distances between any two close enough points and their
images is bounded from below and from above. The maximum local
relative length change is called quasi-isometry constant.
When distance is measured using given metric, the construction
of QI mapping can be considered as a maximum norm approximation
by mapping with prescribed metric properties.
It is possible to construct QI mappings using variational
principle [1]. It is well posed [2] and allows to
minimize QI constant, while retaining mapping smoothness.
It allows to fit n-dimensional mapping to given metric tensor.
Applications:
1. Optimal parameterization of surfaces (surface meshing,
texture mapping with minimal distortion). This approach
is based on flattening of surface with given parameterization,
construction of planar parameterization and backward
mapping [3,4]. In QI approach the presence of distortion bounds
is provable. Results do not depend on initial surface
parameterization [5], i.e. ``mesh independence'' is observed.
2. Morphing problems. Construction of deformation path between two
objects which provides one-to-one texture mapping along deformation
path. Simplest approach is based on linear path and may fail easily.
Approach from [6] is based on interpolation of coefficients of matrices.
It is robust for convex objects. In QI approach the metric tensor of
deformation is interpolated. Intermediate shapes are max-norm fitted
to given metric data. No constraints on domain shapes are imposed.
Method is applicable in n-dimensional case.
3. Morphing via mesh untangling. Given two surfaces and deformation path,
then some transient shapes can be folded and/or local invertibility of
deformation is lost. This problem can be solved via variational untangling
of spatial mesh [7,8]. As a result all transient shapes are
provably good [8]. Can be applied in n-dimensions.
[1] V.A.Garanzha.
"Barrier variational generation of quasi-isometric grids",
Comp. Math. and Math. Phys. Vol. 40, No. 11, 2000, pp. 1617-1637.
[2] V.A.Garanzha, N.L.Zamarashkin.
"Quasi-isometric mappings as minimizers of polyconvex functionals",
In: Grid generation: theory and applications, eds. S.A.Ivanenko,
V.A.Garanzha, 2002, pp. 150-168 (in Russian).
[3] M.S.Floater.
"Parametrization and smooth approximation of surface triangulations",
Comp. Aided Geom. Design, 1997, Vol 14, pp. 231-250.
[4] K.Hormann and G.Greiner.
"MIPS: an efficient global parameterization method",
In: Laurent, Sablonniere, Schumacher (Eds). Curve and surface design.
Vanderbilt Univ. Press, Nashville, 2000, pp. 163-172.
[5] V.A.Garanzha
"Max-norm optimization of spatial mappings with application
to grid generation, construction of surfaces and shape design"
In: Grid generation: new trends and applications in real-world
simulations, eds. S.A.Ivanenko, V.A.Garanzha,
Comm. on Appl. Math., Computer Centre, Russian Academy
of Sciences, Moscow, 2001.
[6] M.S.Floater and C.Gotsman.
"How to Morph Tilings Injectively",
J. Comp. Appl. Math. 1999, Vol101, pp. 117-129.
[7] V.A.Garanzha and I.E.Kaporin.
"Regularization of the barrier variational grid generation method",
Comp. Math. and Math. Phys., 1999, No, 9. pp. 1489-1503.
[8] L.V.Branets and V.A.Garanzha.
"Global condition number of trilinear mapping.
Application to 3-D grid generation",
In: Grid Generation: New Trends and Applications in Real-Life
Simulations, eds. S.A.Ivanenko and V.A.Garanzha, Comm. on Applied
Math., Computing Center Russian Acad. Sci., 2001, pp. 45--60.