
Who was Ulugh Beg? A prince who governed a province in the central Asian
empire built by his grandfather Tamerlane. Above all, he was a scholar who
founded the Samarkand astronomical observatory, whose work predated that of the
best astronomers in Europe one and a half centuries later.

From the geocentric, closed world model of Antiquity to the wraparound
universe models of relativistic cosmology, the parallel history of space
representations in science and art illustrates the fundamental role of
geometric imagination in innovative findings. Through the analysis of works of
various artists and scientists like Plato, Durer, Kepler, Escher, Grisey or the
present author, it is shown how the process of creation in science and in the
arts rests on aesthetical principles such as symmetry, regular polyhedra, laws
of harmonic proportion, tessellations, group theory, etc., as well as beauty,
conciseness and emotional approach of the world.

The twin paradox is the best known thought experiment associated with
Einstein's theory of relativity. An astronaut who makes a journey into space in
a highspeed rocket will return home to find he has aged less than a twin who
stayed on Earth. This result appears puzzling, since the situation seems
symmetrical, as the homebody twin can be considered to have done the travelling
with respect to the traveller. Hence it is called a "paradox". In fact, there
is no contradiction and the apparent paradox has a simple resolution in Special
Relativity with infinite flat space. In General Relativity (dealing with
gravitational fields and curved spacetime), or in a compact space such as the
hypersphere or a multiply connected finite space, the paradox is more
complicated, but its resolution provides new insights about the structure of
spacetime and the limitations of the equivalence between inertial reference
frames.

Aims: We investigate the stellar pancake mechanism during which a solartype
star is tidally flattened within its orbital plane passing close to a 10^6
solar masses black hole. We simulate the relativistic orthogonal compression
process and follow the associated shock waves formation.
Methods: We consider a onedimensional hydrodynamical stellar model moving in
the relativistic gravitational field of a nonrotating black hole. The model is
numerically solved using a Godunovtype shockcapturing sourcesplitting method
in order to correctly reproduce the shock waves profiles.
Results: Simulations confirm that the spacetime curvature can induce several
successive orthogonal compressions of the star which give rise to several
strong shock waves. The shock waves finally escape from the star and repeatedly
heat up the stellar surface to high energy values. Such a shockheating could
interestingly provide a direct observational signature of strongly disruptive
star  black hole encounters through the emission of hard X or soft gammaray
bursts. Timescales and energies of such a process are consistent with some
observed events such as GRB 970815.

The full threeyear WMAP results (WMAP3) reinforce the absence of largeangle
correlations at scales greater than 60 degrees. The Poincare dodecahedral space
(PDS) model model, which may naturally explain such features, thus remains a
plausible cosmological model, despite recent controversy about whether matched
circle searches would or would not push the topology beyond the horizon. Here,
we have used new eigenmode calculations of the dodecahedral space to predict
the cosmic microwave background (CMB) temperature fluctuations in such models,
with an improved angular resolution. We have simulated CMB maps and confirmed
the expected presence of matching circles. For a set of plausible cosmological
parameters, we have derived the angular power spectrum of the CMB up to large
wavenumbers. Comparison with the WMAP3 observations confirms a remarkable fit
with a PDS model, for a value $\Omega_0 = 1.018$ of the average total energy
density.

Cosmology's standard model posits an infinite flat universe forever expanding
under the pressure of dark energy. Firstyear data from the Wilkinson Microwave
Anisotropy Probe (WMAP) confirm this model to spectacular precision on all but
the largest scales (Bennett {\it et al.}, 2003 ; Spergel {\it et al.}, 2003).
Temperature correlations across the microwave sky match expectations on scales
narrower than $60^{\circ}$, yet vanish on scales wider than $60^{\circ}$.
Researchers are now seeking an explanation of the missing wideangle
correlations (Contaldi {\it et al.}, 2003 ; Cline {\it et al.}, 2003). One
natural approach questions the underlying geometry of space, namely its
curvature (Efstathiou, 2003) and its topology (Tegmark {\it et al.}, 2003). In
an infinite flat space, waves from the big bang would fill the universe on all
length scales. The observed lack of temperature correlations on scales beyond
$60^{\circ}$ means the broadest waves are missing, perhaps because space itself
is not big enough to support them.
Here we present a simple geometrical model of a finite, positively curved
space  the Poincar\'e dodecahedral space  which accounts for WMAP's
observations with no finetuning required. Circle searching (Cornish, Spergel
and Starkman, 1998) may confirm the model's topological predictions, while
upcoming Planck Surveyor data may confirm its predicted density of $\Omega_0
\simeq 1.013 > 1$. If confirmed, the model will answer the ancient question of
whether space is finite or infinite, while retaining the standard
FriedmannLema\^\i{}tre foundation for local physics.