Re: Fw: The Mere Creation Discussion

Bill Hamilton (hamilton@predator.cs.gmr.com)
Fri, 10 Jan 1997 14:46:50 -0500

Brian D. Harper wrote:
>>BH>Darwinism involves random mutations + natural selection. Natural
>>>selection is about as opposite from pure chance as one could get.
>>
>Stephen Jones wrote

>>Agreed. But it is the *combination* of "random mutations" with
>>"natural selection" that makes the *whole* process essentially
>>random:

This statement is quite misleading. There are a number of problems in
engineering and the sciences in which the deliberate introduction of
randomness actually helps solve the problem and obtain a better result.
Some examples of this are

1. Nonlinear control systems. Control engineers have for years known that
random noise can help cause a nonlinear control system to settle down more
accurately to a desired target state. One of the most obvious examples is
the case of stiction in a servo system. A small random perturbation can
keep the controller from getting stuck on the stiction and stopping before
it has brought the error to zero.

2. Optimization. In some optimization problems the optimal solution can
only be obtained by using a mixed strategy: selecting between two or more
strategies randomly.

3. Adaptive control. In some adaptive controlers a random noise process
is used to perturb the system in such a way that the resulting variations
can be used to obtain the best control strategy for the current state of
the parameters that are being adapted to.

4. System identification. Random inputs are used to provide a
sufficiently rich set of inputs to ensure that all pertinent system modes
are correctly identified.

In each of the above cases random noise is deliberately injected into a
system or process and the result is an improvement in the accuracy of the
result.

Technically Stephen is correct: the resulting process is random in that it
contains random elements. However, the uncertainty of the final result is
significantly _smaller_ than the uncertainty that could be obtained without
introducing random perturbations.

Bill Hamilton
--------------------------------------------------------------------------
William E. Hamilton, Jr, Ph.D. | Staff Research Engineer
Chassis and Vehicle Systems | General Motors R&D Center | Warren, MI
810 986 1474 (voice) | 810 986 3003 (FAX) | whamilto@mich.com (home email)