Controllability of nonholonomic

Black Holes systems

Constantin Udriste, Vincenzo Ciancio, Francesco Farsaci
September 26, 2011
This paper studies the sub-Lorentz-Vranceanu geometry and the optimal control of non-
holonomic black hole systems. This is strongly connected to the possibility of describing a
nonholonomic black hole system as kernel of a Gibbs-Pfa form or by the span of four ap-
propriate vector elds. Joining techniques from sub-Riemannian geometry, optimal control
and thermodynamics, we bring into attention new models of black holes systems. These
are reected by the original results: a Lorentz-Vranceanu geometry on the total space,
a new sub-Lorentz-Vranceanu geometry, a new stress-energy-momentum tensor, original
solutions to Einstein eld equations, and the controllability of nonholonomic black holes
systems by uni-temporal or bi-temporal controls.
Mathematics Subject Classication 2010: 83C57, 93B05, 37J60,
57R15.
Keywords: nonholonomic black hole, optimal control, Lorentz-Vranceanu
geometry, nonholonomic dynamical systems.
1 Introduction
At the beginning, the physics of black holes was largely based on Einsteins
general theory of relativity (GR), which is a theory of gravitation in terms
of Dierential Geometry. In this way, the GR physics is expressed in terms
of tensors and each kind of spacetime is assigned with the proper time and
space interval which are described by the line element or metric. In 1972,
James Bardeen, Brandon Carter and Stephen Hawking discover that the
black holes are intimately related to ordinary thermodynamics in the sense
1
that their mechanical laws are seemingly identical to the laws of the ther-
modynamics. Following the framework of Carathodorys approach to ther-
modynamics, our papers [4], [7], [10]-[12], [17]-[20] implemented a model of
black hole thermodynamics as nonholonomic system described by a Gibbs-
Pfa equation (a 4-dimensional nonholonomic hypersurface [29] as collection
of integral curves and integral surfaces). Our aim is to join the classical ideas
in dierential geometry to the contemporary ideas regarding sub-Riemannian
geometry, optimal control, subelliptic PDEs, Carnot and Caratheodory the-
ory, Hamilton-Jacobi theory and calibrations (see [2], [3], [6], [8], [13]) in
order to create fresh and new information about nonholonomic black hole
systems (see [4], [7], [10]-[12], [17]-[20]).
Section 2 is a survey relating the sub-Riemannian geometry, the optimal
control theory, the nonholonomic geodesics, the subelliptic PDEs, the second
law of thermodynamics and Hamilton-Jacobi theory and calibrations. The
original results in Section 3 include a nonholonomic black hole system as a
4-dimensional distribution, the Vranceanu congruences method, a Lorentz-
Vr anceanu metric on the total space and its geodesics. A new stress-energy-
momentum tensor on Lorentz-Vr anceanu manifold and an original polyno-
mial solution for associated Einstein eld equations are given in Section 4.
The geodesics on the black hole nonholonomic thermodynamic system are
discussed in Section 5, via solutions to variational problem of geodesics and
an optimal control problem of geodesics. Section 6 formulates and solves a
two-time optimal control on black hole system. The main results of Sections
5 and 6 show that the black hole nonholonomic thermodynamic system is
controllable by uni-temporal or bi-temporal controls. Section 6 underlines
the mathematical roots of our original results regarding the black holes.
2 Sub-Riemannian geometry and
optimal control
Here we refer to the sub-Riemannian geometry, but we have also in mind
the sub-semi-Riemannian geometry which is specic for many problems in
Physics.
2
2.1 Sub-Riemannian geometry
Let M be an n-dimensional manifold and TM be its tangent bundle. Sup-
pose, for each point x M, we attach an m-plane D
x
T
x
M, and by letting
x vary we obtain a smoothly varying family of m-planes on M. This rule
denes a smooth linear subbundle D TM. Put a smoothly varying family
g of inner products on each m-plane, called sub-Riemannian metric.
The pair (M, D) is called distribution of dimension m on M, the triple
(M, D, g) is called a sub-Riemannian manifold and the pair (D, g) is called
sub-Riemannian structure. This structure determines a sub-Riemannian ge-
ometry [2], [3], [6], [8], [13].
A more simple alternative to refer on the foregoing notions start from C
2
smooth vector elds X
1
(x), ..., X
m
(x) on M, which are linearly independent
at each point x M. These elds dene an m-dimensional distribution D
via x D
x
= span{X
1
(x), ..., X
m
(x)}. A real function q(x, x) of x M
and x D
x
is called a submetric on D
x
if, for any chosen x M, this
function denes a positive sublinear functional of x. We assume that q is
continuous with respect to (x, x). In particular, if q
2
(x, x) = g(x)( x, x) is a
positive-denite quadratic form of x D
x
, then q is called a sub-Riemannian
metric, and D together with q dene a sub-Riemannian structure on M. A
more general case (which is still special) is related to a sub-Finsler metric
for which q(x, x) = q(x, x). Representing vectors x D
x
in the form x =
u
a
X
a
(x), a = 1, m, we can dene and use the function (x, u) = q(x, x) =
q(x, u
a
X
a
(x)) [3].
We assume that the distribution D is not integrable. A piecewise smooth
curve (t), t [a, b] in M is called horizontal if (t) D
(t)
a.e. t [a, b].
Using the kinetic energy
1
2
g( , ), we dene the energy E() of a horizontal
curve (t), t [a, b] as the integral E() =
1
2
_
t
1
t
0
g( , )dt.
Take the viewpoint that we can explore M only by traveling along paths
tangent to D, called horizontal paths. Since the metric g determines lengths
and angles of horizontal D-vectors, we can compute lengths of horizontal
paths and formulate the sub-Riemannian geodesic problem: nd the shortest
horizontal path connecting two given points. Gromov and his school use
the term Carnot-Caratheodory geometry for what we call sub-Riemannian
geometry.
Usually the sub-Riemannian geometry is divided into three streams of
3
research, according to the methods and motivations used: (1) A stream
arising out of optimal control and surrounding the endpoint map and sub-
Riemannian geodesic problem. We are most familiar with this stream. (2) A
stream coming from partial dierential equations (PDEs), exemplied by a
theorem of Hormander. (3) A stream concerning sub-Riemannian geometries
arising out of limits of Riemannian or other geometries.
2.2 Optimal control
Single-time case. The recent deepest results on sub-Riemannian geodesic
problems have been obtained by mathematicians trained in single-time opti-
mal control [2], [3], [6], [8], [13]. One of the basic problem in single-time non-
linear control theory is to construct paths (t) connecting two points x
0
, x
1
of
a manifold M, subject to path constraints of the form (t) V ((t)), where
V (x) T
x
M are subsets varying with the point x. The sets V (x) are repre-
sented as images of a map u f(x, u), where the control u varies over some
auxiliary set, usually a subset of a Euclidean space. To relate this setup
to sub-Riemannian geometry, set V (x) = D
x
. If we use a Lagrange func-
tion L(x, u) and look to minimize the functional
_
t
1
t
0
L((t), u(t))dt over the
space of all controls u(t) which steer from x
0
to x
1
, i.e., = f(, u(t)), (t
0
) =
x
0
, (t
1
) = x
1
, then we have formulated a basic problem in the single-time
optimal control. For V (x) = D
x
, (t) = u
a
(t)X
a
((t)) and L =
1
2
g( , ),
we represent the sub-Riemannian geodesic problem as a problem in optimal
control. Usually, the set X

is considered as an orthonormal frame and the

components u

are functions in L
2
, i.e.,
1
2
_
t
1
t
0

ab
u
a
(t)u
b
(t)dt < .
The rst question that arises when facing the sub-Riemannian geodesic
problem is accessibility: is there any horizontal path at all which connects the
two given points? If D is involutive, i.e., Lie brackets of horizontal vector
elds remain horizontal, then, by the Frobenius integrability theorem, the
answer is no for generic pairs of points. The Frobenius theorem asserts that
the set of points connected to x
0
by a horizontal path form an immersed
submanifold called the leaf of D hrough x
0
. A given point x
1
that is not
on this leaf cannot be horizontally connected to x
0
. The Chow-Rashevskii
theorem [6], [13] asserts that for D at the opposite extreme of involutive,
the answer is yes, locally. The technical term for at the opposite extreme of
involutive is bracket generating. A distribution D is called bracket generat-
4
ing if its horizontal vector elds, X
1
, X
2
, ..., together with their iterated Lie
brackets, [X
a
, X
b
], [X
a
, [X
b
, X
c
]], ..., span the tangent bundle to M.
Theorem [6]. Let be the space of all absolutely continuous paths on
the interval [t
0
, t
1
], whose derivatives are a.e. horizontal and in L
2
with
respect to the sub-Riemannian metric supported on the distribution D. Let

0
be the subspace of horizontal square-integrable paths beginning at a
xed point x
0
M. Then
0
is a Hilbert manifold with coordinate charts
u L
2
([t
0
, t
1
], R
m
) dened by the ODE (t) = u
a
(t)X
a
((t)). The
endpoint map (1) is a smooth map M relative to this manifold
structure. The endpoint map is open if D is bracket generating. The set
inherits both a strong and a weak topology from the coordinate charts and
these topologies on L
2
. The inclusion map C
0
([t
0
, t
1
], M) is continuous
with respect to both topologies.
Multitime case. The basic problem in multitime nonlinear control the-
ory [21]-[28] is to construct q-sheets x(t), t = (t

)
t
0
t
1
R
q
+
, = 1, q,
connecting two points x
0
, x
1
of a manifold M, subject to sheet constraints
of the form x(t) V (x(t)), where V (x) T
x
M are subsets varying with the
point x. The sets V (x) are represented as images of a map u F(x, u) =
(F

(x, u)), where the control u varies over some auxiliary set, usually a subset
of a Euclidean space. To relate this setup to sub-Riemannian geometry, set
V (x) = D
x
and x(t) at most the maximal integral submanifolds of the distri-
bution D
x
. If we use a Lagrange function L(x, u) and look to minimize the
functional
_
t
0
t
1
L(x(t), u(t))dt
1
...dt
m
over the space of all controls u(t) which
steer from x
0
to x
1
, i.e.,
x
t

(t) = F

(x(t), u(t)), x(t

0
) = x
0
, x(t
1
) = x
1
, then
we have formulated the basic problem in optimal control. For V (x) = D
x
,
x
t

(t) = u
a

(t)X
a
(x(t)), a = 1, m, = 1, q, and L =
1
2
g
ij

x
i

x
j

, we rep-
resent the sub-Riemannian q-sheet problem as a problem in multitime op-
timal control. Usually, the set X
a
is considered as an orthonormal frame
and the components u
a

of the control matrix u are functions in L

2
, i.e.,
1
2
_
t
1
t
0

ab

u
a

(t)u
b

(t)dt
1
...dt
m
< .
2.3 Nonholonomic geodesics
Nonholonomic systems are mechanical systems with constraints on their ve-
locity that are not derivable from position constraints. In nonholonomic
Lagrangian mechanics, there are two well known approaches for the study
5
of the constrained mechanics: dAlembertian nonholonomic mechanics and
the variational nonholonomic mechanics. The variational nonholonomic me-
chanics, also called vakonomic mechanics by the Russian school, is to solve
a constrained variational problem: to nd a curve such that minimizes
the Lagrangian functional
_
t
1
t
0
L((t), (t))dt among all curves satisfying
the (nonholonomic) constraints (t) D
(t)
, where D is a (nonintegrable)
distribution. When the Lagrangian L is regular, and quadratic with respect
to the velocity component, the required curve is just a sub-Riemannian
geodesic. The dynamics studied by dAlembertian nonholonomic mechanics
is governed by the Lagrange-dAlembert principle: the equations of motion
of a curve x(t) in a conguration space are obtained by setting to zero the
variations in the integral of the Lagrangian subject to variations lying in
the constraint distribution, L =
_
d
dt
L
x
i

L
q
i
_
x
i
= 0, for all variations
x D
x
, and that the velocity of the curve x(t) itself satises the constraints
x(t) D
x(t)
. There is a huge literature on nonholonomic mechanics.
The following theorem is well known [6].
Theorem. Assume (M, D, g) be a sub-Riemannian manifold, where g
is the restriction to D of a given Riemannian metric g on the manifold M.
If the Lagrangian L is the kinetic energy L(x, x) =
1
2
g( x, x), then by using
the horizontal connection , the Euler-Lagrange equations of motion can be
rewritten as
x
x = 0.
Denition (nonholonomic geodesic). On a sub-Riemannian mani-
fold (M, D, g), any horizontal curve satisfying

= 0 is called a non-
holonomic geodesic.
Suppose that the orthonormal covector elds
a
i
, a = 1, 2, ..., n m; i =
1, 2, ..., n, dene the distribution D of dimension m on the manifold M and

i
jk
are the Christoel symbols on (M, g). Denote

a
i;j
=

a
i
x
j

k
ji

a
k
and dene the horizontal connection

k
ji
=
k
ji
+
1
2
_

a
i;j
+
a
j;i
_

k
a
.
Now, the equation of nonholonomic geodesics is a system of second order
ODEs
d
2
x
i
dt
2
(t) +
i
jk
(x(t))
dx
j
dt
(t)
dx
k
dt
(t) = 0.
6
It is well known that Riemannian geodesics are the projections on M of the
integral curves of a spray. For nonholonomic geodesics, we have a similar
characterization.
2.4 Subelliptic PDEs
Chow-Rashevskii bracket generating condition reappeared in Hormander pa-
pers. Let X
a
, a = 1, m be a (local) orthonormal frame for the distribution D.
These vector elds dene a sub-Laplacian =
m

a=1
X
2
a
. Hormander theorem
asserts that acts like the usual Laplacian (close analogues of the usual
elliptic regularity theorems), provided that D is bracket generating. That is
why the bracket generating condition is often referred to as Hormander con-
dition in certain PDE literature. Most problems in Riemannian geometric
analysis involve the Riemannian Laplacian in time that the sub-Riemannian
Laplacians are central to sub-Riemannian geometric analysis. Over the last
ten years there has been enormous work on the sub-Riemannian versions of
the minimal surface equations, constant mean curvature equations, Yamabes
problem, etc. Most of the analysis has been on Carnot groups (see also [6]).
2.5 Carnot and Caratheodory
Carnot, working on heat engines, realized that certain thermodynamic states
are inaccessible from other states. At the urging of his friend the physicist
Max Born, Caratheodory used a local version of inaccessibilty to derive the
second law of thermodynamics, and of entropy. What he established was
a kind of converse to Chow-Rashevskii for co-rank 1 (m = n 1) distribu-
tions D. Caratheodory showed that if every point in (M, D) has a nearby
inaccessible point, then the distribution D is involutive.
2.6 Hamilton-Jacobi theory and calibrations
In order to prove that the normal geodesics are locally minimizing geodesics,
we use the method of calibrations (see also [6]).
Denition. A closed 1-form on a sub-Riemannian manifold which
satises |(v)| ds(v) =

