doc-src/TutorialI/basics.tex
author nipkow
Wed Jan 24 12:29:10 2001 +0100 (2001-01-24)
changeset 10971 6852682eaf16
parent 10885 90695f46440b
child 10978 5eebea8f359f
permissions -rw-r--r--
*** empty log message ***
nipkow@8743
     1
\chapter{Basic Concepts}
nipkow@8743
     2
nipkow@8743
     3
\section{Introduction}
nipkow@8743
     4
nipkow@10971
     5
This is a tutorial on how to use the theorem prover Isabelle/HOL as a specification and
nipkow@8743
     6
verification system. Isabelle is a generic system for implementing logical
nipkow@8743
     7
formalisms, and Isabelle/HOL is the specialization of Isabelle for
nipkow@8743
     8
HOL, which abbreviates Higher-Order Logic. We introduce HOL step by step
nipkow@8743
     9
following the equation
nipkow@8743
    10
\[ \mbox{HOL} = \mbox{Functional Programming} + \mbox{Logic}. \]
nipkow@8743
    11
We assume that the reader is familiar with the basic concepts of both fields.
nipkow@8743
    12
For excellent introductions to functional programming consult the textbooks
nipkow@8743
    13
by Bird and Wadler~\cite{Bird-Wadler} or Paulson~\cite{paulson-ml2}.  Although
nipkow@8743
    14
this tutorial initially concentrates on functional programming, do not be
nipkow@8743
    15
misled: HOL can express most mathematical concepts, and functional
nipkow@8743
    16
programming is just one particularly simple and ubiquitous instance.
nipkow@8743
    17
nipkow@10971
    18
Isabelle~\cite{paulson-isa-book} is implemented in ML~\cite{SML}.
nipkow@10971
    19
This has influenced some of HOL's concrete syntax but is otherwise
nipkow@10971
    20
irrelevant for us because this tutorial is based on
nipkow@10971
    21
Isabelle/Isar~\cite{isabelle-isar-ref}, an extension of Isabelle
nipkow@10971
    22
with structured proofs.\footnote{Thus the full name of the system should be
nipkow@10971
    23
  Isabelle/Isar/HOL, but that is a bit of a mouthful.}
nipkow@8743
    24
nipkow@8743
    25
A tutorial is by definition incomplete.  Currently the tutorial only
nipkow@8743
    26
introduces the rudiments of Isar's proof language. To fully exploit the power
nipkow@8743
    27
of Isar you need to consult the Isabelle/Isar Reference
nipkow@8743
    28
Manual~\cite{isabelle-isar-ref}. If you want to use Isabelle's ML level
nipkow@8743
    29
directly (for example for writing your own proof procedures) see the Isabelle
nipkow@8743
    30
Reference Manual~\cite{isabelle-ref}; for details relating to HOL see the
nipkow@8743
    31
Isabelle/HOL manual~\cite{isabelle-HOL}. All manuals have a comprehensive
nipkow@8743
    32
index.
nipkow@8743
    33
nipkow@8743
    34
\section{Theories}
nipkow@8743
    35
\label{sec:Basic:Theories}
nipkow@8743
    36
nipkow@8743
    37
Working with Isabelle means creating theories. Roughly speaking, a
nipkow@8743
    38
\bfindex{theory} is a named collection of types, functions, and theorems,
nipkow@8743
    39
much like a module in a programming language or a specification in a
nipkow@8743
    40
specification language. In fact, theories in HOL can be either. The general
nipkow@8743
    41
format of a theory \texttt{T} is
nipkow@8743
    42
\begin{ttbox}
nipkow@8743
    43
theory T = B\(@1\) + \(\cdots\) + B\(@n\):
nipkow@8743
    44
\(\textit{declarations, definitions, and proofs}\)
nipkow@8743
    45
end
nipkow@8743
    46
\end{ttbox}
nipkow@8743
    47
where \texttt{B}$@1$, \dots, \texttt{B}$@n$ are the names of existing
nipkow@8743
    48
theories that \texttt{T} is based on and \texttt{\textit{declarations,
nipkow@8743
    49
    definitions, and proofs}} represents the newly introduced concepts
nipkow@8771
    50
(types, functions etc.) and proofs about them. The \texttt{B}$@i$ are the
nipkow@8743
    51
direct \textbf{parent theories}\indexbold{parent theory} of \texttt{T}.
