(*<*)

 theory Itrev = Main:;

 (*>*)

 text{* Function \isa{rev} has quadratic worst-case running time

 because it calls function \isa{\at} for each element of the list and

 \isa{\at} is linear in its first argument.  A linear time version of

 \isa{rev} reqires an extra argument where the result is accumulated

 gradually, using only \isa{\#}:*}

 consts itrev :: "'a list \\<Rightarrow> 'a list \\<Rightarrow> 'a list";

 primrec

 "itrev []     ys = ys"

 "itrev (x#xs) ys = itrev xs (x#ys)";

 text{*\noindent The behaviour of \isa{itrev} is simple: it reverses

 its first argument by stacking its elements onto the second argument,

 and returning that second argument when the first one becomes

 empty. Note that \isa{itrev} is tail-recursive, i.e.\ it can be

 compiled into a loop.

 Naturally, we would like to show that \isa{itrev} does indeed reverse

 its first argument provided the second one is empty: *};

 lemma "itrev xs [] = rev xs";

 txt{*\noindent

 There is no choice as to the induction variable, and we immediately simplify:

 *};

 apply(induct_tac xs, auto);

 txt{*\noindent

 Unfortunately, this is not a complete success:

 \begin{isabellepar}%

 ~1.~\dots~itrev~list~[]~=~rev~list~{\isasymLongrightarrow}~itrev~list~[a]~=~rev~list~@~[a]%

 \end{isabellepar}%

 Just as predicted above, the overall goal, and hence the induction

 hypothesis, is too weak to solve the induction step because of the fixed

 \isa{[]}. The corresponding heuristic:

 \begin{quote}

 {\em 3. Generalize goals for induction by replacing constants by variables.}

 \end{quote}

 Of course one cannot do this na\"{\i}vely: \isa{itrev xs ys = rev xs} is

 just not true---the correct generalization is

 *};

 (*<*)oops;(*>*)

 lemma "itrev xs ys = rev xs @ ys";

 txt{*\noindent

 If \isa{ys} is replaced by \isa{[]}, the right-hand side simplifies to

 @{term"rev xs"}, just as required.

 In this particular instance it was easy to guess the right generalization,

 but in more complex situations a good deal of creativity is needed. This is

 the main source of complications in inductive proofs.

 Although we now have two variables, only \isa{xs} is suitable for

 induction, and we repeat our above proof attempt. Unfortunately, we are still

 not there:

 \begin{isabellepar}%

 ~1.~{\isasymAnd}a~list.\isanewline

 ~~~~~~~itrev~list~ys~=~rev~list~@~ys~{\isasymLongrightarrow}\isanewline

 ~~~~~~~itrev~list~(a~\#~ys)~=~rev~list~@~a~\#~ys%

 \end{isabellepar}%

 The induction hypothesis is still too weak, but this time it takes no

 intuition to generalize: the problem is that \isa{ys} is fixed throughout

 the subgoal, but the induction hypothesis needs to be applied with

 @{term"a # ys"} instead of \isa{ys}. Hence we prove the theorem

 for all \isa{ys} instead of a fixed one:

 *};

 (*<*)oops;(*>*)

 lemma "\\<forall>ys. itrev xs ys = rev xs @ ys";

 txt{*\noindent

 This time induction on \isa{xs} followed by simplification succeeds. This

 leads to another heuristic for generalization:

 \begin{quote}

 {\em 4. Generalize goals for induction by universally quantifying all free

 variables {\em(except the induction variable itself!)}.}

 \end{quote}

 This prevents trivial failures like the above and does not change the

 provability of the goal. Because it is not always required, and may even

 complicate matters in some cases, this heuristic is often not

 applied blindly.

 *};

 (*<*)

 by(induct_tac xs, simp, simp);

 end

 (*>*)