[time 944] Goedel's incompleteness implies the existence of time

Hitoshi Kitada (hitoshi@kitada.com)
Thu, 21 Oct 1999 20:25:07 +0900

Here is some excerpt from time_VI.tex. LaTeX file is attached, which is
available also at

http://www.kitada.com/time_VI.tex (the link is not yet made in index.html)

A key is Goedel's incompleteness theorem, which assures the existence of
(local) time.

Hitoshi Kitada

A possible solution for the non-existence of time

(October 21, 1999 version)

Abstract. A possible solution for the problem of
 non-existence of universal time is given.

In a recent book \cite{Barbour}, Barbour presented
 a thought that time is an illusion, by noting that
 Wheeler-DeWitt equation yields the non-existence of
 time, whereas time around us looks flowing.
 However, he does not seem to have a definite idea
 to actualize his thought. In the present note I present
 a concrete way to resolve the problem of non-existence
 of time, which is partly a repetition of my works
 \cite{K1}, \cite{K2}, \cite{Ki-Fl1}, \cite{Ki-Fl2}.

1. Time seems not exist

According to equation (5.13) in Hartle \cite{CP},
 the non-existence of time would be expressed by
 an equation:
H \Psi=0. \label{1}
Here $\Psi$ is the ``state" of the universe
belonging to a suitable Hilbert space $\HH$, and
$H$ denotes the total Hamiltonian of the universe
 defined in $\HH$. This equation implies that there
 is no global time of the universe, as the state
 $\Psi$ of the universe is an eigenstate for the total
 Hamiltonian $H$, and hence does not change.
One might think that this implies the non-existence
 of local time because any part of the universe is
 described by a part of $\Psi$. Then we have no time,
 contradicting our observation. This is a restatement
 of the problem of time, which is a general problem to
 identify a time coordinate with preserving the
 diffeomorphism invariance. In fact, equation \eq{1}
 follows if one assumes the existence of a preferred
 foliating family of spacelike surfaces in spacetime
 (see section 5 of \cite{CP}).

We give a solution in the paper to this problem that
 on the level of the total universe, time does not exist,
 but on the local level of our neighborhood, time looks

2. G\"odel's theorem

Our starting point is the incompleteness theorem proved by
 G\"odel \cite{G}. It states that any formal theory that can
 describe number theory includes an infinite number of undecidable
 propositions. The physical world we describe includes at least
 natural numbers, and it is described by a system of words, which
 can be translated into a formal theory. The theory of physics
 therefore includes an undecidable proposition, i.e. a proposition
 whose correctness cannot be known by human beings until one finds
 a phenomenon or observation that supports the proposition or denies
 the proposition. Such propositions exist infinitely according to
 G\"odel's theorem. Thus human beings can never reach the final
 theory that can express the totality of the phenomena in the

Thus we have to assume that any human observer sees a part
 or subsystem $L$ of the universe and never gets the total
 Hamiltonian $H$ in \eq{1} by his observation. Here the
 total Hamiltonian $H$ is an {\it ideal} Hamiltonian
that might be gotten by ``God." In other words, a consequence
  from G\"odel's theorem is that the Hamiltonian that an
 observer assumes with his observable universe
 is a part $H_L$ of $H$. Stating explicitly,
the consequence from G\"odel's theorem is the
 following proposition
H=H_L+I+H_E, H_E not = 0, \label{2}
where $H_E$ is the unknown Hamiltonian describing
 the system $E$ exterior to the realm of the observer,
 whose existence, i.e. $H_E not = 0$, is assured by G\"odel's
 theorem. This unknown system $E$ includes
all what is unknown to the observer.
E.g., it might contain particles which
 exist near us but have not been discovered yet.
The term $I$ is an unknown interaction between
 the observed system $L$ and the unknown system $E$.
Since the exterior system $E$ is assured to exist
by G\"odel's theorem, the interaction $I$ does not vanish:
In fact if $I$ vanished, then one could not know that
 the observed system $L$ and the exterior system $E$ interacts
 and hence could not know that the exterior system $E$ exists,
 which contradicts G\"odel's theorem. By the same reason,
 $I$ is not a constant operator:
I not = constant operator. \label{3}
For suppose it is a constant operator. Then
the systems $L$ and $E$ do not change how far or
 near they are located because the interaction
 between $L$ and $E$ is a constant operator.
 Hence the observer cannot know that $E$ exists,
 contradicting G\"odel's theorem.

