances are misleading ML 1 2 2 4πε = 2 + , (43) in the case of (41). A well-defined initial value problem |BBB| |DDD| for the fields is obtained only if (20) & (22) and (42) are
ML treated as genuine evolution equations for the fields , , 4πcΠ = × . (44) D B and , while (41) is not treated as an evolution equationB D for E — against all appearances. Recall that = in Maxwell–Lorentz field theory. B For the Maxwell–Born–InfeldD E (MBI) field theory, the Indeed, given the field initial data (0, s), also E field energy density εMBI and field momentum vector- ∇× (0, s) is fixed initially, so (20) yields ˙ (0, s). And MBI E B density Π are given by given the field initial data ˙ (0, s), also ∇× ˙ (0, s) is fixed E E initially, so the time derivative of (20) yields ¨(0, s). 4πεMBI = b4 + b2( 2 + 2)+ 2 b2 , (45) And with the initial data (0, s) given, also ∆B (0, s) |BBB| |DDD| |BB × DD| − is fixed, and then r.h.s.(41) isB completely determinedB ini- p 4πcΠMBI = × . (46) tially. Thus (41) defines initially in terms of and D B its second partial derivatives;H note that this also impliesB ∇ For the Maxwell–Bopp–Land´e–Thomas–Podolsky that = 0. And then, with so defined initially, MBLTP and the· particle’s H initial position andH velocity given, (22) (MBLTP) field theory, the field energy density ε and field momentum vector-density ΠMBLTP are given by now yields ˙ (0, s). Lastly, with the initial field data D ( , ˙ )(0, s), and (0, s) given, ¨(0, s) is initially deter- MBLTP 1 2 2 4πε = + 2 + (47) minedE E by (42). D E B · H E · DD− |BBB| |EEE| 1 2 1 ∂ 2 2 + ×
, This scheme now propagates in time, i.e. (41) remains − 2κ ∇·EE ∇ BB− c ∂tE the defining equation for in terms of and its second h
i partial derivatives, while (H20), (22), andB (42) are genuine ΠMBLTP × × × evolution equations for , , and . In [36] it is shown 4πc = + (48) B D E D B E HH−EE B that MBLTP field initial data ( , , , ˙ )(0, s) launch a 1 × 1 ∂ κ2 . unique global distributional solutionB D ofE theE MBLTP field − ∇·EE ∇ BB− c ∂tE equations, conditioned on the motions being given.
  There is a small variation on this theme, which takes D. Field data yielding an electromagnetic force advantage of the convenience of having the general dis- initially tributional solution of the pre-metric Maxwell field equa- tions for the pair ( , ) available with (26) & (27). Thus, Consider first the ML and MBI field theories. Both prescribing the motionD H t q(t) for t 6 0 conveniently, operate with the same formula for the field momentum though twice continuously7→ differentiable with subluminal density, (44) respectively (46). Initial data for the fields velocity v(t), and choosing smooth and spatially rapidly and compatible with the constraint equations (21) ∂ B D decaying fields Z(0, s), (0, s), and ( ∂t )(0, s), such & (23) for which (14) is initially well-defined are easily that ∇ = 0, equationsZ (26) & (27) fixZ (0, s) and obtained as follows. · H D ∇ 1 ∇ 1 ∂ (0, s). Prescribing also ( , ˙ )(0, s) fixes ∇× (0, s) and Set (0, s) = e s s Z(0, s) ( )(0, s) H E E E D | − (0)| − − c ∂tZ ∇× ˙ (0, s) initially, so (20) and its time derivative yield and (0, s)= ∇× (0, s), with Z(0, s) and ( 1 ∂ )(0, s) E B A c ∂tZ 8 and (0, s) smooth and rapidly decaying at spatial in- also ∇× (0, s) is smooth and so is ∇× (0, s). Given in finityA together with their derivatives. Then ( × )(0, s) additionD the assumed decay at spatial B, steps (a), (b), is integrable over R3, i.e. pfield(0) exists. D B and (c) from section II.A are now justified∞ to manipulate As to the derivative of pfield(t) at t = 0, consider first (16) and yield the ML field theoy. With the above choice of initial data,

em ×∇× ×∇× 1 × 3 f (0) = + +4π c j (0, s)d s, (49) R3 B B D D B Z   which is well-defined. Note though that (ρ )(0, s) is not Consider next the MBI field theory. If we formally well-defined, so that one cannot apply stepsD (d) and (e) of carry out steps (a), (b), and the analog of (c) of section section II.A and arrive at the Lorentz formula for f em(0). II.A, we get

