sign in sign up
<<
Home , Curved space, Einstein field equations, Equivalence principle, Field equation, General relativity, Gravitation (book)

How Einstein Got His Equations

S. Walters

In commemoration of ’s centennial

ABSTRACT. We study the pages in ’s 1916 landmark paper in the where he derived his field equations for . Einstein made two heuristic and physically insightful steps. The first was to obtain the field equations in in a rather geometric fashion. The step was obtaining the field equations in the presence of from the field equations in vacuum. (This transition is an essential principle in , much as the principle of local gauge invariance in quantum field .) To this end, we go over some quick differential geometric background related to , vectors, , , , tensor, Ricci tensor, and see how Einstein used to model gravity.

This paper is a more detailed version of my talk given at the Math-Physics Symposium at UNBC on February 25, 2016. It is in reference to Einstein’s paper: A. Einstein, The Foundation of the General , Annalen der Physik, 49, 1916. (For an English see: H. A. Lorentz, A. Einstein, H. Minkowski, H. Weyl, The .) The paper has two sections. The first section is a smash course on the semi- tools needed to understand Einstein’s theory. The second section looks at Einstein’s derivation of his field equations in vac- uum and in the presence of matter and/or as he worked them out in his paper. This paper commemorates the 100th centennial of Einstein’s General The- ory of Relativity, which he finalized near the end of November 1915 and pub- lished in 1916. arXiv:1608.05752v1 [physics.hist-ph] 19 Aug 2016

Date: Jan./Feb. 2016 (last update July 2016) LATEX File: EinsteinRelativityLectureFeb2016.tex. 2000 Subject Classification. 35Q76 83C05 35Q75 37N20 83-01 83-XX. Key words and phrases. Einstein field equations, tensors, relativity, gravity, curvature, . 1 2 S. Walters 1. Semi-Riemannian to the rule ∂x0ν V 0ν = V µ . (1.1) Geometry ∂xµ In this section we outline briefly some of Covariant vector fields are written using the basic and standard concepts known from subscripts Vµ and they transform in a dual manner via semi-Riemannian geometry that are used in ∂xµ Einstein’s formulation of his theory of grav- V 0 = V . (1.2) ν x0ν µ ity. ∂ Spacetime Curvilinear coordinates. Spacetime coordinates are written using superscripts simply as xµ where µ = 1,2,3,4. This is short for vectors (x1,x2,x3,x4) describ- ing points in spacetime with respect to a cer- tain (cartesian, cylindri- cal, spherical, etc). The coordinates x1,x2,x3 refer to spatial coordinates and x4 to the coordinate. Any other coordinate system can be written as x0µ . (An observer and the coor- dinate system used by him/her will be iden- tified.) Einstein’s Summation Convention. Figure 1. At each p in time Any time an upper index is repeated as a one has natural vectors e1,e2 that are tan- lower index we are automatically summing gent to the curved coordinate axes, quite ˆi, jˆ, kˆ over that index. For example, we write analogous to the vectors in . 4 4 Examples. V µU ≡ V µU , Aµν ≡ Aµν . ∑ µ µ ∑ τµ τµ (1) If a is moving in space and µ=1 µ=1 its spacetime coordinates are expressed in (In such cases one drops the ∑ notation.) terms of its τ, i.e. xµ = xµ (τ), A vector field V is a differentiable func- then its 4-vector is defined by tion defined on a certain region of space time dxµ whose values are “tangent” vectors to space- V µ = . time (much as vectors one the that dτ are tangent to it at each point of a certain Defining a particle’s velocity in this way is region on it). There is a natural way to de- quite convenient since it gives us a “covari- scribe V by means of its components V µ rela- ant” way of describing . This is so tive to a given coordinate system xµ . (Essen- simply because from the chain rule from cal- tially, each coordinate axis xµ has a vector culus we see that the velocity 4-vector trans- along it similar to the usual ˆi, jˆ, kˆ vectors in forms just like a contravariant vector field: 3-space 3, as illustrated in Figure 1.) dx0ν ∂x0ν dxµ R = . µ The components of such vector fields are dτ ∂x dτ written with superscripts and called contra (2) As a second example, consider a scalar µ 0ν variant because their components in differ- field ϕ = ϕ(x ) = ϕ(x ), and look at its gradi- µ µ ∂ϕ ent coordinate systems – V relative to x ent vector field ∇ϕ with components Vν = ∂xν . and V 0µ relative to x0µ – transform according Given any other coordinate system x0µ , the How Einstein Got His Field Equations 3 chain rule tells us hows the com- by 4 symmetric with real entries) be- ponents are related: ing “− + ++” means that it has one one neg- ∂ϕ ∂xµ ∂ϕ ative eigenvalue and 3 positive eigenvalues. = . (The only negative eigenvalue is related to ∂x0ν ∂x0ν ∂xµ the time coordinate.) This is exactly how covariant vectors trans- In Relativity we are interested in semi- Rie- form – as in equation (1.2). mannian (or Lorentzian) metric tensors ex- A tensor (field) is like a vector (field) except emplified by the Minkowski metric that it can have two or more index compo- 1 0 0 0  nents, e.g. 0 1 0 0 η =   (1.3) µν µ ρτ µν 0 0 1 0  Aµν B C νρ R κα .   0 0 0 −1 The number of indices is called the rank of the tensor. (So vectors have rank 1, scalar (used in ) and its associ- µ fields have rank 0, and C νρ has rank 3.) ated element of proper spacetime interval The way tensor components transform be- ds2 = η dxµ dxν = −dτ2 tween different coordinate systems is just µν like equation (1.1) for each upper index, and where dτ2 is called the proper time (by anal- like equation (1.2) for each lower (covariant) ogy with element of arc in Riemann- β index. Thus, for the rank 2 tensor Aα , its ian geometry). components in another coordinate system In General Relativity, the spacetime inter- 0ν Aµ are related by val extends to any coordinate system using the components g according 0ν α µν 0ν ∂x ∂x β to Aµ = Aα . ∂xβ ∂x0µ 2 µ ν 2 ds = gµν dx dx = −dτ . It is for such elegant transformation prop- The essential property of proper time is erties of tensors that Einstein’s principle of that it is an for all coordinate sys- (discussed in the next sec- tems: tion) requires that the laws of physics be 2 02 expressed in covariant form using tensors. dτ = dτ . Once such a law holds in one coordinate All observers (coordinate systems) will agree system the same form of the law holds in on this quantity. (Again, this follows by the any other coordinate system. chain rule.)

The most important rank 2 tensor is the Since the metric tensor gµν is an invertible metric tensor gµν because it contains the matrix, its inverse is written in superscript essential ingredients of gravity, , form gµν . Thus using the usual rules of ma- time, and curvature of spacetime. It has the trix multiplication one has essential property that it is symmetric µν µ g gνλ = δλ gµν = gνµ . where δ µ is the function (which One can have either a Riemannian or a semi- λ is 1 when µ = λ and is 0 when µ 6= λ). Riemannian metric tensor depending on its signature being + + ++ or − + ++, respec- Both forms of the metric tensor are used tively. The signature of gµν (treated as a 4 to raise and lower indices on tensors. For 4 S. Walters example, Einstein regarded the Christoffel symbols λ ρτ Γ the components of the gravitational A = g Aµ , g R = R τ . µν as “ ν νµ ρµ κα µ κα field.” In his paper, he also referred to the The contraction of a is ob- metric tensor gµν as “describing the grav- tained by setting an upper index and a lower itational field in relation to the chosen index equal and then summing. For instance, system of coordinates.” So, in a manner ρτ if one contracts the rank 4 tensor R κα with of speaking, gravity affects how we take the respect to the indices ρ and κ, one gets the derivative!2 rank 2 tensor A test particle moving “freely” in vacuum 4 under the influence of the geometry of space Sτ = Rρτ ≡ Rρτ α ρα ∑ ρα (i.e., under gravity) obeys the equa- ρ=1 tion 1 d2xρ dxµ dxν where the index ρ now disappeared. + Γρ = 0 dτ2 µν dτ dτ . Since spacetime is (where τ is the particle’s proper time). This generally curved, the notion of the deriva- equation describes the “straight lines” tive from calculus becomes curved also. It () of spacetime. Thus a ray of becomes what we call the covariant deriva- or moving freely in (vacuum) space- tive. The covariant derivative of a covariant time will follow a path that is the analogue of vector field A is given by µ straight lines (or great on a sphere). ∂Aµ λ For a slowly moving particle in low grav- Aµ;ν = − Γµν Aλ ∂xν ity, this equation essentially comes down to where Newton’s second law F = ma. If Γ ≡ 0 (zero d2xρ 1 ∂g ∂g ∂g  gravity) the geodesic equation becomes 2 = Γλ = gλρ ρν + ρµ − µν dτ µν 2 ∂xµ ∂xν ∂xρ 0, i.e. zero or constant velocity, so it moves along a straight (i.e., we’re is called the Christoffel symbol. The covari- in flat spacetime). ant derivative of a contravariant vector field is quite similar except with a “+” in place of The Riemann-Christoffel Curvature Ten- the “−”; thus sor is the rank 4 tensor µ λ ∂ λ ∂ λ η λ η λ µ ∂A µ R = Γ − Γ + Γ Γ − Γ Γ . A = + Γ Aλ . µνκ ∂xκ µν ∂xν µκ µν κη µκ νη ;ν ∂xν νλ It governs all aspect of the curvature of space- For example, if the metric tensor gµν is time. If it is zero everywhere then our space- constant in some coordinate system, then time is flat. This looks like a fairly com- Γ’s are all 0 and so the covariant derivative plicated expression, but here’s a way to re- reduces to the usual partial derivative. The late it to something that is more familiar. advantage of the covariant derivative is that From advanced calculus we know that par- it gives rise to tensors: the covariant deriva- tial derivatives commute on most functions tive of tensors are tensors also. 2The same thing happens in QFT where the de- 1One can contract between any upper index and rivative is similarly modified by means of the 4-vector any lower index. We can also “contract” between two potential of the electromagnetic field in order that the lower or two upper indices after one raises or lowers Lagrangian remain invariant – e.g., see [3], Section one of them. It’s just like taking the trace of a matrix. 10.3. How Einstein Got His Field Equations 5

∂ ∂ ∂ ∂ 5 that appear in practice: i.e., ∂x ∂y f − ∂y ∂x f = 0 1915, he conjectured that the two should (usually called Clairaut’s Theorem). If one be proportional: Rµν = κTµν . Though this works out a similar difference for the covari- equation is not quite correct, since Ricci does ant derivative in , one does not not generally satisfy the , get 0 in general, but instead gets an expres- it’s a significant step closer to his ultimate λ field equation (which he eventually reached sion that involves the components Rµνκ of the . So the cur- by the end of November). vature tensor can be seen as the degree to We now turn to how Einstein, in his pa- which covariant derivatives depend on the per, finally arrived at his field equation after order that they are performed.3 a long and elusive search which ended No- The Riemann tensor has an important con- vember 25, 1915. traction called the Ricci4 tensor: λ 2. Einstein’s Derivation of Rµκ = Rµλκ (of rank 2). Lastly, contracting the Ricci ten- the Field Equations sor once more we obtain the scalar curva- Principle of Equivalence. This princi- ture tensor ple arose from what Einstein described as µ R = Rµ the “happiest thought of my life.” Einstein 6 which is a scalar field on spacetime (related wrote : to the ). “The gravitational field has only There is a rank 2 tensor that has special a relative existence in a way sim- physical significance. It is called the - ilar to the electric field gener- - tensor denoted by Tµν . ated by magentoelectric induc- It contains or models all the essential physics tion. Because for an observer of a system pertinent to and space falling freely from the roof of a time - specifically, it contains information house there exists – at least in regarding energy , energy flow, and his immediate surroundings – no momenta of matter particles and/or radia- gravitational field. tion. It is an essential component of Ein- Einstein’s comes down stein’s field equations. It is normally as- to saying that spacetime is locally Minkowski sumed that one has local energy-momentum (locally inertial). That is, at each point p conservation, which translates into saying of spacetime there is a coordinate system that the energy momentum tensor satisfies where the metric tensor gµν is exactly the the covariant form of the continuity equa- Minkowski metric ηµν at p (see equation (1.3)) tion and where it’s derivatives vanish at p (i.e., µν 7 T ;µ = 0. gravity vanishes). In essence, then, in curved (That is, its covariant vanishes.) spacetime each point has a local coordinate Einstein knew that the Ricci tensor Rµν has strong to energy-momentum 5Pais [4], Chapter 14 (page 253). tensor. And at one point, early in November 6Pais [4], p. 178. 7Here we are referring to the usual partial deriva- 3 τ For a detailed calculation see Weinberg [5], Sec- tives ∂gµν /∂x – not the covariant derivative, since it tion 6.5. is a theorem that the covariant derivatives of the met- 4Pronounced “Reeshi.” ric tensor vanish everywhere. 6 S. Walters system where the Special Theory of Relativ- by suitably scaling the coordinate system so p ity holds good. that −det(gµν ) = 1; he then gets the result Einstein’s Principle of General Covari- ∂ λ η λ ance states that all the laws of physics (at Rµν = Γµν + Γ Γνη = 0. (2.1) ∂xλ µλ least those pertaining to gravitational effects) are to be expressed in covariant form using Using the Lagrangian/ formalism, Ein- tensors. That is, the laws of Physics should stein rewrote this equation in a more phys- be expressible in terms of tensors since ten- ically meaningful form that has an impor- sors transform in a covariant manner – i.e., tant term in it that represents the energy- σ the laws should have the same form regard- momentum “tensor” tµ of the gravitational less of which coordinate system (observer) is field11; so Einstein recomputes equation (2.1) 8 used. to the following Einstein’s is also   a method for finding how physical equations νσ ∂  σβ λ  σ 1 σ g Rµν = g Γ + κ t − δ t = 0 look in any coordinate system: just find out ∂xλ µβ µ 2 µ what the physical equation looks like in in- (2.2) µ σ ertial coordinate systems (which exist locally for some constant κ and t = tµ , where tµ is by the Equivalence Principle), and then make given by the replacements σ 1 σ αβ η λ αβ σ λ ∂A κtµ = δµ g Γαλ Γβη − g Γαλ Γβ µ . η −→ g , −→ A 2 µν µν ∂xµ ;µ Einstein’s next insight: to get the grav- so usual derivatives get replaced by covari- itational field equations in the presence of ant derivatives. The resulting equation will matter and , in the previous equa- be the covariant form of the physical law in 9 tion he replaced the energy momentum ten- any new (accelerated) coordinate system. σ sor tµ of the gravitational field by the total σ σ Einstein’s Postulate: in vacuum10, the energy momentum tensor tµ + Tµ of gravity contracted curvature tensor, the Ricci ten- and of matter12 together! Once this is done, sor, vanishes: he then retraced his steps back the way he R = 0. arrived to (2.2) to obtain his field equations, µν σ which now incorporate matter Tµ . Since this vacuum equation has 4 terms (see σ σ σ Replacing tµ by tµ +Tµ in the vacuum equa- definition above of the Riemann tensor given tion (2.2), Albert Einstein obtains the matter by 4 terms), Einstein simplified it to 2 terms field equation:

8 This means, for example, that if Aµν = Bµν is a co- ∂     gσβ Γα + κ (tσ + T σ ) − 1 δ σ (t + T) = 0 variant (tensor) form for a law of physics in one coor- ∂xα µβ µ µ 2 µ dinate system, then in another coordinate system it 0 0 would read quite similarly: Aµν = Bµν . For instance, Maxwell’s equations can be so expressed in covariant 11It isn’t a tensor in general relativity though it is form - e.g., see [5], Chapter 5, Section 2. a tensor in Special Relativity since it is acts covari- 9See [5], Section 7.6. antly with respect to Lorentz transformations – see 10In this context, “vacuum” means no matter, no [5], Section 7.6, page 167. electromagnetic fields, no that are nongravita- 12Non-gravitational matter and radiation now be- σ tional, etc. – only spacetime (i.e., gravity). ing represented by the tensor Tµ . How Einstein Got His Field Equations 7

µ µ where t + T = tµ + Tµ is the contraction; or - Einstein calculates the precise preces- sion of Mercurys = 43 second of arc ∂       gσβ α + tσ − 1 σ t = − T σ − 1 σ T . per century. α Γµβ κ µ 2 δµ κ µ 2 δµ ∂x - He calculates the bending of light as well σ Now he eliminates tµ from view (since it is as its in a gravitational field. (This not really a tensor in General Relativity) and was verified experimentally in 1919, three he calculates the left side, which in view of years after his paper was published.) νσ (2.2) is g Rµν , and obtains it in terms of the - The slowing down of in a gravita- Ricci tensor thus: tional field. (This was verified by the Eotv¨ os¨ νσ  σ 1 σ  experiment which Einstein was already aware g Rµν = −κ T − δ T . (2.3) µ 2 µ of.) Notice that if there is no matter , so There are numerous tests and applications σ that the tensor Tµ is zero, then this equation of his General Theory which Einstein him- reduces to the vacuum field equation (2.2) self did not anticipate. He did however pre- νσ νσ g Rµν = 0. Taking the g in (2.3) on the dict the existence of gravitational [1] other side he gets which have just recently (in 2015-16) been detected thru the intense gravitational bat-  σ 1 σ  1  Rµν = −κgνσ Tµ − 2 δµ T = −κ Tµν − 2 gµν T tles between two holes. (A good short or historical account of Einstein’s work on grav- 1 itational waves is [6].) Rµν = −κTµν + 2 κgµν T. (2.4) Now contract this equation with respect to REFERENCES the indices µ and ν to get the scalar fields [1] A. Einstein, Approximative Integration of the 1 Field Equations of Gravitation, Prussian Acad- R = −κT + 2 κ4T = κT emy of , June 1916. Sitzungsber and express the T in (2.4) in terms of R so Preuss. Akad. Wiss. Berlin (Math. Phys.) (1916), that the general field equation can now be 688-696. https://inspirehep.net/record/42612 [2] H. A. Lorentz, A. Einstein, H. Minkowski, H. written in the form that is most familiar in Weyl, The Principle of Relativity. (Includes Eng- the literature: lish translation of Einstein’s general relativity paper which appeared in Annalen der Physik 1 Rµν − gµνR = −κTµν. 49,1916.) 2 [3] D. Griffiths, Introduction to , This is Einstein’s gravitational field equa- Wiley-VCH Verlag, 2008, 2nd edition. tion in the presence of matter. (Matter and [4] , , The electromagnetism are represented by the ten- and The Life of Albert Einstein, Oxford University Press, 1982. sor Tµν .) He later calculates the constant [5] , Gravitation and , κ = 8πG (in units where the of light John Wiley and Sons, 1972. c is set to 1). [6] G. Weinstein, Einstein’s Discovery of Gravitational Waves 1916-1918, 2016, Applications/. Einstein ends http://arxiv.org/abs/1602.04040 his paper with these applications: - Einstein verifies that energy-momentum DEPARTMENTOF MATHEMATICSAND STATISTICS,UNI- σ σ VERSITYOF NORTHERN B.C.,PRINCE GEORGE,B.C. tensor for gravity and matter tµ + Tµ has en- ergy conservation. V2N 4Z9, CANADA. E-mail address: [email protected] - He shows that to first order approxima- URL: http://hilbert.unbc.ca/ tion he gets Newtons laws.