9285
|
1 |
%title%
|
|
2 |
Isabelle's Logics
|
|
3 |
|
|
4 |
%body%
|
|
5 |
<h3>What is Isabelle?</h3>
|
|
6 |
|
|
7 |
Isabelle can be viewed from two main perspectives. On the one hand it
|
|
8 |
may serve as a generic framework for rapid prototyping of deductive
|
|
9 |
systems. On the other hand, major existing logics like
|
|
10 |
<strong>Isabelle/HOL</strong> provide a theorem proving environment
|
|
11 |
ready to use for sizable applications.
|
|
12 |
|
|
13 |
|
|
14 |
<h2>Isabelle's Logics</h2>
|
|
15 |
|
|
16 |
The Isabelle distribution includes a large body of object logics and
|
|
17 |
other examples (see the <a href="library/">Isabelle theory
|
|
18 |
library</a>).
|
|
19 |
|
|
20 |
<dl>
|
|
21 |
|
|
22 |
<dt><a href="library/HOL/"><strong>Isabelle/HOL</strong></a><dd> is a
|
|
23 |
version of classical higher-order logic resembling that of the <A
|
|
24 |
HREF="http://www.cl.cam.ac.uk/Research/HVG/HOL/HOL.html">HOL
|
|
25 |
System</A>.
|
|
26 |
|
|
27 |
<dt><a href="library/HOLCF/"><strong>Isabelle/HOLCF</strong></a><dd>
|
|
28 |
adds Scott's Logic for Computable Functions (domain theory) to HOL.
|
|
29 |
|
|
30 |
<dt><a href="library/FOL/"><strong>Isabelle/FOL</strong></a><dd>
|
|
31 |
provides basic classical and intuitionistic first-order logic. It is
|
|
32 |
polymorphic.
|
|
33 |
|
|
34 |
<dt><a href="library/ZF/"><strong>Isabelle/ZF</strong></a><dd> offers
|
|
35 |
a formulation of Zermelo-Fraenkel set theory on top of FOL.
|
|
36 |
|
|
37 |
</dl>
|
|
38 |
|
|
39 |
<p>
|
|
40 |
|
|
41 |
Isabelle/HOL is currently the best developed object logic, including
|
|
42 |
an extensive library of (concrete) mathematics, and various packages
|
|
43 |
for advanced definitional concepts (like (co-)inductive sets and
|
|
44 |
types, well-founded recursion etc.). The distribution also includes
|
|
45 |
some large applications, for example correctness proofs of
|
|
46 |
cryptographic protocols (<a href="library/HOL/Auth/">HOL/Auth</a>) or
|
|
47 |
communication protocols (<a href="library/HOLCF/IOA/">HOLCF/IOA</a>).
|
|
48 |
|
|
49 |
<p>
|
|
50 |
|
|
51 |
Isabelle/ZF provides another starting point for applications, with a
|
|
52 |
slightly less developed library. Its definitional packages are
|
|
53 |
similar to those of Isabelle/HOL. Untyped ZF provides more advanced
|
|
54 |
constructions for sets than simply-typed HOL.
|
|
55 |
|
|
56 |
<p>
|
|
57 |
|
|
58 |
There are a few minor object logics that may serve as further
|
|
59 |
examples: <a href="library/CTT/">CTT</a> is an extensional version of
|
|
60 |
Martin-Löf's Type Theory, <a href="library/Cube/">Cube</a> is
|
|
61 |
Barendregt's Lambda Cube. There are also some sequent calculus
|
|
62 |
examples under <a href="library/Sequents/">Sequents</a>, including
|
|
63 |
modal and linear logics. Again see the <a href="library/">Isabelle
|
|
64 |
theory library</a> for other examples.
|
|
65 |
|
|
66 |
|
|
67 |
<h3>Defining Logics</h3>
|
|
68 |
|
|
69 |
Logics are not hard-wired into Isabelle, but formulated within
|
|
70 |
Isabelle's meta logic: <strong>Isabelle/Pure</strong>. There are
|
|
71 |
quite a lot of syntactic and deductive tools available in generic
|
|
72 |
Isabelle. Thus defining new logics or extending existing ones
|
|
73 |
basically works as follows:
|
|
74 |
|
|
75 |
<ol>
|
|
76 |
|
|
77 |
<li> declare concrete syntax (via mixfix grammar and syntax macros),
|
|
78 |
|
|
79 |
<li> declare abstract syntax (as higher-order constants),
|
|
80 |
|
|
81 |
<li> declare inference rules (as meta-logical propositions),
|
|
82 |
|
|
83 |
<li> instantiate generic automatic proof tools (simplifier, classical
|
|
84 |
tableau prover etc.),
|
|
85 |
|
|
86 |
<li> manually code special proof procedures (via tacticals or
|
|
87 |
hand-written ML).
|
|
88 |
|
|
89 |
</ol>
|
|
90 |
|
|
91 |
The first three steps above are fully declarative and involve no ML
|
|
92 |
programming at all. Thus one already gets a decent deductive
|
|
93 |
environment based on primitive inferences (by employing the built-in
|
|
94 |
mechanisms of Isabelle/Pure, in particular higher-order unification
|
|
95 |
and resolution).
|
|
96 |
|
|
97 |
For sizable applications some degree of automated reasoning is
|
|
98 |
essential. Instantiating existing tools like the classical tableau
|
|
99 |
prover involves only minimal ML-based setup. One may also write
|
|
100 |
arbitrary proof procedures or even theory extension packages in ML,
|
|
101 |
without breaching system soundness (Isabelle follows the well-known
|
|
102 |
<em>LCF system approach</em> to achieve a secure system).
|