ITP-Course/exercises/e3.tex

117 lines
5.4 KiB
TeX
Raw Normal View History

\documentclass[a4paper,10pt,oneside]{scrartcl}
\usepackage[utf8]{inputenc}
\usepackage[a4paper]{geometry}
\usepackage{hyperref}
\usepackage{url}
\usepackage{color}
\usepackage{amsfonts}
\input{../hol_commands.inc}
\title{Exercise 3}
\begin{document}
\begin{center}
\usekomafont{sectioning}\usekomafont{part}ITP Exercise 3
\webversion{}{\\\small{due Friday 12th May}}
\end{center}
\bigskip
\section{Self-Study}
\subsection{Tactics and Tacticals}
You should read background information on the tactics mentioned in the
lecture. For each of the tactics and tacticals mentioned on the
slides, read the entry in the description manual.
\subsection{hol-mode}
Carefully study the \texttt{Goalstack} submenu of hol-mode. Learn the keycodes to
set goals, expand tactics, undo the last tactic expansion, restart the current proof and
drop the current goal.
\subsection{num and list Theories}
We will use the natural number theory \texttt{numTheory} as well as the list theories
\texttt{listTheory} and \texttt{rich\_listTheory} a lot. Please familiarise yourself with
these HOL theories (e.\,g.\ by looking at their signature in the HTML version of the HOL Reference). I also recommend reading up on other common theories like \texttt{optionTheory}, \texttt{oneTheory} and \texttt{pairTheory}. These won't be needed for this weeks exercises, though.
\section{Backward Proofs}
Part of this exercise is searching for useful existing lemmata (\eg
useful rewrite rules). Another part is understanding the effect of
different rewrite rules and their combination. Even if you have
experience with using HOL, please refrain from using automated rewrite
tools that spoil these purposes and have not been covered in the
lecture yet. Please don't use HOL's simplifier and especially don't
use stateful simp-sets. Similarly, please don't use the compute lib
via \eg \texttt{EVAL\_TAC}.
For these exercise, it might be beneficial to open the modules \texttt{listTheory} and \texttt{rich\_listTheory} via \ml{open listTheory rich\_listTheory}. You will notice that when opening them a lot of definitions are printed. This can consume quite some time when opening many large theories. Play around with the hol-mode commands
\texttt{Send region to HOL - hide non-errors} and \texttt{Quite - hide output except errors} to avoid this printout and the associated waiting time.
\subsection{Replay Proofs from Lecture}
If you never did a tactical proof in HOL before, I recommend following the example interactive proofs from Part VIII.
Type them in your own HOL session, make the same mistakes shown in the lecture. Use hol-mode to control the goalStack via
commands like expand, back-up, set goal, drop goal. Get a feeling for how to interactively develop a tactical proof.
This exercise is optional. If you are confident enough, feel free to skip it.
\subsection{Formalise Induction Proofs from Exercise 1}
For exercise sheet 1, some simple properties of appending lists were proved with pen and paper via induction. Let's now try to prove them formally using HOL. Prove \hol{!l.\ l ++ [] = l} via induction using the definition of \hol{APPEND} (\hol{++}). Similarly, prove the associativity of \texttt{APPEND}, \ie prove \hol{!l1 l2 l3.\ l1 ++ (l2 ++ l3) = (l1 ++ l2) ++ l3}.
\subsection{Reverse}
In HOL, \texttt{revAppend} is called \hol{REV}. Using any useful lemmata you can find, prove
the lemma \hol{!l1 l2.\ LENGTH (REV l1 l2) = (LENGTH l1 + LENGTH l2)}. Now, let us as an exercise reprove the existing lemmata \hol{REVERSE\_REV} and \hol{REV\_REVERSE\_LEM}. This means, prove, first prove \hol{!l1 l2.\ REV l1 l2 = REVERSE l1 ++ l2}. Then prove \hol{!l.\ REVERSE l = REV l []} using this theorem. You should not use the theorems \texttt{REVERSE\_REV} or \texttt{REV\_REVERSE\_LEM} in these proofs.
\subsection{Length of Drop}
Prove \hol{!l1 l2.\ LENGTH (DROP (LENGTH l2) (l1 ++ l2)) = LENGTH l1} with induction, \ie
without using lemmata like \texttt{LENGTH\_DROP}. Do one proof with \texttt{Induct\_on} and a very similar proof with \texttt{Induct}. This is a bit tricky.
Please play around with the proof for some time. If you can't figure it out, look at the hints at the end of this exercise sheet.
\subsection{Making Change}
On exercise sheet 1, you were asked to implement a function \texttt{make\_change} in SML. Let's now define it in HOL and prove some properties. Define the function \texttt{MAKE\_CHANGE} in HOL via
\begin{verbatim}
val MAKE_CHANGE_def = Define `
(MAKE_CHANGE [] a = if (a = 0) then [[]] else []) /\
(MAKE_CHANGE (c::cs) a = (
(if (c <= a /\ 0 < a) then
(MAP (\l. c::l) (MAKE_CHANGE cs (a - c)))
else []) ++ (MAKE_CHANGE cs a)))`;
\end{verbatim}
Prove that \hol{!cs.\ MAKE\_CHANGE cs 0 = [[]]} and
\hol{!cs a l.\ MEM l (MAKE\_CHANGE cs a) ==> (SUM l = a)} hold.
\section{Hints}
\subsection{Length of Drop}
For proving \hol{!l1 l2.\ LENGTH (DROP (LENGTH l2) (l1 ++ l2)) = LENGTH l1} induction on
the structure of \hol{l2} is a good strategy. However, one needs to be careful that \hol{l1} stays
universally quantified. Expanding naively with \hol{GEN\_TAC >> Induct} will
remove the needed universal quantification of \hol{l1}.
To solve this, you can either use \hol{Induct\_on `l2`} or get rid of both universal quantifiers and then introduce them in a different order again. This is achieved by \hol{REPEAT GEN\_TAC >> SPEC\_TAC (``l1:'a list``, ``l1:'a list``) >> SPEC\_TAC (``l2:'a list``, ``l2:'a list``)}.
\end{document}
%%% Local Variables:
%%% mode: latex
%%% TeX-master: t
%%% End: