|
Organizers |
Ergodic Theory of One-Dimensional Model of Drilling
by
Georgi Chakvetadze
Moscow State Aviation Institute
We consider a family of interval maps serving as a model
for the process of rock destruction by drilling bit.
These maps depend on two parameters - a scalar and a function - and
are defined as follows.
Let p:R --> R be a 1-periodic
function, twice continuously differentiable on [0, 1],
p''(r) < 0,
p(1-r)=p(r), r in R.
We will refer to the graph
of p as basic profile. The basic
profile is determined by geometry of the drilling bit.
Set
m=infr in [0, 1]|p''(r)|, M=supr in [0, 1]|p''(r)|,
a=[ M/m] and assume that 0 < b < 1.
Given s in R the graph of the function
| (1) |
Lemma 1 There are points s1, ..., sn, sn+1,
0=sn+1 < sn < ... < s1 < [ 1/2] < s0=1,
such that the transformation F
a continuously differentiable and strictly decreasing on the set
I1=(s1, s0) from 1-s1 till 0, on the sets
Ik=(sk, sk-1), k=2, ..., n,
from 1-sk till F(sk-1-0)=F(1-sk-1) and on the set
In+1=(sn+1, sn) from
F(0+0) till F(sn-0)=F(1-sn).
Lemma 2 Assume that b > [ a/2]. Then all the phase space
is attracted to the unique fixed point of the transformation F.
Theorem 1 Assume that p'' has a bounded variation on [0, 1]
and the values of the parameters a and b satisfy the
inequality
(1-b)3/b > (a-b)2+[ 1/8](a-1)(1-b).
Then the transformation F has an absolutely continuous
invariant probability (acip).
The next theorem describes the set of absolutely continuous invariant probabilities of F.
Theorem 2 Assume that p'' has a bounded variation on [0, 1].
Then the measure v is ergodic if the inequality
(1-b)3/b > 2(a-b)2
holds.
Theorem 3 Under the assumptions of the theorem 2 the measure v is
weakly mixing.
Finally, we discuss the stability property of the measure v with respect to stochastic perturbations Yc, c > 0, introduced as follows. Given c > 0 and s in [0, 1) we replace p(s) (p'(s+0)) in the equality (1) by the stochastic variable distributed with some density on the interval [p(s)-c, p(s)+c] ([p'(s+0)-c, p'(s+0)+c]). Then F(s) is replaced by the stochastic variable distributed on some set including F(s) with the density uc(s, ·). The Markov chain Yc has uc(s, ·) as the densities of transition probabilities. In the proof of the next theorem additionally some technical conditions are used which we omit.
Theorem 4 Assume that p in C3([0, 1]) and the
transformation F
has no periodic turning points. Then the densities of
stationary distributions of the chains Yc tend
to the density of v in L1-metric as c --> 0.
Date received: August 6, 1999
Copyright © 1999 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cacy-48.