< v, v >, for all horizontal vectors v, is called
7
calibration. A horizontal curve is called calibrated by the calibration if
( ) = ds( ).
Lemma. If is a calibrated horizontal curve, then each suciently short
subarc of is a minimizing geodesic.
Proof. Let
1
be any horizontal uncalibrated curve with the same end-
points as that is homologous to . Then
_

1
=
_

by Stokes formula
and
_

1
ds
_

1
(uncalibrated) and
_

ds =
_

(calibrated). It follows
(
1
) ().
The Hamilton-Jacobi PDE H(x, dS(x)) =
1
2
(partial dierential equation
for S) provides a method for constructing calibrations.
Lemma. If S is a solution of the Hamilton-Jacobi PDE, then the 1-form
= dS is a calibration.
Proof. The 1-form dS is closed. Let X
a
, a = 1, m be a local orthonor-
mal frame from the horizontal distribution D. Write v D as v = v
a
X
a
.
Then ds(v)
2
=
ab
v
a
v
b
and dS(v) = v
a
dS(X
a
) and by the Cauchy-Schwarz
inequality we nd
|dS(v)| ds(v)
_

ab
dS(X
a
)dS(X
b
).
On the other hand
H(x, dS) =
1
2

ab
dS(X
a
)dS(X
b
)
gives
_

ab
dS(X
a
)dS(X
b
) = 1, since S is a solution of the Hamilton-Jacobi
PDE.
3 Lorentz-Vranceanu geometry
associated to a black hole system
3.1 Nonholonomic black hole systems
A system whose state depends on the path taken to achieve it is called
nonholonomic. Such a system cannot move in arbitrary directions in its
conguration space. In most cases, it is described by a set of parameters
subject to kinematic Pfa constraints (Pfa equations). Consequently when
8
the system evolves continuously along a path in its parameter space, although
at each conguration the instantaneous velocity (Aristotelian motion) of the
system is restricted to a subspace of the tangent space, it is still possible
to reach any conguration. Alternatively, a nonholonomic system is one
in which any two points in the state space can be connected by a path
(Caratheodory Theorem).
Classicaly, a black hole is an open thermodynamical system since matter
may ow in and out of the system boundary. From a macroscopical view-
point, the Gibbs-Pfa equation of a black hole (reecting the rst law of
thermodynamics) is
dM = TdS + dJ +b
a
dQ
a
,
where T =
k
H
2
is the temperature of the black hole (here k
H
is the surface
gravity on the horizon H), is the angular velocity of the horizon, M is
the mass of the black hole, J is the angular momentum, S =
A
H
4
is the en-
tropy (A
H
is the area of the horizon), the parameters b
a
are the Lagrangian
multipliers for (possible) gauge charges Q
a
. The parameters T and are
assumed in this framework to be external parameters of the macroscopical
thermodynamical theory, which have to be obtained by means of physical
considerations on the geometry and the dynamics of the black hole. Partic-
ularly,
dM =
k
8
dA + dJ + dQ,
where is the electrostatic potential and Q is the electric charge.
In this paper, accepting the Caratheodorys point of view, we dene a
black hole nonholonomic thermodynamic system as the geometrical distribu-
tion
(R
5
, D), R
5
= {(M, T, S, , J)}, D : (M, T, S, , J) = dMTdSdJ = 0,
where the Gibbs-Pfa equation is not completely integrable ( is a contact
form). Since the equation = 0 is not completely integrable, the integral
submanifolds of the distribution D are either curves or surfaces in R
5
=
{(M, T, S, , J)}. A closed curve of evolution in the hyperplane M = 0, i.e.,
in the variables (T, S, , J) gives rise to a spiral in R
5
= {(M, T, S, , J)}.
Denition (1) An integral surface of the Gibbs-Pfa equation = 0 is
called simple black hole.
9
(2) The nonholonomic hypersurface (R
5
, = 0) is called black hole non-
holonomic thermodynamic system.
From local point of view the states of a simple black hole can be chosen
as being two of the ve coordinates M, T, S, , J. In this sense there will
appear 10 types of naturally simple black holes, the basic cases being those
in which the two selected parameters are measurable. As example, a general
integral surface is
M = f(S, J), T =
f
S
, =
f
J
,
where f is an arbitrary C
2
function. The most convenient solution is to x
the function f via the extremals of the least squares Lagrangian
2L = (M f(S, J))
2
+
_
T
f
S
_
2
+
_

f
J
_
2
.
Fixing an extremal by appropriate boundary conditions, we obtain a least
squares approximation for the general integral surface.
It is impossible to observe a black hole directly and so any black hole
candidate have to be identied by its eect on the surrounding matter. Our
foregoing statements spotlight that when we look from exterior to a black
hole we detect only a bundle, i.e., either a texture of bidimensional sheets
or a texture of curves or ejecting jets. For related mathematical theory,
reecting this point of view, see [4], [5], [7]-[12], [13]-[20].
Accretion onto a black hole will look similar in many respects to accre-
tion onto a neutron star or white dwarf. The energy extraction mechanism
involves macroscopic magnetic elds that are produced by currents owing
in a disk or torus surrounding a black hole with electric charge.
3.2 Vranceanu congruences method
Basic ideas of Lorentzian geometry such as Lorentzian metric, covariant dif-
ferentiation, geodesics and curvature belong to the core of mathematical
knowledge and are widely used in applications that range from physics to
mechanics and engineering.
From geometrical point of view, we prefer to use the Vranceanu congru-
ences method [29] on a 5-dimensional Minkowski space-time [1] to study the
10
geometry associated to the nonholonomic Gibbs contact structure
R
5
= {(M, T, S, , J)} , D : = dM TdS dJ,
following similar ideas to those in [4], [7], [10]-[12], [14]-[20]. We preserve the
names mass M, temperature T, entropy S, angular velocity , angular mo-
mentum J for the independent variables, but neither is restricted to positive
values as in thermodynamics of black holes. The result will be a Lorentz-
Vr anceanu geometry associated to a black hole nonholonomic thermodynamic
system.
To prevent contradictory discussions, we accept the dimensionless expres-
sion of the used 1-forms. As example, the dimensionless form of requires
the dimensionless ratio