nipkow@8743
    52
Everything defined in the parent theories (and their parents \dots) is
nipkow@8743
    53
automatically visible. To avoid name clashes, identifiers can be
nipkow@8743
    54
\textbf{qualified} by theory names as in \texttt{T.f} and
nipkow@8743
    55
\texttt{B.f}.\indexbold{identifier!qualified} Each theory \texttt{T} must
nipkow@8771
    56
reside in a \bfindex{theory file} named \texttt{T.thy}.
nipkow@8743
    57
nipkow@8743
    58
This tutorial is concerned with introducing you to the different linguistic
nipkow@8743
    59
constructs that can fill \textit{\texttt{declarations, definitions, and
nipkow@8743
    60
    proofs}} in the above theory template.  A complete grammar of the basic
nipkow@8743
    61
constructs is found in the Isabelle/Isar Reference Manual.
nipkow@8743
    62
paulson@10885
    63
HOL's theory collection is available online at
nipkow@8743
    64
\begin{center}\small
nipkow@8743
    65
    \url{http://isabelle.in.tum.de/library/}
nipkow@8743
    66
\end{center}
paulson@10885
    67
and is recommended browsing. Note that most of the theories 
nipkow@9541
    68
are based on classical Isabelle without the Isar extension. This means that
nipkow@9541
    69
they look slightly different than the theories in this tutorial, and that all
nipkow@9541
    70
proofs are in separate ML files.
nipkow@9541
    71
nipkow@8743
    72
\begin{warn}
nipkow@9792
    73
  HOL contains a theory \isaindexbold{Main}, the union of all the basic
paulson@10885
    74
  predefined theories like arithmetic, lists, sets, etc.  
paulson@10885
    75
  Unless you know what you are doing, always include \isa{Main}
nipkow@10971
    76
  as a direct or indirect parent of all your theories.
nipkow@8743
    77
\end{warn}
nipkow@8743
    78
nipkow@8743
    79
paulson@10885
    80
\section{Types, Terms and Formulae}
nipkow@8743
    81
\label{sec:TypesTermsForms}
nipkow@8743
    82
\indexbold{type}
nipkow@8743
    83
paulson@10795
    84
Embedded in a theory are the types, terms and formulae of HOL\@. HOL is a typed
nipkow@8771
    85
logic whose type system resembles that of functional programming languages
nipkow@8771
    86
like ML or Haskell. Thus there are
nipkow@8743
    87
\begin{description}
nipkow@8771
    88
\item[base types,] in particular \isaindex{bool}, the type of truth values,
nipkow@8771
    89
and \isaindex{nat}, the type of natural numbers.
nipkow@8771
    90
\item[type constructors,] in particular \isaindex{list}, the type of
nipkow@8771
    91
lists, and \isaindex{set}, the type of sets. Type constructors are written
nipkow@8771
    92
postfix, e.g.\ \isa{(nat)list} is the type of lists whose elements are
nipkow@8743
    93
natural numbers. Parentheses around single arguments can be dropped (as in
nipkow@8771
    94
\isa{nat list}), multiple arguments are separated by commas (as in
nipkow@8771
    95
\isa{(bool,nat)ty}).
nipkow@8743
    96
\item[function types,] denoted by \isasymFun\indexbold{$IsaFun@\isasymFun}.
nipkow@8771
    97
  In HOL \isasymFun\ represents \emph{total} functions only. As is customary,
nipkow@8771
    98
  \isa{$\tau@1$ \isasymFun~$\tau@2$ \isasymFun~$\tau@3$} means
nipkow@8771
    99
  \isa{$\tau@1$ \isasymFun~($\tau@2$ \isasymFun~$\tau@3$)}. Isabelle also
nipkow@8771
   100
  supports the notation \isa{[$\tau@1,\dots,\tau@n$] \isasymFun~$\tau$}
nipkow@8771
   101
  which abbreviates \isa{$\tau@1$ \isasymFun~$\cdots$ \isasymFun~$\tau@n$
nipkow@8743
   102
    \isasymFun~$\tau$}.
nipkow@8771
   103
\item[type variables,]\indexbold{type variable}\indexbold{variable!type}
paulson@10795
   104
  denoted by \ttindexboldpos{'a}{$Isatype}, \isa{'b} etc., just like in ML\@. They give rise
nipkow@8771
   105
  to polymorphic types like \isa{'a \isasymFun~'a}, the type of the identity
nipkow@8771
   106
  function.
nipkow@8743
   107
\end{description}
nipkow@8743
   108
\begin{warn}
nipkow@8743
   109
  Types are extremely important because they prevent us from writing
nipkow@8743
   110
  nonsense.  Isabelle insists that all terms and formulae must be well-typed
nipkow@8743
   111
  and will print an error message if a type mismatch is encountered. To
nipkow@8743
   112
  reduce the amount of explicit type information that needs to be provided by
nipkow@8743
   113
  the user, Isabelle infers the type of all variables automatically (this is
nipkow@8743
   114
  called \bfindex{type inference}) and keeps quiet about it. Occasionally
nipkow@8743
   115
  this may lead to misunderstandings between you and the system. If anything
nipkow@8743
   116
  strange happens, we recommend to set the \rmindex{flag}
nipkow@9792
   117
  \isaindexbold{show_types} that tells Isabelle to display type information
nipkow@8743
   118
  that is usually suppressed: simply type
nipkow@8743
   119
\begin{ttbox}
nipkow@8743
   120
ML "set show_types"
nipkow@8743
   121
\end{ttbox}
nipkow@8743
   122
nipkow@8743
   123
\noindent
nipkow@10971
   124
This can be reversed by \texttt{ML "reset show_types"}. Various other flags,
nipkow@10971
   125
which we introduce as we go along,
nipkow@8771
   126
can be set and reset in the same manner.\indexbold{flag!(re)setting}
nipkow@8743
   127
\end{warn}
nipkow@8743
   128
nipkow@8743
   129
nipkow@8743
   130
\textbf{Terms}\indexbold{term} are formed as in functional programming by
nipkow@8771
   131
applying functions to arguments. If \isa{f} is a function of type
nipkow@8771
   132
\isa{$\tau@1$ \isasymFun~$\tau@2$} and \isa{t} is a term of type
nipkow@8771
   133
$\tau@1$ then \isa{f~t} is a term of type $\tau@2$. HOL also supports
nipkow@8771
   134
infix functions like \isa{+} and some basic constructs from functional
nipkow@8743
   135
programming:
nipkow@8743
   136
\begin{description}
nipkow@8771
   137
\item[\isa{if $b$ then $t@1$ else $t@2$}]\indexbold{*if}
nipkow@8743
   138
means what you think it means and requires that
nipkow@8771
   139
$b$ is of type \isa{bool} and $t@1$ and $t@2$ are of the same type.
nipkow@8771
   140
\item[\isa{let $x$ = $t$ in $u$}]\indexbold{*let}
nipkow@8743
   141
is equivalent to $u$ where all occurrences of $x$ have been replaced by
nipkow@8743
   142
$t$. For example,
nipkow@8771
   143
\isa{let x = 0 in x+x} is equivalent to \isa{0+0}. Multiple bindings are separated
nipkow@8771
   144
by semicolons: \isa{let $x@1$ = $t@1$; \dots; $x@n$ = $t@n$ in $u$}.
nipkow@8771
   145
\item[\isa{case $e$ of $c@1$ \isasymFun~$e@1$ |~\dots~| $c@n$ \isasymFun~$e@n$}]
nipkow@8743
   146
\indexbold{*case}
nipkow@8771
   147
evaluates to $e@i$ if $e$ is of the form $c@i$.
nipkow@8743
   148
\end{description}
nipkow@8743
   149
nipkow@8743
   150
Terms may also contain
nipkow@8743
   151
\isasymlambda-abstractions\indexbold{$Isalam@\isasymlambda}. For example,
nipkow@8771
   152
\isa{\isasymlambda{}x.~x+1} is the function that takes an argument \isa{x} and
nipkow@8771
   153
returns \isa{x+1}. Instead of
nipkow@8771
   154
\isa{\isasymlambda{}x.\isasymlambda{}y.\isasymlambda{}z.~$t$} we can write
nipkow@8771
   155
\isa{\isasymlambda{}x~y~z.~$t$}.