We now arrive at the following observation:
For an observer, the observable universe is a part $L$
of the total universe and it looks following the
 Hamiltonian $H_L$, not following the total Hamiltonian $H$.
 And the state of the system $L$ is described by a part
 $\Psi(.,y)$ of the state $\Psi$ of the total universe,
 where $y$ is an unknown coordinate of the system $L$ inside
 the total universe, and $.$ is the variable controllable
 by the observer, which we will denote by $x$.

3. Local Time Exists

Assume now, as is expected usually, that there is no
 local time of $L$, i.e. that the state $\Psi(x,y)$
 is an eigenstate of the local Hamiltonian $H_L$ for
 some $y=y_0$:
H_L\Psi(x,y_0)=0. \label{4}
Then from \eq{1}, \eq{2} and \eq{4} follows that
&&\ \hskip5pt=I(x,y_0)\Psi(x,y_0)+H_E\Psi(x,y_0). \label{5}
Here $x$ varies over the possible positions of the particles
 inside $L$. On the other hand, since $H_E$ is the Hamiltonian
 describing the system $E$ exterior to $L$, it does not
 affect the variable $x$ and acts only on the variable $y$.
 Thus $H_E\Psi(x,y_0)$ varies as a bare function $\Psi(x,y_0)$
 insofar as the variable $x$ is concerned.
Equation \eq{5} is now written: For all $x$
H_E\Psi(x,y_0)=-I(x,y_0)\Psi(x,y_0). \label{6}
As we have seen in \eq{3}, the interaction $I$
is not a constant operator and varies when $x$
 varies\footnote[2]{Note that G\"odel's theorem
 applies to any fixed $y=y_0$ in \eq{3}. Namely,
 for any position $y_0$ of the system $L$ in the
 universe, the observer must be able to know
 that the exterior system $E$ exists because
 G\"odel's theorem is a universal statement
 valid throughout the universe.
 Hence $I(x,y_0)$ is not a constant operator
 with respect to $x$ for any fixed $y_0$.},
 whereas the action of $H_E$ on $\Psi$ does not.
 Thus there is a nonempty set of points $x_0$
 where $H_E\Psi(x_0,y_0)$ and $-I(x_0,y_0)\Psi(x_0,y_0)$
 are different, and \eq{6} does not hold at such points
 $x_0$. If $I$ is assumed to be continuous in the variables
 $x$ and $y$, these points $x_0$ constitutes a set of
 positive measure. This then implies that our assumption
 \eq{4} is wrong. Thus a subsystem $L$ of the universe cannot
 be a bound state with respect to the observer's Hamiltonian
 $H_L$. This means that the system $L$ is observed as
 a non-stationary system, therefore there must be observed
 a motion inside the system $L$. This proves that the
 ``time" of the local system $L$ {\it exists for the
 observer} as a measure of motion, whereas the total
 universe is stationary and does not have ``time."

4. A refined argument

(A rather technical part and is omitted.)

5. Conclusion

G\"odel's proof of the incompleteness theorem relies on the
 following type of proposition $P$ insofar as the meaning is concerned:
P = ``P cannot be proved." \label{8}
Then if P is provable it contradicts P itself, and if P is not
 provable, P is correct and seems to be provable. Both cases lead
 to contradiction, which makes this kind of propositions undecidable
 in a given formal theory.

This proposition reminds us the following type of self-referential
A person P says ``I am telling a lie." \label{9}
Both of this and proposition P in \eq{8} are non-diagonal statements
 in the sense that both denies themselves. Namely the core of
 G\"odel's theorem is in proving the existence of non-diagonal
 ``elements" (i.e. propositions) in any formal theory that includes
 number theory. By constructing such propositions in number theory,
 G\"odel's theorem shows that any formal theory has a region exterior
 to the knowable world.

On the other hand, what we have deduced from G\"odel's theorem in
 section 2 is that the interaction term $I$ is not a constant operator.
 Moreover the argument there implies that $I$ does not commute with
 at least one of $H_L$ and $H_E$. For suppose that $I$ commutes with
 both of $H_L$ and $H_E$. Then by spectral theory for
 selfadjoint operators, $I$ is decomposed as $I=f(H_L)+g(H_E)$ for
 some functions $f(H_L)$ and $g(H_E)$ of $H_L$ and $H_E$.
Thus $H$ is decomposed as a sum of mutually commuting operators:
 $H=(H_L+f(H_L))+(H_E+g(H_E))$. Here the Hilbert space $\HH$ is
 decomposed as a direct sum of two direct integrals:
\HH=\int^\oplus \HH_L(\lambda)d\lambda\oplus
\int^\oplus \HH_E(\mu)d\mu, \label{10}
where the first term on the RHS is the decomposition of $\HH$ with
 respect to the spectral representation of $H_L$, and the second is
 the one with respect to that of $H_E$. In this decomposition,
 $H$ is decomposed as a diagonal operator:
H=\int^\oplus (\lambda+f(\lambda))d\lambda\oplus
\int^\oplus (\mu+g(\mu))d\mu.
Namely the total Hamiltonian $H$ is decomposed into a sum of
 mutually independent operators in the decomposition of the
 total system into the observable and unobservable systems $L$
 and $E$. This means that there are no interactions between $L$
 and $E$, contradicting G\"odel's theorem as in section 2.
 Thus $I$ does not commute with one of $H_L$ and $H_E$.
 Therefore $I$ is not diagonalizable with respect to the direct
 integral decomposition \eq{10} of the space $\HH$.

Now a consequence of G\"odel's theorem in the context
 of the decomposition of the total universe into observable and
 unobservable systems $L$ and $E$ is the following:

    In the spectral decomposition \eq{10} of
    $\HH$ with respect to a decomposition
    of the total system into the observable and
    unobservable ones, $I$ is non-diagonalizable.
    In particular so is the total Hamiltonian

Namely G\"odel's theorem yields the existence of non-diagonal
 elements in the spectral representation of $H$ with respect to
 the decomposition of the universe into observable and
 unobservable systems. The existence of non-diagonal
 elements in this decomposition is the cause that the
 observable state $\Psi(.,y)$ is not a stationary state
 and local time arises, and that decomposition is inevitable by
 the existence of the region unknowable to human beings.

 From the standpoint of the person P in \eq{9}, his universe needs
 to proceed to the future for his statement to be decided true or
 not, the decision of which requires his system an infinite ``time."
 This is due to the fact that his self-destructive statement does
 not give him satisfaction in his own world and forces him
 to go out to the region exterior to his universe.
 Likewise, the interaction $I$ in the decomposition above
 forces the observer to anticipate the existence of the
 region exterior to his knowledge. In both cases the unbalance
 caused by the existence of the exterior region yields time.
 In other words, time is an indefinite desire to reach the
 balance that only the universe has.


\bibitem{Barbour} Julian Barbour, ``The End of Time,"
 Weidenfeld & Nicolson, 1999.

\bibitem{G} K. G\"odel, On formally undecidable propositions of
 Principia mathematica and related systems I, in ``Kurt G\"odel
 Collected Works, Volume I, Publications 1929-1936," Oxford University
 Press, New York, Clarendon Press, Oxford, 1986, pp.144-195,
 translated from \"Uber formal unentsceidebare S\"atze der
 Principia mathematica und verwandter Systeme I, Monatshefte
 fur Mathematik und Physik, 1931.

\bibitem{CP} J. B. Hartle, Time and Prediction in
 Quantum Cosmology, in ``Conceptual
 Problems of Quantum Gravity," Einstein Studies Vol. 2,
 Edited by A. Ashtekar and J. Stachel, Birkh\"auser,
 Boston-Basel-Berlin, 1991.

\bibitem{K1} H. Kitada, Theory of local times,
Il Nuovo Cimento 109 B, N. 3 (1994), 281-302.

\bibitem{K2} H. Kitada, Quantum Mechanics and Relativity
--- Their Unification by Local Time,
 in ``Spectral and Scattering Theory,"
Edited by A.G.Ramm, Plenum Publishers, New York,
 pp. 39-66, 1998. (http://xxx.lanl.gov/abs/gr-qc/9612043,

\bibitem{Ki-Fl1} H. Kitada and L. Fletcher, Local time and
 the unification of physics, Part I: Local time, Apeiron
3 (1996), 38-45. (http://kims.ms.u-tokyo.ac.jp/time_III.tex,

\bibitem{Ki-Fl2} H. Kitada and L. Fletcher,
Comments on the Problem of Time.

This archive was generated by hypermail 2.0b3 on Sun Oct 24 1999 - 19:01:01 JST