% !TEX program = xelatex
\documentclass{article}
\usepackage{../style}
\usepackage{../langs}
\begin{document}
\section*{Coursework (Strand 2)}
\noindent This coursework is worth 20\% and is due on 13 December at
18:00. You are asked to prove the correctness of the regular expression
matcher from the lectures using the Isabelle theorem prover. You need to
submit a theory file containing this proof and also a document
describing your proof. The Isabelle theorem prover is available from
\begin{center}
\url{http://isabelle.in.tum.de}
\end{center}
\noindent This is an interactive theorem prover, meaning that
you can make definitions and state properties, and then help
the system with proving these properties. Sometimes the proofs
are also completely automatic. There is a shortish user guide for
Isabelle, called ``Programming and Proving in Isabelle/HOL''
at
\begin{center}
\url{http://isabelle.in.tum.de/documentation.html}
\end{center}
\noindent
and also a longer (free) book at
\begin{center}
\url{http://www.concrete-semantics.org}
\end{center}
\noindent The Isabelle theorem prover is operated through the
jEdit IDE, which might not be an editor that is widely known.
JEdit is documented in
\begin{center}
\url{http://isabelle.in.tum.de/dist/Isabelle2014/doc/jedit.pdf}
\end{center}
\noindent If you need more help or you are stuck somewhere,
please feel free to contact me (christian.urban at kcl.ac.uk). I
am one of the main developers of Isabelle and have used it for
approximately 16 years. One of the success stories of
Isabelle is the recent verification of a microkernel operating
system by an Australian group, see \url{http://sel4.systems}.
Their operating system is the only one that has been proved
correct according to its specification and is used for
application where high assurance, security and reliability is
needed (like in helicopters which fly over enemy territory).
\subsection*{The Task}
In this coursework you are asked to prove the correctness of the
regular expression matcher from the lectures in Isabelle. The matcher
should be able to deal with the usual (basic) regular expressions
\[
\ZERO,\; \ONE,\; c,\; r_1 + r_2,\; r_1 \cdot r_2,\; r^*
\]
\noindent
but also with the following extended regular expressions:
\begin{center}
\begin{tabular}{ll}
$r^{\{n\}}$ & exactly $n$-times\\
$r^{\{..m\}}$ & zero or more times $r$ but no more than $m$-times\\
$r^{\{n..\}}$ & at least $n$-times $r$\\
$r^{\{n..m\}}$ & at least $n$-times $r$ but no more than $m$-times\\
$\sim{}r$ & not-regular-expression of $r$\\
\end{tabular}
\end{center}
\noindent
You need to first specify what the matcher is
supposed to do and then to implement the algorithm. Finally you need
to prove that the algorithm meets the specification. The first two
parts are relatively easy, because the definitions in Isabelle will
look very similar to the mathematical definitions from the lectures or
the Scala code that is supplied at KEATS. For example very similar to
Scala, regular expressions are defined in Isabelle as an inductive
datatype:
\begin{lstlisting}[language={},numbers=none]
datatype rexp =
ZERO
| ONE
| CHAR char
| SEQ rexp rexp
| ALT rexp rexp
| STAR rexp
\end{lstlisting}
\noindent The meaning of regular expressions is given as
usual:
\begin{center}
\begin{tabular}{rcl@{\hspace{10mm}}l}
$L(\ZERO)$ & $\dn$ & $\varnothing$ & \pcode{ZERO}\\
$L(\ONE)$ & $\dn$ & $\{[]\}$ & \pcode{ONE}\\
$L(c)$ & $\dn$ & $\{[c]\}$ & \pcode{CHAR}\\
$L(r_1 + r_2)$ & $\dn$ & $L(r_1) \cup L(r_2)$ & \pcode{ALT}\\
$L(r_1 \cdot r_2)$ & $\dn$ & $L(r_1) \,@\, L(r_2)$ & \pcode{SEQ}\\
$L(r^*)$ & $\dn$ & $(L(r))^*$ & \pcode{STAR}\\
\end{tabular}
\end{center}
\noindent You would need to implement this function in order
to state the theorem about the correctness of the algorithm.
The function $L$ should in Isabelle take a \pcode{rexp} as
input and return a set of strings. Its type is
therefore
\begin{center}
\pcode{L} \pcode{::} \pcode{rexp} $\Rightarrow$ \pcode{string set}
\end{center}
\noindent Isabelle treats strings as an abbreviation for lists
of characters. This means you can pattern-match strings like
lists. The union operation on sets (for the \pcode{ALT}-case)
is a standard definition in Isabelle, but not the
concatenation operation on sets and also not the
star-operation. You would have to supply these definitions.
The concatenation operation can be defined in terms of the
append function, written \code{_ @ _} in Isabelle, for lists.