nipkow@8743
   156
nipkow@8771
   157
\textbf{Formulae}\indexbold{formula} are terms of type \isaindexbold{bool}.
nipkow@8771
   158
There are the basic constants \isaindexbold{True} and \isaindexbold{False} and
nipkow@8771
   159
the usual logical connectives (in decreasing order of priority):
nipkow@8771
   160
\indexboldpos{\isasymnot}{$HOL0not}, \indexboldpos{\isasymand}{$HOL0and},
nipkow@8771
   161
\indexboldpos{\isasymor}{$HOL0or}, and \indexboldpos{\isasymimp}{$HOL0imp},
nipkow@8743
   162
all of which (except the unary \isasymnot) associate to the right. In
nipkow@8771
   163
particular \isa{A \isasymimp~B \isasymimp~C} means \isa{A \isasymimp~(B
nipkow@8771
   164
  \isasymimp~C)} and is thus logically equivalent to \isa{A \isasymand~B
nipkow@8771
   165
  \isasymimp~C} (which is \isa{(A \isasymand~B) \isasymimp~C}).
nipkow@8743
   166
nipkow@8743
   167
Equality is available in the form of the infix function
nipkow@8771
   168
\isa{=}\indexbold{$HOL0eq@\texttt{=}} of type \isa{'a \isasymFun~'a
nipkow@8771
   169
  \isasymFun~bool}. Thus \isa{$t@1$ = $t@2$} is a formula provided $t@1$
nipkow@8743
   170
and $t@2$ are terms of the same type. In case $t@1$ and $t@2$ are of type
nipkow@8771
   171
\isa{bool}, \isa{=} acts as if-and-only-if. The formula
nipkow@8771
   172
\isa{$t@1$~\isasymnoteq~$t@2$} is merely an abbreviation for
nipkow@8771
   173
\isa{\isasymnot($t@1$ = $t@2$)}.
nipkow@8743
   174
paulson@10795
   175
Quantifiers are written as
nipkow@8771
   176
\isa{\isasymforall{}x.~$P$}\indexbold{$HOL0All@\isasymforall} and
nipkow@8771
   177
\isa{\isasymexists{}x.~$P$}\indexbold{$HOL0Ex@\isasymexists}.  There is
nipkow@8771
   178
even \isa{\isasymuniqex{}x.~$P$}\index{$HOL0ExU@\isasymuniqex|bold}, which
nipkow@8771
   179
means that there exists exactly one \isa{x} that satisfies \isa{$P$}.  Nested
nipkow@8771
   180
quantifications can be abbreviated: \isa{\isasymforall{}x~y~z.~$P$} means
nipkow@8771
   181
\isa{\isasymforall{}x.\isasymforall{}y.\isasymforall{}z.~$P$}.
nipkow@8743
   182
nipkow@8743
   183
Despite type inference, it is sometimes necessary to attach explicit
nipkow@8771
   184
\textbf{type constraints}\indexbold{type constraint} to a term.  The syntax is
nipkow@8771
   185
\isa{$t$::$\tau$} as in \isa{x < (y::nat)}. Note that
nipkow@10538
   186
\ttindexboldpos{::}{$Isatype} binds weakly and should therefore be enclosed
nipkow@8771
   187
in parentheses: \isa{x < y::nat} is ill-typed because it is interpreted as
paulson@10795
   188
\isa{(x < y)::nat}. The main reason for type constraints is overloading of
nipkow@10538
   189
functions like \isa{+}, \isa{*} and \isa{<}. See {\S}\ref{sec:overloading} for
nipkow@10695
   190
a full discussion of overloading and Table~\ref{tab:overloading} for the most
nipkow@10695
   191
important overloaded function symbols.
nipkow@8743
   192
nipkow@8743
   193
\begin{warn}
nipkow@8743
   194
In general, HOL's concrete syntax tries to follow the conventions of
nipkow@8743
   195
functional programming and mathematics. Below we list the main rules that you
nipkow@8743
   196
should be familiar with to avoid certain syntactic traps. A particular
nipkow@8743
   197
problem for novices can be the priority of operators. If you are unsure, use
paulson@10795
   198
additional parentheses. In those cases where Isabelle echoes your
nipkow@10971
   199
input, you can see which parentheses are dropped --- they were superfluous. If
nipkow@8743
   200
you are unsure how to interpret Isabelle's output because you don't know
paulson@10795
   201
where the (dropped) parentheses go, set the \rmindex{flag}
nipkow@9792
   202
\isaindexbold{show_brackets}:
nipkow@8743
   203
\begin{ttbox}
nipkow@8743
   204
ML "set show_brackets"; \(\dots\); ML "reset show_brackets";
nipkow@8743
   205
\end{ttbox}
nipkow@8743
   206
\end{warn}
nipkow@8743
   207
nipkow@8743
   208
\begin{itemize}
nipkow@8743
   209
\item
nipkow@8771
   210
Remember that \isa{f t u} means \isa{(f t) u} and not \isa{f(t u)}!
nipkow@8743
   211
\item
nipkow@8771
   212
Isabelle allows infix functions like \isa{+}. The prefix form of function
nipkow@8771
   213
application binds more strongly than anything else and hence \isa{f~x + y}
nipkow@8771
   214
means \isa{(f~x)~+~y} and not \isa{f(x+y)}.
nipkow@8743
   215
\item Remember that in HOL if-and-only-if is expressed using equality.  But
nipkow@8743
   216
  equality has a high priority, as befitting a relation, while if-and-only-if
nipkow@8771
   217
  typically has the lowest priority.  Thus, \isa{\isasymnot~\isasymnot~P =
nipkow@8771
   218
    P} means \isa{\isasymnot\isasymnot(P = P)} and not
nipkow@8771
   219
  \isa{(\isasymnot\isasymnot P) = P}. When using \isa{=} to mean
nipkow@8771
   220
  logical equivalence, enclose both operands in parentheses, as in \isa{(A
nipkow@8743
   221
    \isasymand~B) = (B \isasymand~A)}.
nipkow@8743
   222
\item
nipkow@8743
   223
Constructs with an opening but without a closing delimiter bind very weakly
nipkow@8743
   224
and should therefore be enclosed in parentheses if they appear in subterms, as
nipkow@10971
   225
in \isa{(\isasymlambda{}x.~x) = f}. This includes \isaindex{if},
nipkow@8771
   226
\isaindex{let}, \isaindex{case}, \isa{\isasymlambda}, and quantifiers.
nipkow@8743
   227
\item
nipkow@8771
   228
Never write \isa{\isasymlambda{}x.x} or \isa{\isasymforall{}x.x=x}
nipkow@8771
   229
because \isa{x.x} is always read as a single qualified identifier that
nipkow@8771
   230
refers to an item \isa{x} in theory \isa{x}. Write
nipkow@8771
   231
\isa{\isasymlambda{}x.~x} and \isa{\isasymforall{}x.~x=x} instead.
nipkow@8771
   232
\item Identifiers\indexbold{identifier} may contain \isa{_} and \isa{'}.
nipkow@8743
   233
\end{itemize}
nipkow@8743
   234
nipkow@8771
   235
For the sake of readability the usual mathematical symbols are used throughout
nipkow@8771
   236
the tutorial. Their ASCII-equivalents are shown in figure~\ref{fig:ascii} in
nipkow@8771
   237
the appendix.
nipkow@8771
   238
nipkow@8743
   239
nipkow@8743
   240
\section{Variables}
nipkow@8743
   241
\label{sec:variables}
nipkow@8743
   242
\indexbold{variable}
nipkow@8743
   243
nipkow@8743
   244
Isabelle distinguishes free and bound variables just as is customary. Bound
nipkow@8743
   245
variables are automatically renamed to avoid clashes with free variables. In
nipkow@8743
   246
addition, Isabelle has a third kind of variable, called a \bfindex{schematic
nipkow@8743
   247
  variable}\indexbold{variable!schematic} or \bfindex{unknown}, which starts
nipkow@8771
   248
with a \isa{?}.  Logically, an unknown is a free variable. But it may be
nipkow@8743
   249
instantiated by another term during the proof process. For example, the
nipkow@8771
   250
mathematical theorem $x = x$ is represented in Isabelle as \isa{?x = ?x},
nipkow@8743
   251
which means that Isabelle can instantiate it arbitrarily. This is in contrast
nipkow@8743
   252
to ordinary variables, which remain fixed. The programming language Prolog
nipkow@8743
   253
calls unknowns {\em logical\/} variables.
