Sqrt[g(X)]
Cone is an integral of Sqrt[g] over the tube
Small t wrt radius of graph; large t wrt 1
Source of change in Sqrt[g] is divergence or convergence of geodesics
Ricci tensor: area of geodesic bundle
Weyl tensor: [trace free part of Riemann tensor]
1. Individual hyperedge
2. Space seems continuous
3.
4. Size of t he universe
Claim is that between scales 1 and 2, you have to have a perfect “geodesic monopole” to achieve correct dimensional space.
Nothing can cancel R t t. But the higher order terms could cancel in various ways.
Action verion:
R Sqrt[g]
Einstein-Hilbert action is:
bulk geodesic volume

Dimension extremization

Limits

Graph metric: distance along a hyperedge = 1
Physical metric: whole hypergraph is unit size (or 10^26 m), and the edges are scaled accordingly
[[ We could rescale our plots
res=
;
In[]:=
ListLinePlot[res]
In[]:=
Out[]=
Function[u,MapIndexed[{First[#2]/Length[u],#1}&,u]]/@res
In[]:=
Out[]=
ListLinePlot[%]
In[]:=
Out[]=
Computing the spacetime average of
R
μν
​Orientations give us RIntegrating over the spacetime volume gives
g
With each new t (or r) you want the growth rate to not differ from r^d
For large t, growth rate has to be t^(d+1)

Paths

If no refinement

[[Transitivity of geodesics ??]]

JG argument

t is in graph coordinates
t
is in scaled coordinates where the graph diameter is 1 (size of whole graph is T)
Let
t
be small enough that we can series expand
It is the length of the cone that we are studying
​
​
C(
t
)~
d+1
t
(1-
2
t
R)
​
​
The R has units of L^-2, measured with a scaled distance
​
​
What is the global dimension, which goes way above
t
size; this is at the “t scale” (i.e. at the size of the whole graph, i.e. where
t
T~1
)?
​
​
If you average over directions, you are only left with the Ricci part (?)
T
ijkl
x
i
x
j
x
k
x
l
tt=Array[a,{3,3,3,3}]
In[]:=
Out[]=
Icosahedron[]
In[]:=
Icosahedron[]
Out[]=
["VertexCoordinates"]
In[]:=
Out[]=
xx=%165;
In[]:=
Sum[tt[[i,j,k,l]]xx[[n,i]]xx[[n,j]]xx[[n,k]]xx[[n,l]],{n,Length[xx]},{i,3},{j,3},{k,3},{l,3}]//FullSimplify
In[]:=
Out[]=

Momentum

Pick a hypersurface
+ neighbors
​
Number of edges of causal graph slicing a hypersurface
[If the hypersurface is energy]
​
​
​
Momentum + energy are frame dependent.
Given a structure in causal graph:
purely timelike
Excess of edges in CG
​
[Derivative of excess of edges ???] <energy conservation?>
​
There are different ways to increase the number of hyperedges
Think about a buckyball....
​
Excess of edges due to energy is compensated by a decrease in edges due to positive curvature
Fixed energy density : 1 + δρ

Energy momentum

vacuum energy +

E = m
2
c

Projecting edges through spacelike surface

Continuum Limit

Foliations are the way we idealize the observer
[foliations are we turn reality into the subjective experience of the observer]
If we want a simple description of physics, need to have simple foliations

Quantum analog

Non-interacting branchial graph [ bunch of separated pieces without entanglement ] : free field case
​
Boost is like a transformation between observers
​
​
Basic indeterminancy: you can go either way on a critical pair. But entanglement in branchial graph.
​
Curvature measures entanglement because it is a sign of non-commutation
​
[x, p] = i ℏ
​
Go across the spatial graph and down the causal graph
​
Operators vs. states ....