23.02.2014 Views

Shape

Shape

Shape

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

283 Parts Are Evanescent—They Change as Rules Are Tried<br />

according to the twin formulas for applying rules. The scenarios are good for shapes<br />

with points and lines, etc. In fact, for lines, they’re already familiar from part I. But<br />

my scenarios are just examples. There are other ways to talk about rules and what<br />

they do to divide shapes into parts. I’ll show this below in an easy approach with identities<br />

and transformations. It expands a little on my second scenario to define equivalence<br />

relations that classify parts. Decompositions and other kinds of taxonomies are<br />

an ideal way for reasoning to conclude. There’s a sense of accomplishment, and a real<br />

chance to understand what happened. Parts are charged with meaning. My scenarios<br />

provide the means to get them, and to change them whenever I go on.<br />

In the first scenario, only rules of a special kind are used to calculate. Every rule<br />

A fi<br />

erases a shape A—or some transformation of A—by replacing it with the empty shape.<br />

So, when I calculate, it looks like this<br />

C Þ C<br />

tðAÞ Þ Þ r<br />

I start with any shape C. Another part of C—in particular, the shape tðAÞ—is subtracted<br />

in each step, when a rule A fi is applied under a transformation t. The shape<br />

r—the remainder—is left at the end of this process. Ideally, r is the empty shape. But<br />

whether or not this happens, a Boolean algebra is defined for C. The remainder r when<br />

it isn’t empty and the parts tðAÞ, picked out as rules are tried, are the atoms of the algebra.<br />

This is an easy way for me to define a vocabulary, or to see just how well a given<br />

one works in a particular case.<br />

In the next scenario, rules are always identities. Every identity<br />

A fi A<br />

has the same shape A in both its left and right sides. When only identities are applied,<br />

calculating is monotonous<br />

C Þ C Þ Þ C<br />

where, again, C is a given shape. In each step, another part of C is resolved in accordance<br />

with an identity A fi A, and nothing else is done. Everything stays exactly the<br />

same because tðAÞ is part of C. Identities are constructively useless. In fact, it’s a common<br />

practice to discard them. But this misses their value as observational devices. The<br />

parts tðAÞ that are resolved as identities are tried, the empty shape, and C can be used<br />

to define a topology for C when they’re all combined in sums and products. The first<br />

scenario is included in this one if the remainder r is empty, or if r is included with the<br />

parts tðAÞ. It’s easy to do—for each application of an erasing rule A fi , simply use<br />

the identity A fi A. But a Boolean algebra isn’t necessary when identities are applied.<br />

Complements aren’t defined automatically. They may have to be specified explicitly.<br />

In the final scenario, rules aren’t restricted. Each rule<br />

A fi B

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!