+ <sect2 id="letrec">
+ <title><emphasis role="bold">letrec</emphasis> &TODO;</title>
+ <titleabbrev>&TODO;</titleabbrev>
+ <para>&TODO;</para>
+ </sect2>
+ <sect2 id="inductive">
+ <title>[<emphasis role="bold">inductive</emphasis>|<emphasis role="bold">coinductive</emphasis>] &id; [&args2;]… <emphasis role="bold">:</emphasis> &term; <emphasis role="bold">≝</emphasis> [<emphasis role="bold">|</emphasis>] [&id;<emphasis role="bold">:</emphasis>&term;] [<emphasis role="bold">|</emphasis> &id;<emphasis role="bold">:</emphasis>&term;]…
+[<emphasis role="bold">with</emphasis> &id; <emphasis role="bold">:</emphasis> &term; <emphasis role="bold">≝</emphasis> [<emphasis role="bold">|</emphasis>] [&id;<emphasis role="bold">:</emphasis>&term;] [<emphasis role="bold">|</emphasis> &id;<emphasis role="bold">:</emphasis>&term;]…]…
+</title>
+ <titleabbrev>(co)inductive types declaration</titleabbrev>
+ <para><userinput>inductive i x y z: S ≝ k1:T1 | … | kn:Tn with i' : S' ≝ k1':T1' | … | km':Tm'</userinput></para>
+ <para>Declares a family of two mutually inductive types
+ <command>i</command> and <command>i'</command> whose types are
+ <command>S</command> and <command>S'</command>, which must be convertible
+ to sorts.</para>
+ <para>The constructors <command>ki</command> of type <command>Ti</command>
+ and <command>ki'</command> of type <command>Ti'</command> are also
+ simultaneously declared. The declared types <command>i</command> and
+ <command>i'</command> may occur in the types of the constructors, but
+ only in strongly positive positions according to the rules of the
+ calculus.</para>
+ <para>The whole family is parameterized over the arguments <command>x,y,z</command>.</para>
+ <para>If the keyword <command>coinductive</command> is used, the declared
+ types are considered mutually coinductive.</para>
+ <para>Elimination principles for the record are automatically generated
+ by Matita, if allowed by the typing rules of the calculus according to
+ the sort <command>S</command>. If generated,
+ they are named <command>i_ind</command>, <command>i_rec</command> and
+ <command>i_rect</command> according to the sort of their induction
+ predicate.</para>
+ </sect2>
+ <sect2 id="record">
+ <title><emphasis role="bold">record</emphasis> &id; [&args2;]… <emphasis role="bold">:</emphasis> &term; <emphasis role="bold">≝</emphasis><emphasis role="bold">{</emphasis>[&id; [<emphasis role="bold">:</emphasis>|<emphasis role="bold">:></emphasis>] &term;] [<emphasis role="bold">;</emphasis>&id; [<emphasis role="bold">:</emphasis>|<emphasis role="bold">:></emphasis>] &term;]…<emphasis role="bold">}</emphasis></title>
+ <titleabbrev>record</titleabbrev>
+ <para><userinput>record id x y z: S ≝ { f1: T1; …; fn:Tn }</userinput></para>
+ <para>Declares a new record family <command>id</command> parameterized over
+ <command>x,y,z</command>.</para>
+ <para><command>S</command> is the type of the record
+ and it must be convertible to a sort.</para>
+ <para>Each field <command>fi</command> is declared by giving its type
+ <command>Ti</command>. A record without any field is admitted.</para>
+ <para>Elimination principles for the record are automatically generated
+ by Matita, if allowed by the typing rules of the calculus according to
+ the sort <command>S</command>. If generated,
+ they are named <command>i_ind</command>, <command>i_rec</command> and
+ <command>i_rect</command> according to the sort of their induction
+ predicate.</para>
+ <para>For each field <command>fi</command> a record projection
+ <command>fi</command> is also automatically generated if projection
+ is allowed by the typing rules of the calculus according to the
+ sort <command>S</command>, the type <command>T1</command> and
+ the definability of depending record projections.</para>
+ <para>If the type of a field is declared with <command>:></command>,
+ the corresponding record projection becomes an implicit coercion.
+ This is just syntactic sugar and it has the same effect of declaring the
+ record projection as a coercion later on.</para>
+ </sect2>
+ </sect1>
+
+ <sect1 id="proofs">
+ <title>Proofs</title>
+ <sect2 id="theorem">
+ <title><emphasis role="bold">theorem</emphasis> &id;[<emphasis role="bold">:</emphasis> &term;] [<emphasis role="bold">≝</emphasis> &term;]</title>
+ <titleabbrev>theorem</titleabbrev>
+ <para><userinput>theorem f: P ≝ p</userinput></para>
+ <para>Proves a new theorem <command>f</command> whose thesis is
+ <command>P</command>.</para>
+ <para>If <command>p</command> is provided, it must be a proof term for
+ <command>P</command>. Otherwise an interactive proof is started.</para>
+ <para><command>P</command> can be omitted only if the proof is not
+ interactive.</para>
+ <para>Proving a theorem already proved in the library is an error.
+ To provide an alternative name and proof for the same theorem, use
+ <command>variant f: P ≝ p</command>.</para>
+ <para>A warning is raised if the name of the theorem cannot be obtained
+ by mangling the name of the constants in its thesis.</para>
+ <para>Notice that the command is equivalent to <command>definition f: T ≝ t</command>.</para>
+ </sect2>
+ <sect2 id="variant">
+ <title><emphasis role="bold">variant</emphasis> &id;<emphasis role="bold">:</emphasis> &term; <emphasis role="bold">≝</emphasis> &term;</title>
+ <titleabbrev>variant</titleabbrev>
+ <para><userinput>variant f: T ≝ t</userinput></para>
+ <para>Same as <command>theorem f: T ≝ t</command>, but it does not
+ complain if the theorem has already been proved. To be used to give
+ an alternative name or proof to a theorem.</para>
+ </sect2>
+ <sect2 id="lemma">
+ <title><emphasis role="bold">lemma</emphasis> &id;[<emphasis role="bold">:</emphasis> &term;] [<emphasis role="bold">≝</emphasis> &term;]</title>
+ <titleabbrev>lemma</titleabbrev>
+ <para><userinput>lemma f: T ≝ t</userinput></para>
+ <para>Same as <command>theorem f: T ≝ t</command></para>
+ </sect2>
+ <sect2 id="fact">
+ <title><emphasis role="bold">fact</emphasis> &id;[<emphasis role="bold">:</emphasis> &term;] [<emphasis role="bold">≝</emphasis> &term;]</title>
+ <titleabbrev>fact</titleabbrev>
+ <para><userinput>fact f: T ≝ t</userinput></para>
+ <para>Same as <command>theorem f: T ≝ t</command></para>
+ </sect2>
+ <sect2 id="remark">
+ <title><emphasis role="bold">remark</emphasis> &id;[<emphasis role="bold">:</emphasis> &term;] [<emphasis role="bold">≝</emphasis> &term;]</title>
+ <titleabbrev>remark</titleabbrev>
+ <para><userinput>remark f: T ≝ t</userinput></para>
+ <para>Same as <command>theorem f: T ≝ t</command></para>
+ </sect2>
+ </sect1>
+
+ <sect1 id="tacticargs">
+ <title>Tactic arguments</title>
+ <para>This section documents the syntax of some recurring arguments for
+ tactics.</para>
+
+ <sect2 id="introsspec">
+ <title>intros-spec</title>
+ <table frame="topbot" rowsep="0" colsep="0" role="grammar">
+ <title>intros-spec</title>
+ <tgroup cols="4">
+ <tbody>
+ <row>
+ <entry id="grammar.intros-spec">&intros-spec;</entry>
+ <entry>::=</entry>
+ <entry>[&nat;] [<emphasis role="bold">(</emphasis>[&id;]…<emphasis role="bold">)</emphasis>]</entry>
+ </row>
+ </tbody>
+ </tgroup>
+ </table>
+ <para>The natural number is the number of new hypotheses to be introduced. The list of identifiers gives the name for the first hypotheses.</para>
+ </sect2>
+
+ <sect2 id="pattern">
+ <title>pattern</title>
+ <table frame="topbot" rowsep="0" colsep="0" role="grammar">
+ <title>pattern</title>
+ <tgroup cols="4">
+ <tbody>
+ <row>
+ <entry id="grammar.pattern">&pattern;</entry>
+ <entry>::=</entry>
+ <entry><emphasis role="bold">in</emphasis>
+ [&id;[<emphasis role="bold">:</emphasis> &path;]]…
+ [<emphasis role="bold">⊢</emphasis> &path;]]</entry>
+ <entry>simple pattern</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">in match</emphasis> &term;
+ [<emphasis role="bold">in</emphasis>
+ [&id;[<emphasis role="bold">:</emphasis> &path;]]…
+ [<emphasis role="bold">⊢</emphasis> &path;]]</entry>
+ <entry>full pattern</entry>
+ </row>
+ </tbody>
+ </tgroup>
+ </table>
+ <table frame="topbot" rowsep="0" colsep="0" role="grammar">
+ <title>path</title>
+ <tgroup cols="4">
+ <tbody>
+ <row>
+ <entry id="grammar.path">&path;</entry>
+ <entry>::=</entry>
+ <entry><emphasis>〈〈any &sterm; whithout occurrences of <emphasis role="bold">Set</emphasis>, <emphasis role="bold">Prop</emphasis>, <emphasis role="bold">CProp</emphasis>, <emphasis role="bold">Type</emphasis>, &id;, &uri; and user provided notation; however, <emphasis role="bold">%</emphasis> is now an additional production for &sterm;〉〉</emphasis></entry>
+ </row>
+ </tbody>
+ </tgroup>
+ </table>
+ <para>A <emphasis>path</emphasis> locates zero or more subterms of a given term by mimicking the term structure up to:</para>
+ <orderedlist>
+ <listitem><para>Occurrences of the subterms to locate that are
+ represented by <emphasis role="bold">%</emphasis>.</para></listitem>
+ <listitem><para>Subterms without any occurrence of subterms to locate
+ that can be represented by <emphasis role="bold">?</emphasis>.
+ </para></listitem>
+ </orderedlist>
+ <para>For instance, the path
+ <userinput>∀_,_:?.(? ? % ?)→(? ? ? %)</userinput>
+ locates at once the subterms
+ <userinput>x+y</userinput> and <userinput>x*y</userinput> in the
+ term <userinput>∀x,y:nat.x+y=1→0=x*y</userinput>
+ (where the notation <userinput>A=B</userinput> hides the term
+ <userinput>(eq T A B)</userinput> for some type <userinput>T</userinput>).
+ </para>
+ <para>A <emphasis>simple pattern</emphasis> extends paths to locate
+ subterms in a whole sequent. In particular, the pattern
+ <userinput>in H: p K: q ⊢ r</userinput> locates at once all the subterms
+ located by the pattern <userinput>r</userinput> in the conclusion of the
+ sequent and by the patterns <userinput>p</userinput> and
+ <userinput>q</userinput> in the hypotheses <userinput>H</userinput>
+ and <userinput>K</userinput> of the sequent.
+ </para>
+ <para>If no list of hypotheses is provided in a simple pattern, no subterm
+ is selected in the hypothesis. If the <userinput>⊢ p</userinput>
+ part of the pattern is not provided, no subterm will be matched in the
+ conclusion if at least one hypothesis is provided; otherwise the whole
+ conclusion is selected.
