123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262 |
- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
- % Twisting type N solutions of GR %
- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
- % The problem is to analyse an ansatz for a particular type of vacuum
- % solution to Einstein's equations for general relativity. The analysis was
- % described by Finley and Price (Proc Aspects of GR and Math Phys
- % (Plebanski Festschrift), Mexico City June 1993). The equations resulting
- % from the ansatz are:
- % F - F*gamma = 0
- % 3 3
- %
- % F *x + 2*F *x + x *F - x *Delta*F = 0
- % 2 2 1 2 1 2 1 2 2 1
- %
- % 2*F *x + 2*F *x + 2*F *x + 2*F *x + x *F = 0
- % 2 3 2 3 2 2 3 3 3 2 2 3 3 2 2 3 2 2 3 3
- %
- % Delta =0 Delta neq 0
- % 3 1
- %
- % gamma =0 gamma neq 0
- % 2 1
- % where the unknowns are {F,x,gamma,Delta} and the indices refer to
- % derivatives with respect to an anholonomic basis. The highest order is 4,
- % but the 4th order jet bundle is too large for practical computation, so
- % it is necessary to construct partial prolongations. There is a single
- % known solution, due to Hauser, which is verified at the end.
- on evallhseqp,edssloppy,edsverbose;
- off arbvars,edsdebug;
- pform {F,x,Delta,gamma,v,y,u}=0;
- pform v(i)=0,omega(i)=1;
- indexrange {i,j,k,l}={1,2,3};
- % Construct J1({v,y,u},{x}) and transform coordinates. Use ordering
- % statement to get v eliminated in favour of x where possible.
- % NB Coordinate change cc1 is invertible only when x(-1) neq 0.
- J1 := contact(1,{v,y,u},{x});
- korder x(-1),x(-2),v(-3);
- cc1 := {x(-v) = x(-1),
- x(-y) = x(-2),
- x(-u) = -x(-1)*v(-3)};
- J1 := restrict(pullback(J1,cc1),{x(-1) neq 0});
- % Set up anholonomic cobasis
- bc1 := {omega(1) = d v - v(-3)*d u,
- omega(2) = d y,
- omega(3) = d u};
- J1 := transform(J1,bc1);
- % Prolong to J421: 4th order in x, 2nd in F and 1st in rest
- J2 := prolong J1$
- J20 := J2 cross {F}$
- J31 := prolong J20$
- J310 := J31 cross {Delta,gamma}$
- J421 := prolong J310$
- cc4 := first pullback_maps;
- % Apply first order de and restrictions
- de1 := {Delta(-3) = 0,
- gamma(-2) = 0,
- Delta(-1) neq 0,
- gamma(-1) neq 0};
- J421 := pullback(J421,de1)$
- % Main de in original coordinates
- de2 := {F(-3,-3) - gamma*F,
- x(-1)*F(-2,-2) + 2*x(-1,-2)*F(-2)
- + (x(-1,-2,-2) - x(-1)*Delta)*F,
- x(-2,-3)*(F(-2,-3)+F(-3,-2)) + x(-2,-2,-3)*F(-3)
- + x(-2,-3,-3)*F(-2) + (1/2)*x(-2,-2,-3,-3)*F};
- % This is not expressed in terms of current coordinates.
- % Missing coordinates are seen from 1-form variables in following
- d de2 xmod cobasis J421;
- % The necessary equation is contained in the last prolongation
- pullback(d de2,cc4) xmod cobasis J421;
- % Apply main de
- pb1 := first solve(pullback(de2,cc4),{F(-3,-3),F(-2,-2),F(-2,-3)});
- Y421 := pullback(J421,pb1)$
- % Check involution
- on ranpos;
- characters Y421;
- dim_grassmann_variety Y421;
- % 15+2*7 = 29 > 28: Y421 not involutive, so prolong
- Y532 := prolong Y421$
- characters Y532;
- dim_grassmann_variety Y532;
- % 22+2*6 = 34: just need to check for integrability conditions
- torsion Y532;
- % Y532 involutive. Dimensions?
- dim Y532;
- length one_forms Y532;
- % The following puts in part of Hauser's solution and ends up with an ODE
- % system (all characters 0), so no more solutions, as described by Finley
- % at MG6.
- hauser := {x=-v+(1/2)*(y+u)**2,delta=3/(8x),gamma=3/(8v)};
- H532 := pullback(Y532,hauser)$
- lift ws;
- characters ws;
- clear v(i),omega(i);
- clear F,x,Delta,gamma,v,y,u,omega;
- off ranpos;
- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
- % Isometric embeddings of Ricci-flat R(4) in ISO(10) %
- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
- % Determine the Cartan characters of a Ricci-flat embedding of R(4) into
- % the orthonormal frame bundle ISO(10) over flat R(6). Reference:
- % Estabrook & Wahlquist, Class Quant Grav 10(1993)1851
- % Indices
- indexrange {p,q,r,s}={1,2,3,4,5,6,7,8,9,10},
- {i,j,k,l}={1,2,3,4},{a,b,c,d}={5,6,7,8,9,10};
- % Metric for R10
- pform g(p,q)=0;
- g(p,q) := 0$ g(-p,-q) := 0$ g(-p,-p) := g(p,p) := 1$
- % Hodge map for R4
- pform epsilon(i,j,k,l)=0;
- index_symmetries epsilon(i,j,k,l):antisymmetric;
- epsilon(1,2,3,4) := 1;
- % Coframe for ISO(10)
- % NB index_symmetries must come after o(p,-q) := ... (EXCALC bug)
- pform e(r)=1,o(r,s)=1;
- korder index_expand {e(r)};
- e(-p) := g(-p,-q)*e(q)$
- o(p,-q) := o(p,r)*g(-r,-q)$
- index_symmetries o(p,q):antisymmetric;
- % Structure equations
- flat_no_torsion := {d e(p) => -o(p,-q)^e(q),
- d o(p,q) => -o(p,-r)^o(r,q)};
- % Coframing structure
- ISO := coframing({e(p),o(p,q)},flat_no_torsion)$
- dim ISO;
- % 4d curvature 2-forms
- pform F(i,j)=2;
- index_symmetries F(i,j):antisymmetric;
- F(-i,-j) := -g(-i,-k)*o(k,-a)^o(a,-j);
- % EDS for vacuum GR (Ricci-flat) in 4d
- GR0 := eds({e(a),epsilon(i,j,k,l)*F(-j,-k)^e(-l)},
- {e(i)},
- ISO)$
- % Find an integral element, and linearise
- Z := integral_element GR0$
- GRZ := linearise(GR0,Z)$
- % This actually tells us the characters already:
- % {45-39,39-29,29-21,21} = {6,10,8,21}
- % Get the characters and dimension at Z
- characters GRZ;
- dim_grassmann_variety GRZ;
- % 6+2*10+3*8+4*21 = 134, so involutive
- clear e(r),o(r,s),g(p,q),epsilon(i,j,k,l),F(i,j);
- clear e,o,g,epsilon,F,Z;
- indexrange 0;
- %%%%%%%%%%%%%%%%%%%%%%%%%%
- % Janet's PDE system %
- %%%%%%%%%%%%%%%%%%%%%%%%%%
- % This is something of a standard test problem in analysing integrability
- % conditions. Although it looks very innocent, it must be prolonged five
- % times from the second jet bundle before reaching involution. The initial
- % equations are just
- %
- % u =w, u =u *y + v
- % y y z z x x
- load sets;
- off varopt;
- pform {x,y,z,u,v,w}=0$
- janet := contact(2,{x,y,z},{u,v,w})$
- janet := pullback(janet,{u(-y,-y)=w,u(-z,-z)=y*u(-x,-x)+v})$
- % Prolong to involution
- involutive janet;
- involution janet;
- involutive ws;
- % Solve the homogeneous system, for which the
- % involutive prolongation is completely integrable
- fdomain u=u(x,y,z),v=v(x,y,z),w=w(x,y,z);
- janet := {@(u,y,y)=0,@(u,z,z)=y*@(u,x,x)};
- janet := involution pde2eds janet$
- % Check if completely integrable
- if frobenius janet then write "yes" else write "no";
- length one_forms janet;
- % So there are 12 constants in the solution: there should be 12 invariants
- length(C := invariants janet);
- solve(for i:=1:length C collect
- part(C,i) = mkid(k,i),coordinates janet \ {x,y,z})$
- S := select(lhs ~q = u,first ws);
- % Check solution
- mkdepend dependencies;
- sub(S,{@(u,y,y),@(u,z,z)-y*@(u,x,x)});
- clear u(i,j),v(i,j),w(i,j),u(i),v(i),w(i);
- clear x,y,z,u,v,w,C,S;
- end;
|