Back

Stress-energy projected components:
Let ρ=nunvTuv = the stress-energy density (2008 Alcubierre, eqn. 2.4.12)
Let ja=nuTuvγav = the stress-energy spatial current (2008 Alcubierre, eqn. 2.4.12)
Let Sab=Tuvγauγbv = the spatial stress tensor
Let S=Tuvγuv=Saa = the spatial stress tensor trace
So Tuvguv=Tuv(γuvnunv)=Sρ

ja=γabjb
=γabnuTuvγvb
=nuTuvγva
=ja
so ja is spatial

jana=nuTuvγvana=0 so ja is spatial

So Tuv=δuaδvbTab
=(γuanuna)(γvbnvnb)Tab
=(γuaγvbγuanvnbnunaγvb+nunanvnb)Tab
=γuaγvbTabγuanvnbTabnunaγvbTab+nunanvnbTab
=Suv+nvju+nujv+nunvρ

So Tuv12guvTabgab
=Suv+nvju+nujv+nunvρ12(γuvnunv)(Sab+nbja+najb+nanbρ)(γabnanb)
=Suv+nvju+nujv+nunvρ12(γuvnunv)(SabγabSabnanb+nbjaγabnbjananb+najbγabnajbnanb+nanbργabnanbρnanb)
=Suv12γuv(Sρ)+nvju+nujv+nunvρ+12nunv(Sρ)
=Suv+nvju+nujv+nunvρ12(γuvnunv)(Sρ)
=Suv+nvju+nujv+nunvρ12guv(Sρ)



Einstein Field Equations in projected components:

Einstein Field Equations, contracting both indexes with the normal vector:
nunvGuv=8πnunvTuv
nunvGuv=8πρ
using 2nunvGuv=R+K2KabKab:
R+K2KabKab=16πρ (2010 Baumgarte & Shapiro, eqn. 2.125)
R+K2KabKab16πρ=0
Hamiltonian constraint: Let...
H=nunvGuv8πρ (2008 Alcubierre, eqn. 2.5.10)
H=12(R+K2KabKab)8πρ (2008 Alcubierre, eqn. 2.5.9)
Now when the Einstein Field Equations are satisfied, we will have H=0

Einstein Field Equations, contracting one index with normal and projecting the other:
nuγavGuv=nuγav8πTuv
using ja=nuγavTuv
nuγavGuv=8πja
using γaunvGuv=aKbKab
bKabaK=8πja
bKabaK8πja=0
Momentum constraint: Let...
Ma=nuγavGuv8πja (2008 Alcubierre, eqn. 2.5.11)
Ma=bKabaK8πja
When the Einstein Field Equations are satisfied, we will have Ma=0

Einstein Field Equations, projecting both indexes.
γauγbvGuv=8πγauγbvTuv
using Sab=Tuvγauγbv:
γauγbvGuv=8πSuv
γauγbvGuv8πSuv=0
Energy constraint: Let...
Eab=γauγbvGuv8πSuv (2008 Alcubierre, eqn. 2.5.12)
... this one is mentioned in 2008 Alcubierre, eqn. 2.5.12, but no mention of which values for Guv are used ... is it the R definition? is it the K definition?
When the Einstein Field Equations are satisfied, we will have Eab=0

starting with (from the "normal projections" worksheet):
Rab+KKabKacKcb=γaqγbsRqs+npnrγaqγbsRpqrs
npnrγaqγbsRpqrs=Rab+KKabKacKcbγaqγbsRqs
Substitute the trace-reversed Einstein field equations, Rab=8π(Tab12gabT)
npnrγaqγbsRpqrs=Rab+KKabKacKcbγauγbv8π(Tuv12guvTcdgcd) (2010 Baumgarte & Shapiro, 2.103)
npnrγaqγbsRpqrs=Rab+KKabKacKcb8π(Sab12γab(Sρ))
... and this route is the one that turns into the time-derivative of the extrinsic curvature ...
... but how do we come up with the Eab constraints from only projected, non-time-derivative terms?
We don't. The Eab constraints give rise to the Kab ivp. So using ADM formalism this constraint is always satisfied.

LnKab=nuuKab+Kubanu+Kaubnu
LnKab=nuuKab+Kubanu+Kaubnu
using Kab=anbnaab so anb=Kabnaab
LnKab=nuu(anbnaab)+Kub(Kaunaau)+Kua(Kbunbau)
LnKab=nuuanbabnuunananuuabKubKauKuaKbuKubnaauKuanbau (2010 Baumgarte & Shapiro, eqn. 2.74)
using aa=nuuna
LnKab=nuuanbaaabnanuuab2KauKubKubnaauKuanbau
using ndRdbua=uanbaunb
LnKab=ndnuRdbuanuaunbaaabnanuuab2KauKubKubnaauKuanbau
TODO after a lot of work
LnKab=1αabαKacKcb+npnrγaqγbsRpqrs (2010 Baumgarte & Shapiro, eqn. 2.82)
using the definition of npnrγaqγbsRpqrs above ... without stress-energy:
LnKab=1αabαKacKcb+Rab+KKabKacKcbγauγbvRuv
... and with stress-energy:
LnKab=1αabαKacKcb+Rab+KKabKacKcb8π(Sab12γab(Sρ))
LnKab=1αabα+Rab+KKab2KacKcb8π(Sab12γab(Sρ))

Extrinsic curvature initial value formulation:
using Ln=1α(LtLβ)
tKabLβKab=αLnKab
tKab=abα+α(Rab+KKab2KacKcb8π(Sab12γab(Sρ)))+LβKab



Ricci tensor contracted with normal:
nanbRab
=nanbRcacb
=nb(cbncbcnc)
=nbcbncnbbcnc
=c(nbbnc)(cnb)(bnc)b(nbcnc)+(bnb)(cnc)
=c(nbbnc)(Kcb+ncab)(Kbc+nbac)b(nbcnc)+(Kbb+nbab)(Kcc+ncac)
=c(nbbnc)(KcbKbcncabKbcnbacKcb+nbncabac)b(nbcnc)+K2
=c(nbbnc)b(nbcnc)KabKab+K2
=c(nbbnc)c(ncbnb)KabKab+K2
nanbRab=K2KabKab+c(nbbncncbnb) (2008 Alcubierre, eqn 2.7.3)

start with from the "normal projections" worksheet:
R=RKabKab+K22nanbRab
...substitute contracted Ricci tensor...
R=RKabKab+K22(K2KabKab+c(nbbncncbnb))
R=RKabKab+K22K2+2KabKab2c(nbbncncbnb)
R=R+KabKabK22c(nbbncncbnb) (2008 Alcubierre, eqn 2.7.4)
Looks like now we disregard the 2c(nbbncncbnb)
because it can be converted to a total integral via divergence theorem.
but this is only the case if we are integrating over a volume, i.e. true for Ldx4 and therefore Rdx4
So how can it still be true for L=gRdx4? Does cgdx4 disappear as well?
R=R+KabKabK2

Ok a suspicious thing about the "cancel total derivatives", that would imply:
nanbRab=K2KabKab
Is only true in the context of a spacetime-volume integral?
If so then how true is all of numerical relativity, where we are doing numerical spatial volume integration / spatial PDEs?

Back from our "normal projection" worksheet, start with our projected Einstein curvature:
γauγbvGuv=Rab+KKabKacKcb12γabRnpnrγaqγbsRpqrs
Substitute R=R+KabKabK2
γauγbvGuv=Rab+KKabKacKcb12γab(R+KabKabK2)npnrγaqγbsRpqrs
γauγbvGuv=Rab12γabR+12γabK2+KKab12γabKabKabKacKcbnpnrγaqγbsRpqrs
... how to get rid of that last npnrγaqγbsRpqrs ?
Maybe you can't. As 2008 Alcubierre after eqn. 2.5.12 says, the Eab=0 turn into the ADM equations.

Lagrangian is equal to scalar curvature:
L=R

Densitized Lagrangian:
L=gL
=gR
=αγR
...substitute Gauss-Codazzi equations...
=αγ(R+KabKabK2
L=αγ(R+KijKijK2) (2008 Alcubierre, eqn. 2.7.6)

Vacuum Action:
S=gRdx4
S=αγ(R+KijKijK2)dx3dt

γ˙ij=2αKij+2(iβj)
Kij=12αγ˙ij+1α(iβj)
γ˙ijKkl=12αδkiδlj

conjugate momentum:
πij=Lγ˙ij
=γ˙ij(αγ(R+KklKklK2))
=γ˙ijαγ(R+γkmγlnKklKmnKklγklKmnγmn)
πij=αγ(δδγ˙ij(γkmγlnKklKmn)δδγ˙ij(KklγklKmnγmn))
πij=αγ(γkmγlnδδγ˙ij(KklKmn)γklγmnδδγ˙ij(KklKmn))
πij=αγ(γkmγlnγklγmn)δδγ˙ij(KklKmn)
πij=αγ(γkmγlnγklγmn)(Kklδδγ˙ijKmn+Kmnδδγ˙ijKkl)
πij=αγ(γkmγlnγklγmn)(12αKklδimδjn+12αKmnδikδjl)
πij=α12αγ((γkmγlnγklγmn)Kklδimδjn+(γkmγlnγklγmn)Kmnδikδjl)
πij=12γ((γkmγlnKklδimδjnγklγmnKklδimδjn)+(γkmγlnKmnδikδjlγklγmnKmnδikδjl))
πij=12γ(KijγijK+KijγijK)
πij=γ(KijγijK)
πij=γ(KijγijK) (2008 Alcubierre, eqn. 2.7.8)
Notice that πij is a tensor density with weight 1.

Extrinsic curvature in terms of conjugate momentum:
γijπij=γij(γ(KijγijK))
solve conjugate momentum trace:
π=γ(γijKijγijγijK)
π=γ(K3K)
π=2γK
K=π/(2γ)
substitute into equation of conjugate momentum in terms of extrinsic curvature:
πij=γ(KijγijK)
πij=γKij+γγijπ/(2γ)
πij=γKij+12γijπ
Kij=1γ(πij12γijπ) (2008 Alcubierre, eqn. 2.7.9)

Extrinsic cuvrature derivative in terms of conjugate momentum:
γ˙ij=2αKij+2(iβj)
substitute Kij=1γ(πij12γijπ):
γ˙ij=2α1γ(πij12γijπ)+2(iβj)

useful identity:
K2KijKij
=(π/(2γ))21γ(πij12γijπ)(πij12γijπ)
=14γπ21γ(πijπij12πγijπij12πγijπij+14πγijγijπ)
=1γ(14π2πijπij+12π2+12π234π2)
=1γ(12π2πijπij)

Hamiltonian density:
H=πijγ˙ijL
=αγ(R+KijKijK2)γ(KijγijK)(2αKij+2(iβj))
=αγ(R+KijKijK2)+2αγ(KijγijK)Kij2γ(KijγijK)(iβj)
=γ(α(R+KijKijK22(KijγijK)Kij)+2(KijγijK)(iβj))
=γ(α(R+KijKijK22KijKij+2K2)+2(KijγijK)(iβj))
=γ(α(RKijKij+K2)+2(KijγijK)(iβj))
=γ(α(RKijKij+K2)2πij1γ(iβj))
=γ(α(R+1γ(12π2πijπij))2πij1γ(iβj))
=α(γR+1γ(πijπij12π2))+2πijiβj
using integration by parts: 2πijjβi=2βijπij
(TODO prove that integration-by-parts works with the spatial projected covariant derivative)
=α(γR+1γ(πijπij12π2))2βiγijkπjk

Somewhere in the middle of this we had:
H=γ(α(RKijKij+K2)2πij1γ(iβj))
...TODO should look like H=2γ(αH+βiMi)
...TODO that means defining earlier ... H=RKijKij+K2 ... but next we define it as the densitized curvature ... which is it?

Constraints:
Let H=γR+1γ(πijπij12π2)
So H=γ(R+K2KijKji)
So H=γ(R+ϵijmϵklmKikKjl)

Let Mi=2jπij
H=αHβiMi
(I'm using Alcubierre's symbols from the numerical relativity community. In the LQG world this usually looks like H=CN+CiNi.)
Assume H=0,Mi=0.
(TODO Alcubierre's book, eqn 2.7.11, defines Hamiltonian as non-densitized and puts the density into this equation. Maybe I should too? Where did I get the densitized version from anyways?)
Let H=Hdx3
S=Hdt

Derivatives of momentum constraint:
βiMi=2βkγkijπij
=2βijπij
using integration-by-parts of spatial covariant derivative (TODO verify you can do this)
=2πijiβj
=πij2(iβj)
using Lβγij=2(iβj)
=πijLβγij
using integration by parts of Lie derivative (TODO verify you can do this)
=γijLβπij
so γij(βkMk)=Lβπij
and πij(βkMk)=Lβγij

Let {f,g} represent the Poisson bracket of f and g, defined as:
{f,g}=fγijgπijfπijgγij

There is also the option to define an extended phase space: {γij,βi,α} and {πij,pi,p}. (think of some letters for those).
The Poisson bracket can then be defined as:
{f,g}=fγijgπijfπijgγij+fβigpifpigβi+fαgpfpgα.
Then choose the phase space hypersurface/subset such that p=0,pi=0. I guess you could pick any values for p and pi so long as they are constant, such that p=0 and pi=0. This causes your Poisson bracket value to reduce to the original definition of {f,g}=fγijgπijfπijgγij. (This doesn't sit right. If a coordinate variable remains constaint, this does not imply that the derivative of another variable wrt that coordinate variable will be constant.)

Poisson bracket of momentum constraint and arbitrary function:
{f(π,γ),βkMk}
=fγijπij(βkMk)fπijγij(βkMk)
=fγijLβγij+fπijLβπij
=Lβf

Derivatives of Hamiltonian constraint:
γijH
=γij(γR+1γ(γkmγlnπklπmn12γklγmnπklπmn))
=(121γRγij(γ)+γγij(R))121γ3/2γij(γ)(γkmγlnπklπmn12γklγmnπklπmn)+2γ(πikπkj12ππij)
=121γRγγijγγij(R)121γ3/2γγij(πklπkl12π2)+2γ(πikπkj12ππij)
=12γRγijγγij(R)121γγij(πklπkl12π2)+2γ(πikπkj12ππij)

πijH
=πij(γR+1γ(γkmγlnπklπmn12γklγmnπklπmn))
=1γ(γimγjnπmn+γkiγljπkl12(γijγmnπmn+γklγijπkl))
=2γ(πij12γijπ)

Derivatives of the Hamiltonian:
γijH=γij(αH+βiMi)
=αγijH
=α(12γRγijγγij(R)121γγij(πklπkl12π2)+2γ(πikπkj12ππij))

πijH=πij(αH+βiMi)
=απijH+βiπijMi
...

Lie bracket of momentum constraint and momentum constraint:
(TODO was this H the Hamiltonian constraint scalar, coupled with α, or was it H the Hamiltonian?).
{βkMk,bkMk}
=γij(βkMk)πij(blMl)πij(βkMk)γij(blMl)
=Lβπijπij(blMl)Lβγijγij(blMl)
=Lβ(bkMk)
=[bkMk,β]
TODO show this is equal to =[b,β]kMk
(i.e. show that LβMk=2Lβjπjk=0)

Lie bracket of momentum constraint and Hamiltonian constraint:
(TODO was this H the Hamiltonian constraint scalar, coupled with α, or was it H the Hamiltonian?).
{βkMk,αH}
=γij(βkMk)πij(αH)πij(βkMk)γij(αH)
=Lβπijπij(αH)Lβγijγij(αH)
=Lβ(αH)
TODO show this is equal to HLβα
i.e. show LβH=Lβ(γR+1γ(πijπij12π2))=0

Lie bracket of Hamiltonian constraint and Hamiltonian constraint:
(TODO was this H the Hamiltonian constraint scalar, coupled with α, or was it H the Hamiltonian?).
{αH,aH}
=γij(αH)πij(aH)γij(aH)πij(αH)
=γij(αH)πij(aH)γij(aH)πij(αH)
TODO show this is equal to (αkaakα)Mk

TODO:
f˙=?ddtf={f,H}, and how does this relate to the Lie derivative, and to the total derivative definition ddt=tLβ and the Lie deriative statements above?
implied by H=Mi=0:
H˙={H,H}=0
M˙i={Mi,H}=0



Back