+ </para>
+ <para>Finally, a <emphasis>full pattern</emphasis> is interpreted in three
+ steps. In the first step the <userinput>match T in</userinput>
+ part is ignored and a set <emphasis>S</emphasis> of subterms is
+ located as for the case of
+ simple patterns. In the second step the term <userinput>T</userinput>
+ is parsed and interpreted in the context of each subterm
+ <emphasis>s ∈ S</emphasis>. In the last term for each
+ <emphasis>s ∈ S</emphasis> the interpreted term <userinput>T</userinput>
+ computed in the previous step is looked for. The final set of subterms
+ located by the full pattern is the set of occurrences of
+ the interpreted <userinput>T</userinput> in the subterms <emphasis>s</emphasis>.
+ </para>
+ <para>A full pattern can always be replaced by a simple pattern,
+ often at the cost of increased verbosity or decreased readability.</para>
+ <para>Example: the pattern
+ <userinput>⊢ in match x+y in ∀_,_:?.(? ? % ?)</userinput>
+ locates only the first occurrence of <userinput>x+y</userinput>
+ in the sequent <userinput>x,y: nat ⊢ ∀z,w:nat. (x+y) * (z+w) =
+ z * (x+y) + w * (x+y)</userinput>. The corresponding simple pattern
+ is <userinput>⊢ ∀_,_:?.(? ? (? % ?) ?)</userinput>.
+ </para>
+ <para>Every tactic that acts on subterms of the selected sequents have
+ a pattern argument for uniformity. To automatically generate a simple
+ pattern:</para>
+ <orderedlist>
+ <listitem><para>Select in the current goal the subterms to pass to the
+ tactic by using the mouse. In order to perform a multiple selection of
+ subterms, hold the Ctrl key while selecting every subterm after the
+ first one.</para></listitem>
+ <listitem><para>From the contextual menu select "Copy".</para></listitem>
+ <listitem><para>From the "Edit" or the contextual menu select
+ "Paste as pattern"</para></listitem>
+ </orderedlist>
+ </sect2>
+
+ <sect2 id="reduction-kind">
+ <title>reduction-kind</title>
+ <para>Reduction kinds are normalization functions that transform a term
+ to a convertible but simpler one. Each reduction kind can be used both
+ as a tactic argument and as a stand-alone tactic.</para>
+ <table frame="topbot" rowsep="0" colsep="0" role="grammar">
+ <title>reduction-kind</title>
+ <tgroup cols="4">
+ <tbody>
+ <row>
+ <entry id="grammar.reduction-kind">&reduction-kind;</entry>
+ <entry>::=</entry>
+ <entry><emphasis role="bold">normalize</emphasis></entry>
+ <entry>Computes the βδιζ-normal form</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">reduce</emphasis></entry>
+ <entry>Computes the βδιζ-normal form</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">simplify</emphasis></entry>
+ <entry>Computes a form supposed to be simpler</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">unfold</emphasis> [&sterm;]</entry>
+ <entry>δ-reduces the constant or variable if specified, or that
+ in head position</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">whd</emphasis></entry>
+ <entry>Computes the βδιζ-weak-head normal form</entry>
+ </row>
+ </tbody>
+ </tgroup>
+ </table>
+ </sect2>
+
+ <sect2 id="auto-params">
+ <title>auto-params</title>
+ <para>&TODO;</para>
+ <table frame="topbot" rowsep="0" colsep="0" role="grammar">
+ <title>reduction-kind</title>
+ <tgroup cols="4">
+ <tbody>
+ <row>
+ <entry id="grammar.autoparams">&autoparams;</entry>
+ <entry>::=</entry>
+ <entry><emphasis role="bold">depth=&nat;</emphasis></entry>
+ <entry>&TODO;</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">width=&nat;</emphasis></entry>
+ <entry>&TODO;</entry>
+ </row>
+ <row>
+ <entry/>
+ <entry>|</entry>
+ <entry><emphasis role="bold">&TODO;</emphasis></entry>
+ <entry>&TODO;</entry>
+ </row>
+ </tbody>
+ </tgroup>
+ </table>
+ </sect2>
+ </sect1>