nipkow@8743
   254
nipkow@8743
   255
Most of the time you can and should ignore unknowns and work with ordinary
nipkow@8743
   256
variables. Just don't be surprised that after you have finished the proof of
nipkow@8743
   257
a theorem, Isabelle will turn your free variables into unknowns: it merely
nipkow@8743
   258
indicates that Isabelle will automatically instantiate those unknowns
nipkow@8743
   259
suitably when the theorem is used in some other proof.
nipkow@9689
   260
Note that for readability we often drop the \isa{?}s when displaying a theorem.
nipkow@8743
   261
\begin{warn}
nipkow@8771
   262
  If you use \isa{?}\index{$HOL0Ex@\texttt{?}} as an existential
nipkow@8771
   263
  quantifier, it needs to be followed by a space. Otherwise \isa{?x} is
nipkow@8743
   264
  interpreted as a schematic variable.
nipkow@8743
   265
\end{warn}
nipkow@8743
   266
paulson@10885
   267
\section{Interaction and Interfaces}
nipkow@8771
   268
nipkow@8771
   269
Interaction with Isabelle can either occur at the shell level or through more
nipkow@8771
   270
advanced interfaces. To keep the tutorial independent of the interface we
nipkow@8771
   271
have phrased the description of the intraction in a neutral language. For
nipkow@8771
   272
example, the phrase ``to abandon a proof'' means to type \isacommand{oops} at the
nipkow@8771
   273
shell level, which is explained the first time the phrase is used. Other
nipkow@8771
   274
interfaces perform the same act by cursor movements and/or mouse clicks.
nipkow@8771
   275
Although shell-based interaction is quite feasible for the kind of proof
nipkow@8771
   276
scripts currently presented in this tutorial, the recommended interface for
nipkow@8771
   277
Isabelle/Isar is the Emacs-based \bfindex{Proof
nipkow@8771
   278
  General}~\cite{Aspinall:TACAS:2000,proofgeneral}.
nipkow@8771
   279
nipkow@8771
   280
Some interfaces (including the shell level) offer special fonts with
nipkow@8771
   281
mathematical symbols. For those that do not, remember that ASCII-equivalents
nipkow@8771
   282
are shown in figure~\ref{fig:ascii} in the appendix.
nipkow@8771
   283
nipkow@9541
   284
Finally, a word about semicolons.\indexbold{$Isar@\texttt{;}} 
nipkow@9541
   285
Commands may but need not be terminated by semicolons.
nipkow@9541
   286
At the shell level it is advisable to use semicolons to enforce that a command
nipkow@8771
   287
is executed immediately; otherwise Isabelle may wait for the next keyword
nipkow@9541
   288
before it knows that the command is complete.
nipkow@8771
   289
nipkow@8771
   290
paulson@10885
   291
\section{Getting Started}
nipkow@8743
   292
nipkow@8743
   293
Assuming you have installed Isabelle, you start it by typing \texttt{isabelle
nipkow@8743
   294
  -I HOL} in a shell window.\footnote{Simply executing \texttt{isabelle -I}
nipkow@8743
   295
  starts the default logic, which usually is already \texttt{HOL}.  This is
nipkow@8743
   296
  controlled by the \texttt{ISABELLE_LOGIC} setting, see \emph{The Isabelle
nipkow@8743
   297
    System Manual} for more details.} This presents you with Isabelle's most
nipkow@8743
   298
basic ASCII interface.  In addition you need to open an editor window to
nipkow@8743
   299
create theory files.  While you are developing a theory, we recommend to
nipkow@8743
   300
type each command into the file first and then enter it into Isabelle by
nipkow@8743
   301
copy-and-paste, thus ensuring that you have a complete record of your theory.
nipkow@8771
   302
As mentioned above, Proof General offers a much superior interface.
paulson@10795
   303
If you have installed Proof General, you can start it by typing \texttt{Isabelle}.