0
J
0
T
0
S
0
= = 1.
The techniques from dierential geometry ask the general notations x
1
=
M, x
2
= T, x
3
= S, x
4
= , x
5
= J. Also Vr anceanu requires to start from
one given 1-form and to attach another four 1-forms such that all together
to be linearly independent. In our case, the starting is the Gibbs contact
1-form

5
= dM TdS dJ,
written as
5
= dx
1
x
2
dx
3
x
4
dx
5
. To that we add four important ther-
modynamic Pfa forms

1
= x
2
dx
3
,
2
= x
4
dx
5

3
= dx
1
x
4
dx
5
x
5
dx
4
,
4
= dx
1
x
2
dx
3
x
3
dx
2
.
The criteria for our selection is the physical meaning, i.e., the foregoing forms
represent: ds
1
= TdS = dQ, elementary heat of the system; ds
2
= dJ =
dW, elementary expansion work of the system; ds
3
= dMdJJd = dH,
elementary enthalpy; ds
4
= dM TdS SdT = dA, elementary Helmholtz
energy; ds
5
= dM TdS dJ, the initial Gibbs-Pfa contact 1-form. Of
course, here (and in most Thermodynamics Books) are used implicitly the
Vr anceanu notations along curves, ds
a
=
a
.
The components (moments)
a
i
, a, i = 1, 5 of the foregoing Pfa forms

a
=
a
i
dx
i
, a, i = 1, 5, determine the lines of the matrix
(
a
i
) =
_
_
_
_
_
_
_
_
0 0 x
2
0 0
0 0 0 0 x
4
1 0 0 x
5
x
4
1 x
3
x
2
0 0
1 0 x
2
0 x
4
_
_
_
_
_
_
_
_
,
11
This matrix is nondegenerate on the set
A : det (
a
i
) = x
2
x
3
x
4
x
5
= 0, A R
5
,
and its inverse is
_

i
a
_
=
_
_
_
_
_
_
_
_
1 1 0 0 1
0
1
x
3
0
1
x
3
1
x
3
1
x
2
0 0 0 0
1
x
5
0
1
x
5
0
1
x
5
0
1
x
4
0 0 0
_
_
_
_
_
_
_
_
.
Consequently, on the subset A R
5
, the Pfa forms
a
=
a
i
dx
i
are linearly
independent, i.e., they dene a system of congruences (moving co-frame),
with the moments
a
i
. These congruences are not orthogonal in the Euclidean
space (R
5
,
ab
). On the other hand, on the subset A R
5
, the columns
of the inverse matrix (
i
a
) determine the linearly independent vector elds
X
a
=
i
a

x
i
, a = 1, 5 (moving frame). Explicitly,
X
1
=

x
1
+
1
x
2

x
3
+
1
x
5

x
4
, X
2
=

x
1

1
x
3

x
2

1
x
4

x
5
X
3
=
1
x
5

x
4
, X
4
=
1
x
3

x
2
, X
5
=
1
x
3

x
2
+
1
x
5

x
4
.
To be clear, we underline that the components of the previous matrices are
related by

a
i

i
b
=
a
b
,
i
a

a
j
=
i
j
, a, b, i, j = 1, 5
and hence

a
=
a
i
dx
i
, dx
i
=
i
a

a
, i, a = 1, 5.
3.3 Lorentz-Vranceanu metric
The manifold R
5
= R
4
R together with the metric of components

11
=
22
=
33
=
44
= 1,
55
= 1
is a 5-dimensional Minkowski space-time (R
5
, ). Given the system of in-
dependent congruences
a
, a = 1, 5, on (A R
5
,
ab
), we associate to this
12
system only one Lorentzian manifold [1] (A R
5
, g
ij
,
a
) (with a Lorentz-
Vr anceanu metric and a moving co-frame [29]), in which the system of con-
gruences is orthonormal. The Lorentz-Vranceanu metric g
ij
is the tensor
eld

ab
ds
a
ds
b
= g
ij
dx
i
dx
j
, a, b = 1, 5; i, j = 1, 5
of components
(g
ij
) = (
ab

a
i

b
j
) =
_
_
_
_
_
_
_
_
_
1 x
3
0 x
5
0
x
3
x
3
2
x
2
x
3
0 0
0 x
2
x
3
x
2
2
0 x
2
x
4
x
5
0 0 x
5
2
x
4
x
5
0 0 x
2
x
4
x
4
x
5
x
4
2
_
_
_
_
_
_
_
_
_
.
The components (signomials) g
ij
are called thermodynamic potentials.
The contravariant components g
ij
=
ab

i
a

j
b
of the Lorentz-Vr anceanu metric
(signomials) are given by the matrix
(g
ij
) =
_
_
_
_
_
_
_
_
1 0
1
x
2
0
1
x
4
0
1
x
3
2
0
1
x
3
x
5
1
x
3
x
4
1
x
2
0
1
x
2
2
1
x
2
x
5
0
0
1
x
3
x
5
1
x
2
x
5
1
x
5
2
0
1
x
4
1
x
3
x
4
0 0
1
x
4
2
_
_
_
_
_
_
_
_
.
Interesting features of the Lorentz-Vr anceanu metric are the independence
of the mass x
1
= M and the set of singularities for x
2
x
3
x
4
x
5
= 0. These
singularities represent points of innite curvature of the space.
The foregoing vector elds X
a
, a = 1, 4 are spacelike and the vector eld
X
5
is timelike; all together are orthonormal vector elds, with respect to
Lorentz-Vr anceanu metric. Also, these vector elds are complete since their
eld lines are straight lines.
3.4 Geodesics of Lorentz-Vranceanu metric
In Dierential Geometry, geodesics are autoparallel curves and are parame-
terized with constant velocity. Let us nd the geodesics of the Lorentz-
Vr anceanu manifold (A R
5
, g
ij
). For that we need the Levi-Civita connec-
tion
i
jk
produced by the Lorentz-Vr anceanu metric g
ij
.
13
Theorem The non-zero components of the Levi-Civita connection asso-
ciated to the metric g
ij
are the signomials

1
12
=
1
2x
2
,
1
14
=
1
2x
4
,
1
25
=
x
4
2x
2
,
1
34
=
x
2
2x
4
,

2
13
=
1
2x
3
2
,
2
14
=
1
2x
3
x
4
,
2
15
=
1
2x
3
x
5
,

2
23
=
1
2x
3
,
2
34
=
x
2
x
3
x
4
,
2
35
=
1
2
x
2
x
3
x
4
x
5
x
3
2
x
5
,

3
12
=
1
2x
2
2
,
3
15
=
1
2x
2
x
5
,
3
23
=
1
2x
2
,
3
25
=
x
4
2x
2
2
,
3
35
=
1
2x
5
,

4
12
=
1
2x
2
x
5
,
4
13
=
1
2x
3
x
5
,
4
15
=
1
2x
5
2
,
4
25
=
x
4
2x
2
x
5
,

4
35
=
1
2
x
2
x
3
x
4
x
5
x
3
x
5
2
,
4
45
=
1
2x
5
,

5
13
=
1
2x
3
x
4
,
5
14
=
1
2x
4
2
,
5
34
=
x
2
2x
4
2
,
5
35
=
1
2x
3
,
5
45
=
1
2x
4
.
The autoparallel curves of the Levi-Civita connection are geodesics. It is
well known that if is a geodesic, then there exists a real constant E()
such that for any t (t
0
, t
1
) we have E() = g( , ). A geodesic is called
timelike, lightlike or spacelike according as E() is negative, null or positive.
A geodesic is called causal if E() 0.
The foregoing names come from the general relativity where we have the
following interpretations: (1) a timelike geodesic represents the trajectory of
a free falling particle; null geodesics represent the light rays, while (3) space-
like geodesics, for a suitable local observer, represent Riemannian geodesics
consisting of simultaneous events.
The geodesics of the Lorentz-Vranceanu metric, associated to a black hole
nonholonomic thermodynamic system, are given in the next
Theorem. The ODEs of geodesics of the Lorentz-Vranceanu manifold
(A R
5
, g
ij
) are
d
2
x
1
dt
2
+
1
x
2
dx
1
dt
dx
2
dt
+
1
x
4
dx
1
dt
dx
4
dt

x
4
x
2
dx
2
dt
dx
5
dt

x
2
x
4
dx
3
dt
dx
4
dt
= 0,
14
d
2
x
2
dt
2

1
x
3
2
dx
1
dt
dx
3
dt
+
1
x
3
x
4
dx
1
dt
dx
4
dt
+
1
x
3
x
5
dx
1
dt
dx
5
dt
+
1
x
3
dx
2
dt
dx
3
dt

x
2
x
3
x
4
dx
3
dt
dx
4
dt

x
2
x
3
x
4
x
5
x
3
2
x
5
dx
3
dt
dx
5
dt
= 0,
d
2
x
3
dt
2
+
1
x
2
2
dx
1
dt
dx
2
dt

1
x
2
x
5
dx
1
dt
dx
5
dt
+
+
1
x
2
dx
2
dt
dx
3
dt

x
4
x
2
2
dx
2
dt
dx
5
dt
+
1
x
5
dx
3
dt
dx
5
dt
= 0,
d
2
x
4
dt
2
+
1
x
2
x
5
dx
1
dt
dx
2
dt
+
1
x
3
x
5
dx
1
dt
dx
3
dt

1
x
5
2
dx
1
dt
dx
5
dt

x
4
x
2
x
5
dx
2
dt
dx
5
dt
+
x
2
x
3
x
4
x
5
x
3
x
5
2
dx
3
dt
dx
5
dt
+
1
x
5
dx
4
dt
dx
5
dt
= 0,
d
2
x
5
dt
2

1
x
3
x
4
dx
1
dt
dx
3
dt
+
1
x
4
2
dx
1
dt
dx
4
dt

x
2
x
4
2
dx
3
dt
dx
4
dt
+
1
x
3
dx
3
dt
dx
5
dt
+
1
x
4
dx
4
dt
dx
5
dt
= 0.
The results in Section 5 shows that the nonspacelike geodesics of the
Lorentz-Vr anceanu manifold (A R
5
, g
ij
) are transversal to the distribution
D : dx
1
x
2
dx
3
x
4
dx
5
= 0.
4 Stress-energy-momentum tensor on
Lorentz-Vranceanu manifold
In the spacetime theory, the stress-energy-momentum tensor describes the
density and ux of energy and momentum. It is an attribute of matter and
non-gravitational force elds. This tensor is the source of the gravitational
eld in the Einstein eld equations of general relativity [1].
Similarly, let us consider that the Einstein eld equations on the Lorentz-
Vr anceanu manifold (A R
5
, g
ij
) are of the form
R
ij

1
2
Rg
ij
=
8G
c
4
T
ij
, i, j = 1, 5,
15
where R
ij
is the Ricci tensor, R is the Ricci scalar, and G is the universal
gravitational constant.
Theorem The nonzero components of the stress-energy-momentum ten-
sor on the Lorentz-Vranceanu manifold (A R
5
+
, g
ij
) are proportional to
t
11
=
g
2
11
2g
23
g
45
, t
12
=
g
2
11
2
g
2
45
2g
23
g
45
+g
2
23
g
12
g
33
g
2
45
, t
13
=
g
2
11
2
g
45
g
23
g
2
45
g
12
t
14
=
g
2
11
2
g
2
45
2g
23
g
45
+g
2
23
g
14
g
55
g
2
23
, t
15
=
g
2
11
2
g
45
g
23
g
2
23
g
14
t
22
=

g
11
2
g
12
(3g
45
+g
23
)

g
33
g
2
45
, t
23
=
g
2
11
2
3g
45
+g
23
g
2
45
, t
25
=
g
11

g
33
g
44
,
t
33
=
3
_
g
3
11
2

g
33
g
12
g
45
, t
34
=

_
g
3
11
g
12

g
55
, t
35
=
g
3
11
2
5g
23
g
45
+g
2
23
+g
2
45
g
2
12
g
2
14
g
45
t
44
=

g
11
2
g
14
(3g
23
g
45
)

g
55
g
2
23
, t
45
=
g
11
2
3g
23
g
45
g
2
23
, t
55
=
3
_
g
3
11
2

g
55
g
14
g
23
.
Proof. The components of the Einstein tensor E
ij
= R
ij

1
2
Rg
ij
, on the
Lorentz-Vr anceanu manifold (A R
5
+
, g
ij
), are
E
11
=
1
2x
2
x
3
x
4
x
5
, E
12
=
1
2
x
4
2
x
5
2
2x
2
x
3
x
4
x
5
+x
2
2
x
3
2
x
3
x
2
2
x
4
2
x
5
2
E
13
=
1
2
x
2
x
3
x
4
x
5
x
3
x
4
2
x
5
2
, E
14
=
1
2
x
4
2
x
5
2
2x
2
x
3
x
4
x
5
+x
2
2
x
3
2
x
5
x
2
2
x
3
2
x
4
2
E
15
=
1
2
x
2
x
3
x
4
x
5
x
5
x
2
2
x
3
2
, E
22
=
1
2
x
3
(x
2
x
3
3x
4
x
5
)
x
2
x
4
2
x
5
2
E
23
=
1
2
x
2
x
3
3x
4
x
5
x
4
2
x
5
2
, E
25
=
1
x
2
x
5
, E
33
=
3
2
x
2
x
3
x
4
x
5
, E
34
=
1
x
3
x
4
E
35
=
1
2
x
4
2
x
5
2
5x
2
x
3
x
4
x
5
+x
2
2
x
3
2
x
2
x
4
x
3
2
x
5
2
, E
44
=
1
2
x
5
(3x
2
x
3
x
4
x
5
)
x
4
x
2
2
x
3
2
E
45
=
1
2
3x
2
x
3
x
4
x
5
x
2
2
x
3
2
, E
55
=
3
2
x
4
x
2
x
3
x
5
.
16
Expressing these components as functions of g
ij
via
g
11
= 1, g
12
= x
3
, g
14
= x
5
, g
22
= x
3
2
, g
23
= x
2
x
3
g
33
= x
2
2
, g
35
= x
2
x
4
, g
44
= x
5
2
, g
45
= x
4
x
5
, g
55
= x
4
2
,
we obtain the formulas from the Theorem. Of course, the component g
11
is used only to create homogeneity of degree zero. Also, the solution from
the Theorem is not unique. In fact, the nonzero components of the Lorentz-
Vr anceanu metric are connected by the relations
g
11
= 1, g
2
12
= g
22
, g
2
14
= g
44
, g
2
23
= g
22
g
33
, g
2
35
= g
33
g
55
, g
2
45
= g
44
g
55
.
Remark When we speak about the Einstein PDEs, the simplest case
possible is those of empty space, which means that T
ij
vanishes. In this case
R will also vanish such that the Einstein eld equations reduce to R
ij
= 0.
The signomials solutions of this kind of Einstein eld equations are known
from the time of L. P. Eisenhart (1957).
5 Geodesics on nonholonomic
black hole system
The nonholonomic geodesics can be described as in Subsection 2.3. However,
we prefer to use the variational techniques.
5.1 Variational problem of geodesics
The nonholonomic black hole thermodynamic system is the distribution
D : dx
1
x
2
dx
3
x
4
dx
5
= 0
(kernel of the Gibbs contact form) of the Lorentz-Vr anceanu manifold (A
R
5
, g
ij
).
Lemma Any C
1
curve in the distribution D is a spacelike curve.
Proof The restriction of the energy g
ij
(x(t)) x
i
(t) x
j
(t) to the distribution
D : x
1
(t) x
2
(t) x
3
(t) x
4
(t) x
5
(t) = 0 is positive.
17
Consequently, a spacelike geodesic on the nonholonomic system D is a
solution of the problem
min
x()
J(x()) =
1
2
_
t
1
t
0
g
ij
(x(t)) x
i
(t) x
j
(t)dt
subject to
x
1
(t) x
2
(t) x
3
(t) x
4
(t) x
5
(t) = 0, x(t
0
) = x
0
, x(t
1
) = x
1
.
Solution To solve this problem we apply the Lagrange multiplier rule [7].
Also, we describe our distribution as
i
(x(t)) x
i
(t) = 0, where

1
(x) = 1,
2
(x) = 0,
3
(x) = x
2
,
4
(x) = 0,
5
(x) = x
4
dene the timelike covector = (
i
), i.e., g
ij

j
= 1. Introducing the
multiplier p(t), we built a new Lagrangian
L(x, x, p) =
1
2
g
ij
(x(t)) x
i
(t) x
j
(t) +p(t)
i
(x(t)) x
i
(t),
changing the constrained optimization problem into a free one
min
x(),p()
1
2
_
t
1
t
0
L(x(t), x(t), p(t))dt.
Theorem. If the pair (x

(t), p

(t)) is a solution of the last problem, then

it is a solution of the ODE system
L
x
i

d
dt
L
x
i
= 0,
i
(x(t)) x
i
(t) = 0, i = 1, 5.
Computing
L
x
i
=
1
2
g
jk
x
i
(x(t)) x
j
(t) x
k
(t) +p(t)

j
x
i
(x(t)) x
j
(t)
L
x
i
= g
ik
(x(t)) x
k
(t) +p(t)
i
(x(t)),
the Euler-Lagrange ODE become
1
2
g
jk
x
i
(x(t)) x
j
(t) x
k
(t) +p(t)

j
x
i
(x(t)) x
j
(t)
18
g
ik
(x(t)) x
k
(t)
g
ik
x
j
(x(t)) x
k
(t) x
j
(t) p(t)
i
(x(t)) p(t)

i
x
j
(x(t)) x
j
(t) = 0
or
1
2
g
jk
x
i
(x(t)) x
j
(t) x
k
(t) +p(t)

j
x
i
(x(t)) x
j
(t)
g
ik
(x(t)) x
k
(t)
g
ik
x
j
(x(t)) x
k
(t) x
j
(t) p(t)
i
(x(t))p(t)

i
x
j
(x(t)) x
j
(t) = 0.
This can be written
g
ik
(x(t)) x
k
(t)
1
2
_
g
ik
x
j
(x(t)) +
g
ij
x
k
(x(t))
g
jk
x
i
(x(t))
_
x
j
(t) x
k
(t)
+p(t)

j
x
i
(x(t)) x
j
(t) p(t)
i
(x(t)) p(t)

i
x
j
(x(t)) x
j
(t) = 0.
Introducing the Levi-Civita connection
k
ij
, the covariant derivative
i;j
, we
can write
x
k
(t) +
_

k
ij
+
1
2
(
i;j
+
j;i
)
k
_
x
i
(t) x
j
(t)
= g
ki
_
p(t)
_

j
x
i
(x(t))

i
x
j
(x(t))
_
x
j
(t) p(t)
i
(x(t))
_
+
1
2
(
i;j
+
j;i
)
k
x
i
(t) x
j
(t).
The rst term in the second hand expression is orthogonal to x
i
, i.e., it is
collinear to
i
, in the Euclidean space (R
5
,
ij
). Consequently, the right hand
member is collinear to
i
. On the other hand, by derivation with respect to
t, the constraint equation
i
(x(t)) x
i
(t) = 0 implies

k
(x(t))
_
x
k
(t) +
_

k
ij
+
1
2
(
i;j
+
j;i
)
k
_
(x(t)) x
i
(t) x
j
(t)
_
= 0,
i.e., the expression in the paranthesis is orthogonal to the nonzero covector

i
, in the Euclidean space (R
5
,
ij
).
Corollary The ODEs of the nonholonomic spacelike geodesics are
x
k
(t) +
_

k
ij
+
1
2
(
i;j
+
j;i
)
k
_
(x(t)) x
i
(t) x
j
(t) = 0.
Remark. The Lagrange multiplier p is a solution of ODE
p(t)
_

j
x
i
(x(t))

i
x
j
(x(t))
_
x
j
(t) p(t)
i
(x(t))
+
1
2
(
k;j
+
j;k
)
i
(x(t)) x
k
(t) x
j
(t) = 0.
19
5.2 Optimal control problem of geodesics
The co-vector eld normal to the black hole thermodynamic distribution
D : dx
1
x
2
dx
3
x
4
dx
5
= 0, (x
1
, x
2
, x
3
, x
4
, x
5
) A R
5
,
i.e., = (1, 0, x
2
, 0, x
4
) is timelike with respect to the Lorentz-Vranceanu
metric. The foregoing spacelike orthonormal vector elds X
a
, a = 1, 4 are
tangent to this distribution and consequently D = span{X
a
, a = 1, 4}. Their
orthonormality means g
ij
X
i
a
X
j
b
=
ab
, a, b = 1, 4.
The optimal control problem of spacelike geodesics is a non-linear-
quadratic regulator problem,
min
u()
J(u()) =
1
2
_
t
1
t
0

ab
u
a
(t)u
b
(t)dt
subject to the driftless control system
x(t) = u
a
(t)X
a
(x(t)), u(t) U, t [t
0
, t
1
]; x(t
0
) = x
0
, x(t
1
) = x
1
.
Solution We apply the Pontryaguin minimum principle (for multitime min-
imum principle, see [21]-[28]). In general notations, we have
x = (x
i
), u = (u
a
), p = (p
i
), a = 1, 4; i = 1, 5
X(x(t), u(t)) = u
a
(t)X
a
(x(t)), X
0
(x(t), u(t)) =
1
2

ab
u
a
(t)u
b
(t)
and the control Hamiltonian is
H(x, p, u) = p
i
(t)X
i
(x, u) +X
0
(x, u) =
1
2

ab
u
a
u
b
+p
i
u
a
X
i
a
(x).
The adjoint dynamics says
(ADJ)
dp
j
dt
=
H
x
j
= p
i
u
a
X
i
a
x
j
.
We have to maximize the Hamiltonian H(x, p, u) with respect to the control
variable u. The critical points u of H are solutions of the algebraic system
H
u
a
= u
a
+p
i
X
i
a
= 0.
20
Consequently the adjoint ODE system is
(ADJ)
dp
j
dt
(t) =
1
2
p
i
(t)p
k
(t)
_
g
ik
x
j
(x(t)) +

x
j
(X
i
5
(x(t))X
k
5
(x(t)))
_
.
and the initial dynamics becomes
(ODE) x
i
(t) = p
j
(t)
_
g
ij
(x(t)) +X
i
5
(x(t))X
j
5
(x(t))
_
.
Looking for a solution in the form p(t) = K(t)x(t), where K(t) = (K
ij
(t)) is
a quadratic (symmetric) matrix, we nd the feed-back control law (Kalman
vector) u
a
= K
ij
X
i
a
x
j
and ODE can be written
(ODE) x
i
(t) = K
jk
(t)x
k
(t)
_
g
ij
(x(t)) +X
i
5
(x(t))X
j
5
(x(t))
_
.
Theorem (optimal control is unitary [8]) The optimal controls u
a
= p
i
X
i
a
satisfy the ODE system u
a
= p
i
[X
a
, X
b
]
i
u
b
. Consequently, the control u =
(u
a
) is unitary, i.e.,
1
2

ab
u
a
u
b
is a rst integral.
Proof. Direct dierentiation. The unitarity of the controls follows from
the fact that the evolution u
a
of u
a
is described by a linear ODE with skew-
symmetric right hand side.
Remark The Lie brackets [X
a
, X
b
] represent the connection of the hori-
zontal distribution and p
i
[X
a
, X
b
]
i
is the component of the curvature in the
p direction.
Theorem The black hole nonholonomic thermodynamic system is con-
trollable (accessible) by spacelike geodesics.
Proof Since the vector elds X
a
, a = 1, 4 are orthonormal, they are
linearly independent. Also, the black hole distribution is bracket generating.
Applying the Chow Theorem, any two suciently close points can be joined
by a minimizing geodesic, i.e., a solution of the boundary problem
x(t) = u
a
(t)X
a
(x(t)), u(t) U, t [t
0
, t
1
]; x(t
0
) = x
0
, x(t
1
) = x
1
.
5.3 Stress-energy-momentum tensor on
nonholonomic black hole system
Let (A R
5
+
, g
ij
) be the Riemann - Vr anceanu manifold and
k
ij
be the
Levi-Civita connection. The new connection

k
ij
=
k
ij
+
1
2
(
i;j
+
j;i
)
k
,
21
which lies in on the nonholonomic black hole system, determines a curvature
tensor R
i
jkl
and hence the Ricci tensor R

ij
. We dene R

= g
ij
R

ij
. Similar
ideas to those in Section 4 lead to the stress-energy-momentum tensor T
ij
on
the nonholonomic black hole system and a solution of the associated Einstein
PDEs in terms of g
ij
.
6 Two-time optimal control on
black hole system
Let (A R
5
, g
ij
) be the Riemann - Vranceanu manifold. Let
t
0
t
1
be the
rectangle determined by two opposite diagonal points t
0
= (t
1
0
, t
2
0
) and t
1
=
(t
1
1
, t
2
1
) in R
2
+
, endowed with the product order. We consider those integral
2-sheets of the macroeconomic distribution D = span{X
a
, a = 1, 4} which
are solutions of the completely integrable driftless control system
x
t

(t) = u
a

(t)X
a
(x(t)), t = (t

)
t
0
t
1
, a = 1, 4, = 1, 2,
in the sense that the piecewise complete integrability conditions
_
u
a

u
a

_
X
a
= u
a

u
b

[X
a
, X
b
]
are satised.
Theorem The black hole nonholonomic thermodynamic system is con-
trollable (accessible) by harmonic spacelike 2-sheets.
Proof. The kinetic energy
1
2
_

t
0
t
1
g
ij
(x(t))

x
i

(t)x
j

(t)dt
1
dt
2
simplies
due to the orthonormality of the vector elds X
a
. Consequently, the multi-
time optimal control problem of harmonic spacelike 2-sheets is a non-linear-
quadratic regulator problem
min
u()
J(u()) =
1
2
_
t
0
t
1

ab

u
a

(t)u
b

(t)dt
1
dt
2
,
subject to
x
t

= u
a

(t)X
a
(x(t)),
_
u
a

u
a

_
X
a
= u
a

u
b

[X
a
, X
b
]
22
t
t
0
t
1
; x(t
0
) = x
0
, x(t
1
) = x
1
, u = (u
a

), a = 1, 4; , = 1, 2.
The solve such a multitime optimum problem we use the techniques from our
papers [21]-[28]. To simplify, let us show that the foregoing problem can be
solved via controls of gradient type, i.e., controls written as u
a

=
f
a
t

, where
f = (f
a
) :
t
0
t
1
R is a C
2
function. Then the second PDE constraint is
reduced to an algebraic equation and the Hamiltonian is
H =
1
2

ab

u
a

u
b

+p

i
(t)u
a

X
i
a
(x) +q

i
(t)u
a

u
b

[X
a
, X
b
]
i
.
The adjoint dynamics is described by the PDEs
(ADJ)
p

j
t

=
H
x
j
= p

i
u
a

X
i
a
x
j
+q

i
u
a

u
b

x
j
[X
a
, X
b
]
i
.
The critical points u of H are solutions of the algebraic system
H
u
c

= u
a

c
a

+ (q

i
q

i
) [X
a
, X
c
]
i
_
+p

i
X
i
c
= 0.
The matrix of the system is nondegenerate, as a sum between the unit matrix
and a skew-symmetric one. Hence we have a compatible linear system of eight
equations with eight unknowns u
c

. Consequently the adjoint PDE system

and the initial PDE dynamics have optimal solutions. The optimal 2-sheet
depends on 20 arbitrary functions q

i
(t). Consequently we can have an
optimal solution through the points x(t
0
) = x
0
, x(t
1
) = x
1
.
7 Conclusions
A new approach to the Gibbs-Pfa dierential 1-form of black holes, includ-
ing the congruences theory of Lorentz-Vr anceanu combined with the point of
view of Montgomery and with Udristes theory of thermodynamical systems,
has been presented. It appears a new Lorentz-Vranceanu manifold and a new
nonholonomic geometry, with appropriate Einstein equations, based on the
physical meanings of the used co-frame or frame. This point of view allows
to prove the controllability of our black hole systems via single-time or via
bi-temporal controls.
The original results are based on the connection between techniques from
dierential geometry, variational calculus, optimal control, and Maple capa-
bilities. We hope that our point of view will stimulate the researchers to use
23
more intensively these ingredients to study the black hole systems and the
economic black hole systems.
References
[1] J. K. Beem, P. E. Ehrlich, K. L. Easley, Global Lorentzian Geometry,
Marcel Dekker, Inc, 1996.
[2] O. Calin, D.-C. Chang, Sub-Riemannian geometry: general theory and
examples, Encyclopedia of Mathematics and Its Applications, 126, Cam-
bridge University Press, New York, 2009.
[3] A. V. Dmitruk, Quadratic sucient conditions for strong minimality of
abnormal sub-Riemannian geodesics, Russian Journal of Mathematical
Physics, Vol. 6, No. 3, 1999, pp. 363 372.
[4] M. Ferrara, C. Udriste, Area conditions associated to thermodynamic
and economic systems, BSG Proceedings 8, 60-68, Geometry Balkan
Press (2002), Bucharest, Romania.
[5] V. Frolov, I. Novikov, Black Hole Physics: Basic Concepts and New
Developments, Kluwer Academic Publishers, 1989.
[6] R. Montgomery, A Tour of Subriemannian Geometries, Their Geodesics
and Applications, Mathematical Surveys and Monographs, 91, American
Mathematical Society, 2002.
[7] V. Radcenco, C. Udriste, D. Udriste, Thermodynamic systems and their
interaction, 22-nd Conference on Dierential Geometry and Topology,
Applications in Physics and Technics, Sept. 9-13 (1991), Bucharest; Sci-
entic Bulletin, Polytechnic Institute of Bucharest, Electrical Engineer-
ing, 53, 3-4 (1991), 285-294.
[8] S. Shankar Sastry, R. Montgomery, The structure of optimal controls for
steering problem, NOLCOS, 1992, Conf. Proc., Bordeaux, France.
[9] J. Shen, R. Cai, B. Wang, R. Su, Thermodynamic geometry and critical
behavior of black holes, (arXiv: gr-qc/0512035), 2005.
24
[10] C. Stamin, C. Udriste, The geometry of Gibbs-Duhem-Pfa thermochem-
ical system, Proceedings of the 7th WSEAS International Conference
on Instrumentation, Measurement, Circuits and Systems (IMCAS08),
Electrical and Computer Engineering Series, pp. 77-89, Hangzhou,
China, April 6-8, 2008.
[11] C. Stamin, C. Udriste, Subriemannian geometry of open thermochemical
system, Proceedings of the 2nd WSEAS International Conference on
Multivariate Analysis and Its Application in Science and Engineering
(MAASE09), pp. 60-67, Istanbul, Turkey, May 30-June 1, 2009.
[12] C. Stamin, C. Udriste, Nonholonomic geometry of Gibbs contact struc-
ture, U.P.B. Sci. Bull., Series A, 72, 1 (2010), 153-170.
[13] Kang-Hai Tan, Convex functions on sub-Riemannian manifolds, arXiv:
math/0701273v1, [math.DG], 10 January 2007.
[14] C. Udriste, Geometric Dynamics, Kluwer Academic Publishers, 2000.
[15] C. Udriste, O. Dogaru, I. T evy, Extrema with Nonholonomic Con-
straints, Geometry Balkan Press, Bucharest, 2002.
[16] C. Udriste, M. Ferrara, D. Opris, Economic Geometric Dynamics, Mono-
graphs and Textbooks 6, Geometry Balkan Press, 2004.
[17] C. Udriste, V. Badescu, V. Ciancio, F. Ghionea, D. Izvoranu, I. T evy,
Black hole geometric thermodynamics, BSG Proceedings 14, 186-194,
Geometry Balkan Press (2007), Bucharest, Romania.
[18] C. Udriste, Thermodynamics versus Economics, U.P.B. Sci. Bull., Series
A, 69, 3, 2007, 89-91.
[19] C. Udriste, D. Izvoranu, V. Ciancio, F. Ghionea, V. Badescu, I. T evy,
Higher-dimensional black hole geometric thermodynamics, BSG Pro-
ceedings 15, 221-231, Geometry Balkan Press (2008), Bucharest, Ro-
mania.
[20] C. Udriste, M. Ferrara, Black hole models in economics, Tensor, N.S.,
vol. 70, 1, 53-62, (2008).
25
[21] C. Udriste, Multitime controllability, observability and bang-bang prin-
ciple, Journal of Optimization Theory and Applications, 139, 1, 141-157
(2008).
[22] Udriste, C.: Simplied multitime maximum principle. Balkan Journal
of Geometry and Its Applications, 14, 1, 102-119 (2009)
[23] Udriste, C.: Nonholonomic approach of multitime maximum principle.
Balkan Journal of Geometry and Its Applications, 14, 2, 111-126 (2009)
[24] Udriste, C., T evy, I.: Multitime linear-quadratic regulator problem
based on curvilinear integral. Balkan Journal of Geometry and Its Ap-
plications, 14, 2, 127-137 (2009)
[25] Udriste, C., T evy, I.: Multitime dynamic programming for curvilinear
integral actions. Journal of Optimization Theory and Applications, 146,
189-207 (2010)
[26] Udriste, C.: Equivalence of multitime optimal control problems. Balkan
Journal of Geometry and Its Applications, 15, 1, 155-162 (2010)
[27] Udriste, C.: Multitime maximum principle for curvilinear integral cost.
Balkan Journal of Geometry and Its Applications, 16, 1, 128-149 (2011)
[28] Udriste, C., Bejenaru, A., Multitime optimal control with area integral
costs on boundary. Balkan Journal of Geometry and Its Applications,
16, 2, 138-154 (2011)
[29] G. Vr anceanu, Lectures of Dierential Geometry, Didactical and Peda-
gogical Editorial House, Bucharest, Vol. I, 1962; Vol. II, 1964 (in Roma-
nian).
Prof. Dr. Constantin Udriste, University Politehnica of Bucharest, Fac-
ulty of Applied Sciences, Departament of Mathematics-Informatics I, Splaiul
Independentei 313, 060042 Bucharest, Romania, E-mail: udriste@mathem.pub.ro,
anet.udri@yahoo.com
Prof. Dr. Vincenzo Ciancio, Department of Mathematics, University
of Messina, Contrada Papardo, Salita Sperone, 31, 98166, Messina, Italy,
E-mail: ciancio@unime.it
Francesco Farsaci IPCF-C.N.R., Messina, Italy. E-mail: farsaci@me.cnr.it
26