em ×∇× ×∇× 1 × 3 f (0) = + +4π c j (0, s)d s, (50) R3 B H D E B Z   which may or may not be well-defined, depending on point charge source of certain classical electromagnetic (0, s). Clearly the j× term is the same as in the ML fields, whereas the Lorentz formula (1) fails to be well- Bsetup, pairing a δ distributionB with a smooth test func- defined. This is encouraging but does not suffice to ac- tion. The ×∇× term is also well-defined, and inte- complish our goal, which is to show that a generically grable, becauseD theE BI law (39) guarantees (i) that is well-defined, in fact well-posed and physically interesting uniformly bounded whenever is, and (ii) that for|EEE| our classical electrodynamics with point charges is possible. initial data (0, s) is differentiable|BBB| everywhere except at Our next step thus is to inquire into field evolutions for s = q(0), havingE uniformly bounded partial derivatives which (16) remains well-defined over time. 1 which (iii) decay not slower than r2 at spatial infinity. However, unless (0, s) vanishes at s = q(0), (38) yields s B 1 s q that (0, ) has a r2 singularity at = (0), and it is 1. The ML vacuum law is electrodynamically inadmissible easyH to check that its curl will then generally blow up 1 like 3 , which is too strongly. If (0, s) vanishes like For the sake of completeness, we note that the electro- r B s q(0) or faster at s = q(0), then the ×∇× term static special case leads to a consistent proper solution of | − | B H is well-defined, and integrable. Note that (1) is still not Lorentz electrodynamics in which the Maxwell–Lorentz well-defined even with such idealized initial data because field equations are coupled with (16) instead of (1). In- (0, s) is not well-definable at s = q(0). deed, assuming that q(t)= q(0) and v(t)= 0 for all time, E Since typically the ( ×∇× )(0, s) term is not abso- field initial data (0, s) = e∇ 1 and (0, s) = 0 B H D |s−s(0)| B lutely integrable over a neighborhood of s = q(0), one propagate in time unchanged, viz. (t, s)= (0, s) and typically is not allowed to carry out the manipulations (t, s) = 0 for all time. But thenD ( × )(t,Ds) = 0 for by means of which one arrives at (50). Nevertheless, Ball time, and so by (16) also f em(tD) =B0 for all time, 2 since (46) has at most 1/r singularities, (16) may still consistent with v(t)= 0 and q(t)= q(0) for all time. be well-defined for MBI fields with a regularly moving Unfortunately this very special situation does not have point charge. This is yet to be sorted out, though. an open dynamical neighborhood in Lorentz electrody- Lastly, as to the MBLTP field theory, it is not neces- namics even when (1) is replaced by (16). Suppose we sary to separately discuss the possibility of an initially perturb the static field data (0, s) = e∇ 1 and D |s−s(0)| well-defined force because below we will report that we (0, s) = 0 by replacing the vanishing magnetic ini- actually were able to obtain a generically well-defined Btial induction by a smooth and compactly supported force which persists into the future. (0, s) = ∇× (0, s) such that (0, s)×∇× (0, s) is Bintegrable overAR3. Even leavingB the particleB initial E. Electrodynamic admissibility of the vacuum laws data as before, by (49) the point charge will now feel a nonzero initial force and begin to move. But then by the The examples discussed in the previous subsection Maxwell–Lorentz field equations the magnetic induction 1 demonstrate that the definition (16) can yield a well- will not evolve into a smooth (t, s); a r2 singularity is defined expression for the electromagnetic force on a formed after an arbitrarily shortB time span after the ini- 9 tial instant, so × is only integrable at t = 0, and the coincides with the tangent world line of the actual point force (16) doesD notB remain well-defined. This disqualifies charge at t = 0, by the retarded Li´enard–Wiechert-type the Maxwell–Lorentz field equations. fields of a fictitious point charge whose subluminal C1,1 world line merely is tangent to the world line of the actual point charge at t =0, does not result in more general ini- 2. The BI vacuum law may be electrodynamically tial data, then, because the difference of two such fields at admissible time t =0 is regular enough and satisfies the source-free MBLTP field equations at t =0. Everything we wrote about the electrostatic special case in Lorentz electrodynamics, with (1) replaced by (16), carries over to Born–Infeld electrodynamics. How- The proof in [36] of the electrodynamical admissibility ever, this time there may be an open neighborhood of of the BLTP law of the electromagnetic vacuum consists initial data for which the ensuing MBI field evolution in showing that the electrodynamically admissible initial with a large open set of assumed motions of their point data stipulated above launch field evolutions for which source leads to a well-defined (16) as time goes on. To the force (16) is well-defined for an open set of physically be sure, no such result has been proven rigorously yet, acceptable motions. This proof is greatly facilitated by but the author is optimistic that the BI law is admissi- the fact that the electromagnetic force in the BLTP vac- ble, i.e. that an open set of solutions to the MBI field uum can be computed explicitly; the details are given equations with point charge sources has integrable field in [36]. For the convenience of the reader the result of momentum densities, and that the total field momentum this computation is summarized next; it has also been is differentiable in time — for not too long time intervals. announced in a conference proceedings, see [37].

3. The BLTP vacuum law is electrodynamically admissible

In [36] we establish the electrodynamic admissibility of IV. THE ELECTROMAGNETIC FORCE IN THE the BLTP law, which here we summarize for the simpli- BLTP VACUUM fied case of fields with a single point charge source. We begin with a definition. By the linearity of the MBLTP field equations the so- Definition III.1 Electrodynamically admissible initial lution launched by the above stated type of initial data data are of the following form: ( , , , ˙ )(0, s) = decomposes into the pertinent sum of a vacuum field plus B D E E a Li´enard–Wiechert(-type) field. The vacuum field need 1 ( , , , ˙ )(0, s). Here, ( , , , ˙ )(0, s) j=0 Bj Dj E j E j B0 D0 E 0 E 0 not be represented explicitly for it suffices to know that is the t = 0 evaluation of a C0,1 finite-energy source- P it has the required regularity. The field solutions 1(t, s) free MBLTP solution which is globally bounded by D and 1(t, s) for t> 0 are for a.e. s q(t) > 0 given by ( , , , ˙ ), having global Lipschitz constants LB H | − | B0 D0 E0 E0 B0 the Li´enard–Wiechert fields (6) and (7), and LE . Moreover, ( , , , ˙ )(0, s) is the “co- E 0 B1 D1 E 1 E 1 moving electromagnetic field” at t =0 of a fictitious point ret 1(t, s)= lw(t, s) , (51) charge whose world line coincides with the tangent world D D ret (t, s)= lw(t, s) . (52) line of the actual point charge at t =0. H1 H

Remark III.2 Replacing the “co-moving electromag- The MBLTP field solutions 1(t, s) and 1(t, s) for t > 0 netic field,” of a fictitious point charge whose world line are given by (cf. [36]) B E

ret s t (t, ) κ 2 ′ 2 ′ 2 κ2 1 n(q,s)−v/c κ2 J2 √c (t−t ) −|s−q(t )| ′ ′ ′ ′ 1(t, s)= e n q s v + e 2 ′ 2 s q ′ 2 c (s q(t ) v(t )(t t )) dt , (53) E − 2 1− ( , )· /c ret c (t−t ) −| − (t )| − − − Z−∞
ret s t (t, ) κ 2 ′ 2 ′ 2 κ2 1 v×n(q,s)/c κ2 J2 √c (t−t ) −|s−q(t )| ′ × ′ ′ 1(t, s)= e 2 1−n(q,s)·v/c + e c2(t−t′)2−|s−q(t′)|2 v(t ) (s q(t )) dt ; (54) B − ret −∞ − Z

note that the time integrations from to 0 here do not expressions (53) and (54) reduce at t = 0. involve some unknown past motion but−∞ only the auxiliary straight-line motion which encodes the Lorentz-boosted With the help of these solution formulas the electro- electrostatic MBLTP field of the point charge, to which magnetic force of the MBLTP field on its point charge source can be computed as follows. Since each electro- 10 magnetic field component is the sum of a vacuum field a “self”-field force in BLTP electrodynamics. Thus, (16) and a sourced field, the bilinear ΠMBLTP decomposes into is given by a sum of three types of terms, the vacuum-vacuum terms, the source-source terms, and the mixed vacuum-source f em(t)= f vacuum[q, v](t)+ f source [q, v; a](t) (55) terms. The vacuum-vacuum contribution to r.h.s.(16) vanishes because the total momentum of a vacuum field where is conserved; the vacuum-source contribution to r.h.s.(16) f vacuum[q, v](t)= e (t, q(t)) + 1 v(t) (t, q(t)) yields the force on the point source due to the vacuum − E 0 c ×BB0 field; lastly, the source-source contribution to r.h.s.(16) is (56) is the Lorentz force (1) evaluated with a vacuum field (i.e. a “test particle contribution” to the total force), and

d source ΠMBLTP ΠMBLTP 3 f [q, v; a](t)= source (t, s) source (0, s q0 v0t) d s (57) −dt B (q ) − − − Z ct 0  

ΠMBLTP with source given by (48) with ( 1, 1, 1, 1) in place complete actual past motion of the point particle and can- of ( , , , ). B D E H not be used to study its initial value problem in which only B D E H the particle’s initial position and velocity are prescribed Remark IV.1 We re-emphasize that there is no such (given the initial fields). thing as the self-field force in electrodynamics; only the total force, i.e. the sum at r.h.s.(55), has an absolute The “self”-field force can be evaluated using retarded meaning (in the chosen Lorentz frame). In particular, spherical coordinates (r, ϑ, ϕ) to carry out the d3s inte- our “self”-field force does generally not agree with the grations over the ball Bct(q0), after which one can differ- expression studied in [39] and [38], which depends on the entiate w.r.t. t. This yields

2 source e [2] [2] f [q, v; a](t)= Z (t,t)+ Z ◦ (t,t) (58) 4π − ξ ξ  t 2−k [k] r [k] r r 1−k r c (2 k) Z t,t Z ◦ t,t (t t ) dt − − ξ − ξ − 06k61 Z0 P t h

i 2−k ∂ [k] r ∂ [k] r r 2−k r c Z t,t Z ◦ t,t (t t ) dt , − ∂t ξ − ∂t ξ − 06k62 Z0  P h

i

◦ [2] [k] r where ξ(t) (q, v, a)(t) and ξ (t) (q + v t, v , 0), where Z (t,t) := lim r Z t,t , and where ≡ ≡ 0 0 0 ξ t →t ξ
2π π Z[k] t,tr = 1 1 v(tr) cos ϑ π[k] t, q(tr)+ c(t tr)n sin ϑdϑdϕ , (59) ξ − c ξ − Z0 Z0


with n = (sin ϑ cos ϕ, sin ϑ sin ϕ, cos ϑ) a normal vector to the retarded sphere of radius r = c(t tr). Also, setting −

κ 2 ′ 2 ′ 2 ′ J1 √c (t−t ) −|s−q(t )| Kξ(t , t, s) := , (60) √c2(t−t′)2−|s−q(t′)|2
κ 2 ′ 2 ′ 2 K ′ J2 √c (t−t ) −|s−q(t )| ′ ′ ′ ξ(t , t, s) := 2 ′ 2 ′ 2 (s q(t ) v(t )(t t )) , (61) c (t−t ) −|s−q(t )| − − −
11 the π[k](t, s) with k 0, 1, 2 read, for s = q, ξ ∈{ } 6

1 n(q,s)− 1 v × 1 v×n(q,s) [0] κ4 ( c ) ( c ) πξ (t, s)= 2 (62) − 4 1− 1 v·n(q,s) " c #ret tret (t,s) 1 n(q,s)− 1 v
ξ κ4 c × v ′ ×K ′ s ′ + 1− 1 v·n(q,s) (t ) ξ(t , t, )dt 2 c ret Z−∞ h i tret(t,s) 1 1 v×n(q,s) ξ κ4 c × K ′ s ′ 1− 1 v·n(q,s) c ξ(t , t, )dt − 2 c ret −∞ h ret i Z ret tξ (t,s) tξ (t,s) 4 ′ ′ ′ ′ ′ κ cKξ(t , t, s)dt × v(t )×Kξ(t , t, s)dt − −∞ −∞ Z ret Zret tξ (t,s) tξ (t,s) 4 ′ ′ ′ ′ ′ κ c Kξ(t , t, s)dt Kξ(t , t, s)v(t )dt , − Z−∞ Z−∞

n(q,s)×[ n(q,s)− 1 v ×a] · 1 v n(q,s)− 1 v ×a [1] κ2 ( ( c ) ) c × ( c ) πξ (t, s)= n(q, s) 4 + n(q, s) 3 (63) − c2 1− 1 v·n(q,s) 2c2 1− 1 v·n(q,s) " c c #ret ret s 1 ×
tξ (t, )
2 (n(q,s)− c v) a ′ ′ ′ κ n(q, s)× 3 × v(t )×Kξ(t , t, s)dt 2 1 − c 1− c v·n(q,s) −∞ " #ret Z ret s
1 × tξ (t, ) 2 (n(q,s) − c v) a ′ ′ + κ n(q, s)× n(q, s)× 3 × cKξ(t , t, s)dt 2 1 c 1− c v·n(q,s) −∞ "  #ret Z ret s tξ (t, )
κ3 1 ′ ret ′ ′ + 1 v n q s Kξ(t , t, s) v(tξ (t, s))) + v(t ) dt , 1− c · ( , ) ret Z−∞ h i  

n(q,s)− 1 v × 1 v×n(q,s) [2] κ2 1 1 1 2 ( c ) ( c ) πξ (t,s)= 2 c v 1 c2 v 4 (64) 1− 1 v·n(q,s) 1− 1 v·n(q,s) − " c − − c # h i ret ret tξ (t,s) 2
n(q,s) − 1 v
κ2 1 × c × K ′ ′ + 1 c2 v n(q, s) 3 c ξ(t , t, s)dt 1− 1 v·n(q,s) " − c # −∞ h i ret Z ret tξ (t,s) 2 n(q,s) −1 v
2 1 c ′ K ′ ′ κ 1 2 v 3 × v(t )× ξ(t , t, s)dt , c 1 − − 1− c v·n(q,s) −∞ " #ret Z h i
q ˜ v ˜ a ˜ ˜ ret s and ret means that (t), (t), (t) are evaluated at t = tξ (t, ). Note that r.h.s.(58) vanishes at t = 0.

Several of the spherical angular integrations can be V. BLTP ELECTRODYNAMICS HAS NO ... carried out explicitly in terms of well-known functions; q -PROBLEM see [36]. Here we are content with the remark that for 1,1 C motions t q(t) (which for t 6 0 coincide with Having established that the BLTP law is electrody- the auxiliary straight7→ line motion t q(0) + v(0)t em- 7→ namically admissible, in the sense that the solutions of ployed to define the field initial data, and should not be the MBLTP field equations for prescribed C1,1 motions confused with any actual motion for t< 0), one can eas- q source t (t) yield a field momentum vector density which ily show that all the terms in f [q, v; a](t) are well- is7→ integrable over space, and its integral differentiable in defined. Since (56) is also well-defined for the regular time, we now explain that the so obtained electromag- vacuum field, the electrodynamical admissibility of the netic force (55), with f vacuum[q, v](t) given by (56) and BLTP vacuum law follows. f source [q, v; a](t) by (58)ff, when substituted at r.h.s.(13), yields a well-posed initial value problem for point charge motion, and thus a well-posed classical BLTP electrody- namics with a point charge. It is understood here that the relativistic velocity-momentum relation (12) is assumed, 12 with mb = 0. This well-posedness result is a special case The careful proof in [36] fills many pages. of a result6 proved in [36] for an arbitrary finite number Essentially as a corollary of the above Proposition of point charges. We here explain the main idea of the we obtain the well-posedness result for the joint initial proof for fields having a single point charge source. value problem of the MBLTP field and its point charge A key feature of the total electrodynamical force in source. Indeed, now substituting α[q( ), p( )](t) for a(t) a BLTP vacuum, f vacuum[q, v]+ f source [q, v; a], as indi- in f source[q, v; a], Newton’s equation of· motion,· supple- cated by our notation, is the dependence on only q, v, a; mented with the relativistic velocity-momentum relation d a third- (and higher-)order time derivative of q(t) does (66) and the definition v(t)= dt q(t) of the velocity, can not show up. Therefore, already purely formally, our be formally integrated to become a fixed point map for initial value problem for the point charge motion is of a curve in the phase space of the point charge, viz. for second order, as desired. some T > 0 and t [0,T ), Next, inserting (12) at l.h.s.(13) and carrying out the ∈ differentiation we obtain a familiar expression which can t 1 p(t′) q(t)= q(0) + dt′, (68) be rewritten as a regular matrix acting on the vector a, |p(t′)|2 0 mb Z 1+ m2 c2 the matrix depending only on p (and mb), not on q and b not on a. Applying the inverse of this matrix at both t q vacuum ′ ′ sides of (13) (with the force in a BLTP vacuum in place), p(t)= p(0) + f [q, v](t ) dt (69) 0 our equation of motion becomes Z t + f source q, v; α[q( ), p( )] (t′) dt′, vacuum source · · a = W [p] f [q, v]+ f [q, v; a] (65) Z0 ·     where v at r.h.s.(69) is given in terms of p through (66). where the velocity v is expressed in terms of the momen- The following theorem is a special case of the N-body tum p by inverting (12), i.e. result proved in [36]. 1 p THEOREM V.2 v = ; m =0, (66) Given q(0) and p(0) and the stipu- 2 b mb |p| 6 lated MBLTP field initial data, there is a T > 0 such that 1+ m2 c2 b the fixed point problem (68)&(69) has a unique C1,1 so- q q p and where lution t ( , )(t) for t (0,T ) extending continuously to [0,T )7→. Moreover, if in∈ a finite time the particle does 1 1 p p not reach the speed of light or infinite acceleration, then W [p] := 1 ⊗ . (67) 2 2 2 2 T = . In any event, total energy-momentum conserva- mb |p| − m c + p 1+ 2 2 b ∞ mbc  | |  tion holds. q d Remark V.3 vacuum Together with the usual definitions a(t) := dt v(t) and When f [q, v] is replaced by an exter- d nal, smooth and short-ranged force field, the dynamics is v(t) := dt q(t) this is a complicated, nonlinear and im- plicit, second-order differential-integral equation of mo- global, i.e. T = . This was shown in [40]. ∞ tion for the position of the point charge. We approach the problem stepwise. Temporarily ig- Remark V.4 For the BLTP electrodynamics it is easy d d to generalize the single-particle formulation to the N-body noring the relationships a(t)= dt v(t) and v(t)= dt q(t), and instead treating the maps t q(t) and t p(t) and formulation. By the linearity of the MBLTP field equa- t a(t) as a-priori unrelated for7→ t > 0 (of7→ course, for tions we can associate each particle with its own Li´enard– t 67→0 the maps are determined by the stipulated auxil- Wiechert(-type) MBLTP field, and thus a “self”-field iary motion), we note that then the acceleration t a(t) force, same expressions as before except that the index 1 is source 7→ enters the BLTP “self” force term f [q, v; a] only in replaced by j , say, also to be attached to position, velocity, a linear fashion. Therefore, when treating t q(t) and momentum, and acceleration vectors, and to the charge t v(t) as given, (65) becomes a linear integral7→ equa- and mass parameters. The total MBLTP field is the sum tion7→ for the acceleration t a(t). Better yet, inspection of a vacuum field of the kind considered above plus all shows that it is a linear7→ Volterra integral equation for the Li´enard–Wiechert(-type) fields. The total force now which we prove the following key result, see [36]. gets an extra contribution in form of the Lorentz force (1) on particle j exerted by the Li´enard–Wiechert-type fields of all the particles but j . This requires extra regularity 0,1 PROPOSITION V.1 Given C maps t q(t) and estimates for all n and n, n 1, ..., N , which are t p(t), with Lip(q) = v, Lip(v) = a, and7→ v(t) 6 established in [36].E In additionB to∈ { the avoidance} of the v

Remark V.5 Above we noted already that our “self”- VI. THE ELECTROMAGNETIC FORCE: field force r.h.s.(58) vanishes at the initial instant t =0. N POINT CHARGES Thus in a BLTP vacuum the inertia of the particle is initially entirely due to its bare mass! In the case of BLTP electrodynamics it is straightfor- ward to generalize the formula for the electromagnetic To summarize, the work of [36] on BLTP electrody- force on a point charge from when there is only a sin- namics demonstrates that it is feasible to set up a classi- gle charge to when in total N point charges are present. cal electrodynamics with point charge sources as a well- This is possible because the linear pre-metric Maxwell posed joint initial value problem for the fields and the equations are then complemented with the linear BLTP particles, which is of second order in the particle posi- vacuum law as closure relation. When nonlinear vacuum tions. No Landau–Lifshitz type approximation has been laws are used, for instance the Born–Infeld law, all these ... invoked because the infamous q -problem does not show linear algebra-based conclusions are not available. In the up; no negative infinite bare mass renormalization [10], following we first give a general distributional definition no additional comparison axioms [41] (cf. also [42]), and of the electromagnetic force on a point charge source no “separating off of singularities” [43] are invoked. when N point charges are present. Then we extract from this definition the N-body analog of the earlier given ba- Whether BLTP electrodynamics is already a physically sic definition (16) of the electromagnetic force when only acceptable classical theory is a different question. Partic- a single charge is present. ularly embarrassing is the fact that the triumph of avoid- ing unphysical third- (and higher-)order time derivatives of the point charge’s position q(t) in its force law is paid A. Distributional definition of the electromagnetic for by a high prize, namely by introducing (presumably) force unphysical higher-order time derivatives in the electro- magnetic field equations! As a consequence, four MBLTP In the electromagnetic field-theory part of electrody- fields, namely , , , ˙ , require initial data. However, namics one already implements the fact that the sys- B D E E according to phenomenological electromagnetism, once tem of N moving charged point particles is associated the fields (0, s) and (0, s) are determined / prescribed with a distribution-valued four-vector field on spacetime, B D initially (constrained by (21) and (23)),1 one does not namely (cρ, j)(t, s), the inhomogeneity term in the linear ˙ have any freedom left to also choose 0(0, s) and 0(0, s), pre-metric Maxwell field equations. It is only natural to yet the MBLTP field equations do requireE such aE choice. also write the mechanical quantities associated with the (Incidentally, neither the founding fathers of the BLTP point particles as distribution-valued fields on spacetime theory, nor Feynman [44], nor recent authors [39], [38], and thereby treat particles and fields on an equal footing. seem to have been worried about these “loose ends” of The momentum vector-density this field theory.) Be that as it may, it is generally agreed Πcharge s p s upon, and mathematically realized in the structures of (t, )= n(t) δqn(t)( ) (70) the Maxwell–Lorentz field equations and of the Maxwell– n X Born–Infeld field equations, that once the fields (0, s) and the symmetric stress tensor field and (0, s) are prescribed, the initial field data areB fixed. D Therefore, to implement this rule also into the MBLTP 1 p (t) p (t) T charge(t, s)= n n δ (s) (71) field theory one needs a prescription which expresses the ⊗ 2 qn(t) mn |pn(t)| data (0, s) and ˙ (0, s) in terms of (0, s) and (0, s). n 1+ m2 c2 E E B D X n In [36] we show that this can be accomplished by pos- q jointly satisfy the local law of particle momentum balance tulating that the fields (0, s) and ˙ (0, s) maximize the E 0 E 0 field energy functional initially, given 0(0, s), 0(0, s), ∂ Πcharge (t, s)+ T charge(t, s)= f em(t)δ (s); B D ∂t n n qn(t) and given the Li´enard–Wiechert(-type) fields associated ∇ · (72) with the data q(0), v(0). Thus the initial field energy em P here, f n (t) is precisely the force term in (13), now un- is made as little negative as possible. One can make derstood for the n-th particle. this constraint Lorentz invariant by stipulating, e.g., that Similarly, for admissible vacuum laws, the electromag- the field energy maximization refers to the Lorentz frame netic field momentum vector-density Πfield and the sym- in which the total particle momentum vanishes. In the metric stress tensor of the fields2 T field jointly satisfy the Lorentz frame in which the initial value problem is for- local law of field momentum balance mulated the relevant constraint is obtained through a Lorentz boost. ∂ Πfield(t, s)+ T field(t, s)= Φ (t)δ (s). (73) ∂t n qn(t) ∇ · n P

1 In the physics literature on classical Lorentz electrodynamics, 2 To avoid awkward minus signs elsewhere, we define T field with one usually finds B(0, s) and E (0, s) prescribed, but recall that the opposite sign compared to the convention introduced by D = E (and B = H) in Lorentz electrodynamics. Maxwell in what nowadays is called the “Maxwell stress tensor.” 14

field The source (sink) terms, Φn(t), for the electromagnetic Similarly, the field energy density ε (t, s) and the field field momentum density & field stress are defined by momentum density Πfield(t, s) jointly satisfy the local law (73). For admissible vacuum laws (73) is indeed well- of field energy balance defined in the sense of distributions for fields with point charge sources at the q (t) which move with subluminal ∂ εfield(t, s)+ c2 Πfield(t, s)= Φ (t) v (t)δ (s). n ∂t n n qn(t) ∇ · n · velocities vn(t). The vector Φn(t) is the electromagnetic field momentum gained (lost) per unit of time at loca- P (81) Φ When the total energy density is postulated to be addi- tion qn(t) of the source (sink). Therefore n(t) has the physical dimension of a “force.” tive, i.e.

Now we postulate that the total momentum vector- field charge density of the interacting system of electromagnetic field ε(t, s) := ε (t, s)+ ε (t, s), (82) and its charged particle sources is given by the sum of the respective field and particle expressions defined earlier, then the postulated local law of total momentum conser- vation (77) now entails the local conservation law for the Π(t, s) := Πfield(t, s)+ Πcharge (t, s) (74) total energy, is the total momentum vector-density. Similarly, the ∂ ε(t, s)+ c2 Π(t, s)=0. (83) ∂t ∇ · symmetric total stress tensor is postulated to be the sum Similarly one can obtain the local conservation law for T (t, s) := T field(t, s)+ T charge(t, s). (75) total angular-momentum.

Our postulates (74) & (75) in concert with the two bal- ance laws (72) & (73) imply the local balance law for the total momentum vector-density, B. Integral formula for the electromagnetic force

em The distributional definition (78) of the electromag- ∂ Π(t, s)+ T (t, s)= Φ (t)+ f (t) δ (s). ∂t n n qn(t) Φ ∇ · n netic force, with n defined by (73), gives rise to the P
(76) following N-body analog of the one-body formula (16). Now postulating that the total momentum density and Let denote the Voronoi cell of the j-th point charge at Vj the stresses jointly satisfy the local conservation law for the initial time, and ∂ j its boundary. Let T > 0 denote the total momentum vector-density, the instant of time untilV which all N point charges re- main in their initial Voronoi cells. Then integrating (73) ∂ Π(t, s)+ T (t, s)= 0, (77) over yields, for t (0,T ) and j 1, ..., N , ∂t ∇ · Vj ∈ ∀ ∈{ } then by comparing (77) with (76), and invoking relativis- d f em(t)= Πfield(t, s)d3s (T field ν )(t, s)d2s, tic locality (spacelike separated events do not affect each j − dt − · j other), one deduces the identities ZVj Z∂Vj (84) em Φ and the initial force is defined as its limit when t 0. n & t > 0 (a.e.): f n (t) n(t). (78) 3 ↓ ∀ ∀ ≡− If N = 1, so that j = 1, we have 1 = R , the surface V This is the general distributional definition of the electro- integral vanishes, and in this case (84) reduces to (16). If T < then to go beyond T one can reset the clock magnetic force on a point charge source of the classical ∞ electromagnetic field when N point charges are present. to a new “initial” time, say T ǫ, and replace the initial Voronoi cells with those at time− t = T ǫ, and repeat. It may be seen as a zero-gravity implementation of what − Einstein–Infeld–Hoffmann surmised [45]. As shown explicitly for BLTP electrodynamics, so also In the next subsection we will extract the N-body ana- for BI electrodynamics one should be able to show that log of (16), the electromagnetic force when only a single with fixed initial data the r.h.s.(84) depends on the vec- tor functions t′ q (t′), t′ v (t′), and t′ a (t′) for charge is present. 7→ n 7→ n 7→ n t′ (0,t), with lim ′ q (t′)= q (0) and lim ′ v (t′)= We close this subsection by recalling that the energy ∈ t ↓0 n n t ↓0 n density of the charged particle distribution vn(0) given particle initial data, i.e. no higher-order time derivative of the position beyond the second one should ′ ′ ′ ′ 2 show up. Then with t qn(t ) and t vn(t ) consid- charge 2 |Pn(t)| 7→ 7→ ε (t, s)= m c 1+ 2 2 δQ (t)(s) (79) n mnc n ered given, the system of relativistic Newton equations n r X of motion (13) with (84) at its right-hand side becomes a system of generally nonlinear integral equations for the and its momentum vector-density (70) jointly satisfy the maps t′ a (t′) as functionals of the q ( . ) and v ( . ) local law of particle energy balance 7→ n n n (or pn( . )). Whenever this system has a unique solu- charge em q ∂ charge s 2 Π s f v s tion which depends Lipschitz continuously on the n( . ) ∂t ε (t, )+c (t, )= n (t) n(t)δqn(t)( ). ∇· n · and vn( . ) (or pn( . )), the electrodynamical initial value P (80) problem is locally (in time) well-posed. 15