The star-operation can be defined as a ``big-union'' of
powers, like in the lectures, or directly as an inductive set.
The functions for the matcher are shown in
Figure~\ref{matcher}. The theorem that needs to be proved is
\begin{lstlisting}[numbers=none,language={},keywordstyle=\color{black}\ttfamily,mathescape]
theorem
"matches r s $\longleftrightarrow$ s $\in$ L r"
\end{lstlisting}
\noindent which states that the function \emph{matches} is
true if and only if the string is in the language of the
regular expression. A proof for this lemma will need
side-lemmas about \pcode{nullable} and \pcode{der}. An example
proof in Isabelle that will not be relevant for the theorem
above is given in Figure~\ref{proof}.
\begin{figure}[p]
\begin{lstlisting}[language={},keywordstyle=\color{black}\ttfamily,mathescape]
fun
nullable :: "rexp $\Rightarrow$ bool"
where
"nullable ZERO = False"
| "nullable ONE = True"
| "nullable (CHAR _) = False"
| "nullable (ALT r1 r2) = (nullable(r1) $\vee$ nullable(r2))"
| "nullable (SEQ r1 r2) = (nullable(r1) $\wedge$ nullable(r2))"
| "nullable (STAR _) = True"
fun
der :: "char $\Rightarrow$ rexp $\Rightarrow$ rexp"
where
"der c ZERO = ZERO"
| "der c ONE = ZERO"
| "der c (CHAR d) = (if c = d then ONE else ZERO)"
| "der c (ALT r1 r2) = ALT (der c r1) (der c r2)"
| "der c (SEQ r1 r2) =
(if (nullable r1) then ALT (SEQ (der c r1) r2) (der c r2)
else SEQ (der c r1) r2)"
| "der c (STAR r) = SEQ (der c r) (STAR r)"
fun
ders :: "rexp $\Rightarrow$ string $\Rightarrow$ rexp"
where
"ders r [] = r"
| "ders r (c # s) = ders (der c r) s"
fun
matches :: "rexp $\Rightarrow$ string $\Rightarrow$ bool"
where
"matches r s = nullable (ders r s)"
\end{lstlisting}
\caption{The definition of the matcher algorithm in
Isabelle.\label{matcher}}
\end{figure}
\begin{figure}[p]
\begin{lstlisting}[language={},keywordstyle=\color{black}\ttfamily,mathescape]
fun
zeroable :: "rexp $\Rightarrow$ bool"
where
"zeroable ZERO = True"
| "zeroable ONE = False"
| "zeroable (CHAR _) = False"
| "zeroable (ALT r1 r2) = (zeroable(r1) $\wedge$ zeroable(r2))"
| "zeroable (SEQ r1 r2) = (zeroable(r1) $\vee$ zeroable(r2))"
| "zeroable (STAR _) = False"
lemma
"zeroable r $\longleftrightarrow$ L r = {}"
proof (induct)
case (ZERO)
have "zeroable ZERO" "L ZERO = {}" by simp_all
then show "zeroable ZERO $\longleftrightarrow$ (L ZERO = {})" by simp
next
case (ONE)
have "$\neg$ zeroable ONE" "L ONE = {[]}" by simp_all
then show "zeroable ONE $\longleftrightarrow$ (L ONE = {})" by simp
next
case (CHAR c)
have "$\neg$ zeroable (CHAR c)" "L (CHAR c) = {[c]}" by simp_all
then show "zeroable (CHAR c) $\longleftrightarrow$ (L (CHAR c) = {})" by simp
next
case (ALT r1 r2)
have ih1: "zeroable r1 $\longleftrightarrow$ L r1 = {}" by fact
have ih2: "zeroable r2 $\longleftrightarrow$ L r2 = {}" by fact
show "zeroable (ALT r1 r2) $\longleftrightarrow$ (L (ALT r1 r2) = {})"
using ih1 ih2 by simp
next
case (SEQ r1 r2)
have ih1: "zeroable r1 $\longleftrightarrow$ L r1 = {}" by fact
have ih2: "zeroable r2 $\longleftrightarrow$ L r2 = {}" by fact
show "zeroable (SEQ r1 r2) $\longleftrightarrow$ (L (SEQ r1 r2) = {})"
using ih1 ih2 by (auto simp add: Conc_def)
next
case (STAR r)
have "$\neg$ zeroable (STAR r)" "[] $\in$ L (r) ^ 0" by simp_all
then show "zeroable (STAR r) $\longleftrightarrow$ (L (STAR r) = {})"
by (simp (no_asm) add: Star_def) blast
qed
\end{lstlisting}
\caption{An Isabelle proof about the function \texttt{zeroable}.\label{proof}}
\end{figure}
\end{document}
%%% Local Variables:
%%% mode: latex
%%% TeX-master: t
%%% End: