Algorithms and application of scale-dependence monotonicity

Conference · Sat Dec 31 04:00:00 EST 1994

OSTI ID:35862

Brooks, M

Scale-dependent monotonicity is a generalized notion of monotonicity that {open_quotes}ignores{close_quotes} numerical differences less than a given scale for real valued functions on a finite real domain, i.e. finite real functions. For each scale {delta} > 0 any finite real function`s domain can be partitioned into intervals upon which the function is {delta}-monotone. Certain optimal piecewise monotone approximations can be computed directly from the {delta}-monotone intervals The intervals at a given scale are computed in linear time and constant space, using only comparison and addition. The set of all scales {delta} for which a function has distinct {delta}-monotone intervals can be computed in quadratic time. Given a scale {delta}, suboptimal piecewise linear approximations having a bounded number of pieces can be computed. This talk will review scale-dependent monotonicity algorithms and describe their application to empirical modelling of a physical process.

• OSTI ID: 35862
• Report Number(s): CONF-9408161--
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
ALGORITHMS
COMPUTERIZED SIMULATION
CONVERGENCE
NONLINEAR PROBLEMS
NUMERICAL SOLUTION
STOCHASTIC PROCESSES