VII. MOTION ALONG A CONSTANT which yields the total electromagnetic force on particle 1, ELECTRIC FIELD d f em(t)= ΠMBLTP(t, s)d3s e hom, (86) 1 − dt − E ZBct(q0) The problem of determining the classical dynamics of for t > 0. Technically we obtain (86) from (84) for j = 1. 3 a single point charge which moves along a static, spa- The point charge’s Voronoi cell 1 R in the limit tially homogeneous electric field (approximately achieved N , and (84) for j = 1 givesV (86→). We remark that by the field between the plates of a charged capacitor) the→hom ∞ term in (86) comes from the boundary integral has already been mentioned in the introductory section, in (84E ); it is easy to see that hom does not contribute d ΠMBLTP 3 E where we recalled that the Eliezer–Ford–O’Connell equa- to dt R3 (t, s)d s. Moreover, in (86) we have re- R3 tion of motion, and also its Landau–Lifshitz approxima- placed the domain of integration by the ball Bct(q0) R MBLTP tion, fail to account for the radiation-reaction on the mo- because Π (t, s) 0 outside of this ball. The “self”- tion and merely reproduce the test particle motion. We d ≡ΠMBLTP 3 field force q (t, s)d s is in this example − dt Bct( 0) now demonstrate that the initial-value problem for BLTP ΠMBLTP s 0 identical with r.h.s.(R 57), for (0, ) . electrodynamics formulated in this paper does take the To evaluate d ΠMBLTP(t, s)d3≡s we invoke radiation-reaction on the motion into account. − dt Bct(q0) Z[2] r 0 equations (58)ff withR ξ◦ (t,t ) . Due to the highly symmetrical setup these expressions≡ simplify drastically, For simplicity we restrict the discussion to the case although one still cannot carry out each and every inte- where the particle is initially at rest and surrounded by gration in terms of elementary functions — but this is its own electrostatic field and by the electrostatic field of not necessary for our purposes here. the capacitor. Note that this completely fixes the initial We now consider first the early time regime. Since, data for the field and for the particle. No “past hypoth- as noted above, the “self”-field force r.h.s.(58) vanishes esis,” about how these initial data were established, is at the initial instant t = 0, the initial force is identical needed. to the force e hom on a charged test particle. Since the “self”-field− forceE varies continuously with t > 0, it will remain small for a certain amount of time (which The textbook idealization of a single point charge we are not going to determine here). During this early placed in a truly uniform capacitor field is obtained as dynamical phase the motion of the particle is therefore a limiting case from our setup, as follows. Consider a well approximated by the test particle dynamics, with system of 2N + 1 charges, with j = 1 for the point elec- the inertia of the particle given by its bare mass. tron whose dynamics we are interested in, and N positive We now show that although the radiation-reaction and N negative singly charged particles distributed uni- force vanishes initially, it typically does not remain zero, formly over the respectively charged two capacitor plates. unlike the Eliezer–Ford–O’Connell radiation-reaction We take the formal N limit in which the capaci- force and its Landau–Lifshitz approximation. By “typ- tor plates become infinitely→ ∞ charged, but also infinitely ically” we mean that for almost all κ parameter values extended and separated, leaving a homogeneous, static the radiation-reaction force does not vanish (we do not vacuum field hom behind in which the point charge is 1 rule out that there might be special, isolated values of situated. WhileE the total electrical field energy diverges κ for which this might happen). For this purpose, as- in this limit, the field momentum is initially zero and re- sume to the contrary that the radiation-reaction force mains well-defined later on; also the particle momentum r.h.s.(58) would vanish for an open interval of κ values. vanishes initially for our data. The electromagnetic force Since r.h.s.(58)ff reveals that the radiation-reaction force on particle is derived from total momentum balance — 1 is an analytic function of κ, it follows that it then has to note that the total momentum of the single-particle-plus- vanish for all κ. This in turn means that each and every field system is not by itself conserved because this is only Maclaurin coefficient of its power series expansion in κ a subsystem of a formally “infinitely-many-particles-plus- ✟ has to vanish. ✟But The lowest-order term, which is κ2 field” system. Even though all the other charges have and can be evaluated exactly in closed form (see below),∝ been “moved to spatial infinity,” they still exert an influ- does ✟not✟ vanish, but the term κ3 does not, hence the ence on the remaining point charge through their field 1 radiation-reaction force cannot∝ typically vanish. hom. The balance equation for the momentum of the To compute the O(κ2) contribution, divide the expres- Esingle-particle-plus-field subsystem thus reads [k] κ2 κ sions for πξ by and take the limit 0. The only two terms which survive in the limit are those→ in the first line of r.h.s.(63) and r.h.s.(64), respectively. Carrying out the pertinent integrations in (59), and noting that r d the result only depends on t , not on t, so that the third (p (t)+ pfield(t)) = e hom, (85) line at r.h.s.(58)ff vanishes at O(κ2), we obtain dt 1 − E 16

1 v(t) c c2 1+ 1 v(t) f em(t)= e2κ2 2 ln c (87) 1 − 2 v(t) v(t) − v(t) 2 1 1 v(t) " | | | | − c # ✘ r 2 t 1 1✘✘r |v(t )| 1 r 3✄ 1 2✘✘ v(t ) 2 c ✓1 1+ v(t ) ✄ e2κ2 a( tr) − 3 c | | − c ✓ ln c | | cdtr + O(κ3) ✄2 v(tr) 2 |v(tr)|2 v(tr) ✓ 6 1 v r − 0 " 1 2 − − 1 c (t ) # Z | | − c | | ✭ − | | ✭✭✭✭ ✟✟ ✭✭✭ 1 1 2 2 ✟v(t) c ✭c ✭✭1 1+ c v(t) 3 = e κ✟ 1 1+✭✭1✭✭ ln +O(κ ) (88) −2 ✟ v  − v(✭t)✭✭ − v(t) 2 1 v  ✟ (t) ✭✭ 1 c (t) ! ✭ | |  | |  −  

The term in the first line at r.h.s.(87) is the contribution from the first line at r.h.s.(64), the term in the second line at r.h.s.(87) is the contribution from the first line at r.h.s.(63). Since for straight-line motion v(t) and a(t) are parallel, and a(t)= v˙ (t), one can carry out the time inte- gration in the second line at r.h.s.(87) in terms of elemen- tary functions of v, and a few algebraic manipulations then give (88). To O(κ2) this is the exact expression for our BLTP radiation-reaction force in this problem where the particle starts from rest. It vanishes. only as v 0 and as v c, and otherwise points against v. A| simi-|↓ lar computation| |↑ (to be published) reveals that the term κ3 does not vanish. This demonstrates that the BLTP ∝electrodynamical initial value problem accounts for the radiation-reaction on the motion of a point charge along a uniform electric field. In the vicinity of the initial time when |v(t)| 1 we FIG. 1. The velocity of a point charge, starting from rest c hom can expand and obtain, to leading order in v /c:≪ in a strong, constant applied electrostatic field E , as ✘✘ | | per test particle theory (dashed curve), and as per BLTP 2κ2✘✘✘ ✘2 ✘✘ 2 em✘✘ ✘1 e✘ ✘|✘v(t)| ✘κ3 electrodynamics with O(κ ) friction only (continuous curve). ✘f 1 (t)✚=✘ v(t) 1+✘✘O 2 +✘O( ). (89) − 6 c ✘ c (Ignore the continuous curve.) h  i Note that (89) is a radiation-friction force of the familiar “Newtonian friction” type (i.e. proportional to v). quate for demonstrating that the radiation-reaction does In Fig. 1 we show v in units of c as a function− of t not vanish identically in the BLTP version of this stan- in units of m c/e2κ2 both for the test particle motion b dard textbook-type problem. Moreover, we have treated and for the BLTP motion with radiation-friction force the dynamics properly as a physical initial value problem given by (88). The applied electric field strength is 0.1 with the same data as in the test particle formulation, in units of eκ2, which is a strong field for this problem. unlike the treatment in [39] where “the whole path tra- The radiation-friction effect is clearly visible. versed by the particle up to the present time contributes We note that for very weak applied field to [the self-force]” (quoted from [39]). strength the radiation-reaction is captured by the “Newtonian-friction” approximation (89), and the point 6c hom charge’s velocity will saturate at v∞ = eκ2 ; see VIII. SUMMARY AND OUTLOOK Fig. 2 for an applied field strength of 0.−01 inE units of eκ2. In this paper we have shown that a well-defined elec- The sharp borderline between the “weak field” and tromagnetic force on a point charge source of the classical “strong field” regimes determines a critical field strength electromagnetic field can be extracted from momentum for this problem. For field strengths just slightly above balance among charges and field, whenever the electro- the critical value 0.0519 the velocity temporarily magnetic vacuum law which supplies the closure relation reaches a quasi-plateau,≈ before it makes the final for the pre-metric Maxwell field equations leads to a fi- transition to approach the speed of light; see Fig. 3 for nite field momentum vector which is differentiable with an applied field strength of 0.052 in units of eκ2. respect to time. Although the κ3 term (88) is not an accurate for- For the BLTP law of the vacuum we even reported that mula for the radiation-reaction∝ force in the physically in- we were able to prove, in collaboration with S. Tahvildar- teresting regime of very large κ values, it is entirely ade- Zadeh, that the BLTP electromagnetic force as defined 17

namical initial value problem by revisiting the standard textbook problem of a point charge released from rest in a constant applied electrostatic field. Our discussion con- firms that the test-particle approximation is valid in the initial dynamical phase, with radiation-reaction correc- tions first in form of a linear , “Newtonian friction”-type term (at least in the small κ regime), and eventually in a non-linear manner. A most interesting finding, valid for arbitrary κ, is that in the initial phase of the dynamics the particle in- ertia is determined entirely by its bare rest mass, not by the mass of the (electromagnetically) “dressed” particle. The latter is generally thought to control the inertia in scattering scenarios. The predominance of scattering ex- periments, in particular in high energy physics, has led to the general belief that only the “dressed particle” mass FIG. 2. The velocity of a point charge, starting from rest hom is observable in experiments, not the bare mass. Our in a very weak, constant applied electrostatic field E , as findings by contrast suggest that the bare mass may be per test particle theory (dashed curve), and as per BLTP 2 observable by cleverly setting up an initial value problem electrodynamics with O(κ ) friction only (continuous curve). (Ignore the continuous curve.) in the laboratory. True, BLTP electrodynamics may not be the most re- alistic classical theory, but it surely is a “proof of con- cept,” signaling that analogous results should be feasible also for putatively more realistic models, in particular the BI electrodynamics. We remark that a well-defined joint initial value problem for the MBI fields and their point charge sources was formulated with the help of a Hamilton–Jacobi-type theory in [46], but it is not clear whether that theory is well-posed, nor is it clear that its dynamics is independent of the invoked foliation of spacetime it needs for its formulation. Since the setup given in the present paper is truly Lorentz co-variant and foliation-independent, it should shed light on the formu- lation given in [46] by comparing the two. Incidentally, neither Born & Infeld, nor Schr¨odinger, nor Dirac, pro- posed the electromagnetic force given in this paper but instead tried to implement the ill-defined Lorentz force formula into the Born–Infeld electrodynamics; cf. [47]. FIG. 3. The velocity of a point charge, starting from hom Inside the family of well-posed classical models one rest in a constant applied electrostatic field E of slightly may hope to find the classical limit of the elusive, math- larger-than-critical field strength, as per test particle the- ematically well-defined and physically viable, special- ory (dashed curve), and as per BLTP electrodynamics with 2 O(κ ) friction only (continuous curve). (Ignore the continu- relativistic quantum theory of electromagnetism. ous curve.) Having obtained a rigorous control over the classical electromagnetic radiation-reaction problem, an impor- tant next goal in the realm of classical physics is to get a rigorous hand on the gravitational radiation-reaction in (16) furnishes a well-posed joint initial value problem problem. As a first step, armed with the insights gained for fields and point particles which is of second-order in from the special-relativistic theory of motion formulated the particle positions; see [36] for the details. To the in this paper we have embarked on an assessment of the best of our knowledge BLTP electrodynamics is the first Einstein–Infeld–Hoffmann [45] legacy; cf. [37]. classical electrodynamical theory of point charges and their electromagnetic fields which has been shown to be dynamically well-posed, free of infinite “self” energies etc. ... and ill-defined Lorentz “self” forces, and free of the q - ACKNOWLEDGMENT problem. Incidentally, neither Bopp, Land´e–Thomas, nor Podolsky considered the definition of the force given in The ideas and insights reported here evolved over this paper, but tried (in vain) to implement the ill-defined many years, during which I have benefited especially Lorentz force formula into their theory. from discussions with (in alphabetical order): Walter We have illustrated the well-posed BLTP electrody- Appel, Holly Carley, Dirk Deckert, Detlef D¨urr, Vu 18

Hoang, Markus Kunze, Volker Perlick, Maria Radosz, prompting me to explain the MBLTP initial value prob- Jared Speck, Herbert Spohn, and Shadi Tahvildar-Zadeh lem more clearly; to the referee for the helpful comments in particular, for which I am very grateful. After the which improved the presentation, and for prompting me first version of this paper was made public, the author to work out the example discussed in section VII. received several comments which were implemented in the revision. Many thanks go: to Bob Wald for draw- ing attention to Ref. [5]; to Mario Hubert for catching a REFERENCES sign error in the Born–Infeld section; to Herb Johnson for

[1] Lorentz, H.A., La th´eorie ´electromagnetique de Maxwell [20] Born, M., and Infeld, L., Foundation of the new field et son application aux corps mouvants, Arch. N´eerl. Sci. theory, Proc. Roy. Soc. London A 144, 425–451 (1934). Exactes Nat. 25, pp. 363–552 (1892). [21] Bia lynicki-Birula, I., Nonlinear electrodynamics: Varia- [2] Li´enard, A., Champ ´electrique et magn´etique produit par tions on a theme by Born and Infeld, pp. 31-48 in “Quan- une charge concentr´ee en un point et anim´ee d’un mou- tum theory of particles and fields,” eds. B. Jancewicz and vement quelconque, L’ Eclairage´ ´electrique 16 p.5; ibid. J. Lukierski, World Scientific, Singapore (1983). p. 53; ibid. p. 106 (1898). [22] Bopp, F., Eine lineare Theorie des Elektrons, Annalen [3] Wiechert, E., Elektrodynamische Elementargesetze, Arch. Phys. 430, 345–384 (1940). N´eerl. Sci. Exactes Nat. 5, 549-573 (1900). [23] Bopp, F., Lineare Theorie des Elektrons. II, Annalen [4] Jackson, J.D., Classical electrodynamics, J. Wiley & Phys. 434, 573–608 (1943). nd rd Sons, New York 2 ed. (1975); 3 ed. (1999). [24] Land´e, A., Finite Self-Energies in Radiation Theory. I, [5] Gralla, S.E., Harte, A., and Wald, R.M., A rigorous Phys. Rev. 60, 121–126 (1941). derivation of electromagnetic self-force, Phys. Rev. D [25] Land´e, A., and Thomas, L.H., Finite Self-Energies in 80:024031 (2009). Radiation Theory. II, Phys. Rev. 60, 514–523 (1941). [6] Lorentz, H.A., Weiterbildung der Maxwell’schen The- [26] Podolsky, B., A generalized electrodynamics. Part I: Non- orie: Elektronentheorie., Encyklop¨adie d. Mathematis- quantum, Phys. Rev. 62, 68–71 (1942). chen Wissenschaften V2, Art. 14, pp. 145–280 (1904). [27] Podolsky, B., and Schwed, P., A review of generalized [7] Yaghjian, A.D., “Relativistic dynamics of a charged electrodynamics, Rev. Mod. Phys. 20, 40–50 (1948). sphere,” Lect. Notes Phys. m11, Springer, Berlin (1992). [28] Mie, G., Grundlagen einer Theorie der Materie, I, An- [8] Spohn, H., Dynamics of charged particles and their radi- nalen Phys. 37, 511-534 (1912). ation fields, Cambridge UP (2004). [29] Mie, G., Grundlagen einer Theorie der Materie, II, An- [9] Appel, W., and Kiessling, M.K.-H., Mass and spin renor- nalen Phys. 39, 1-40 (1912). malization in Lorentz electrodynamics, Annals Phys. 289, [30] Schwarzschild, K. Zur Elektrodynamik.I, Nachr. Ges. 24–83 (2001). Wiss. G¨ottingen, 126–131 (1903). [10] Dirac, P. A. M., Classical theory of radiating electrons, [31] Speck, J.R., The nonlinear stability of the trivial solution Proc. Roy. Soc. A 167, 148–169 (1938). to the Maxwell-Born-Infeld system, J. Math. Phys. 53, [11] Fermi, E., Uber¨ einen Widerspruch zwischen der elektro- 083703(83pp) (2012). dynamischen und der relativistischen Theorie der elektro- [32] Serre, D., Les ondes planes en ´ectromagn´et´etisme non- magnetischen Masse, Phys. Zeitschr. 23, 340-344 (1922). lin´eaire, Physica D 31, 227-251 (1988). [12] Landau, L. D., and Lifshitz, E., The Classical Theory of [33] Brenier, Y., Hydrodynamic structure of the augmented Fields, Pergamon (1951). Born–Infeld equations, Arch. Rat. Mech. Anal. 172, 65- [13] Poisson, E., Pound, A., and Vega, I., The motion of point 91 (2004). particles in curved spacetime, Living Rev. Rel. 14,7(190) [34] Kiessling, M.K.-H., On the quasi-linear elliptic PDE 2 s (2011). −∇ · (∇u/p1 − |∇u| ) = 4π Pk akδ k in physics and [14] Burton, D.-A., and Noble, A., Aspects of electromagnetic geometry, Commun. Math. Phys. 314, 509–523 (2012); radiation in strong fields, Contemp. Phys. 55, 110–121 Correction ibid. 364, 825–833 (2018). (2014). [35] Carley, H. K., Kiessling, M. K.-H., and Perlick, V., On [15] Ares de Parga, G., Mares, R., and Dominguez, S. An un- the Schr¨odinger spectrum of a hydrogen atom with BLTP physical result for the Landau–Lifshitz equation of motion interactions between electron and proton, Int. J. Mod. for a charged particle, Rev. Mex. Fis. 52, 139–142 (2006). Phys. A 34, 1950146 (2019). [16] Deckert, D.-A., and Hartenstein, V., On the initial [36] Kiessling, M.K.-H., and Tahvildar-Zadeh, A. S., Bopp- value formulation of classical electrodynamics J. Phys. Land´e-Thomas-Podolsky electrodynamics as initial value A: Math. Theor. 49, 445202 (19pp.) (2016). problem, in preparation (2020). [17] Miller, A. I., Albert Einstein’s special theory of relativity, [37] Kiessling, M.K.-H., and Tahvildar-Zadeh, A. S., The Springer, New York (1998). Einstein–Infeld–Hoffmann legacy in mathematical rela- [18] Abraham, M., Die Grundhypothesen der Elektronenthe- tivity. I: The classical motion of charged point particles, orie, Phys. Z. 5, 576-579 (1904). 16pp. in Proc. of the 15th Marcel Grossmann Meeting [19] Hehl, F. W., and Obukov, Y. N., “Foundations of Clas- on General Relativity, Rome, Italy (2018); Int. J. Mod. sical Electrodynamics,” Birkh¨auser, Basel (2003). Phys. D 28 1930017 (2019). 19

[38] Gratus, J., Perlick, V., and Tucker, R.W., On the self- [43] Detweiler, S., and Whiting, B. F., Self-force via a Greens force in Bopp–Podolsky electrodynamics, J. Phys. A: function decomposition, Phys. Rev. D 67, 024025 (2003). Math. Theor. 48, 435401 (28pp.) (2015). [44] Feynman, R. P., A relativistic cut-off for classical elec- [39] Zayats, A.E., Self-interaction in the Bopp-Podolsky elec- trodynamics, Phys. Rev. 74, 939–946 (1948). trodynamics: Can the observable mass of a charged par- [45] Einstein, A., Infeld, L., and Hoffmann, B., The gravita- ticle depend on its acceleration?, Annals Phys. (NY)342, tional equations and the problem of motion, Annals Math. 11–20 (2014). 39, 65-100 (1938). [40] Hoang, V., and Radosz, M., On the self-force [46] Kiessling, M.K.-H., Electromagnetic field theory without in higher-order electrodynamics, Preprint (41pp.) divergence problems. 1. The Born legacy, J. Stat. Phys. arXiv:1902.06386 (2019). 116, 1057-1122 (2004). [41] Quinn, T.C., and Wald, R.M., An axiomatic approach [47] Kiessling, M.K.-H., On the Motion of Point Defects in to electromagnetic and gravitational radiation reaction of Relativistic Fields, pp.299–335 in: Finster F., M¨uller O., particles in curved spacetime, Phys. Rev. D 56, 3381– Nardmann M., Tolksdorf J., Zeidler E. (eds) “Quantum 3394 (1997). Field Theory and Gravity.” Springer, Basel (2012). [42] Kijowski, J., Electrodynamics of moving particles, Gen. Rel. Grav. 26, 167-192 (1994).