Unique

Posted on 2017-05-26 by nbloomf
Tags: arithmetic-made-difficult, literate-haskell

This page is part of a series on Arithmetic Made Difficult.

This post is literate Haskell; you can load the source into GHCi and play along.


{-# LANGUAGE NoImplicitPrelude #-}
module Unique
  ( unique, _test_unique, main_unique
  ) where

import Testing
import Booleans
import Not
import And
import NaturalNumbers
import Lists
import ConsumingFold
import Snoc
import Reverse
import Map
import Range
import AllAndAny
import Filter
import Elt
import Sublist
import Select

Today we’ll introduce a boolean function \(\unique\) to detect whether or not a list has any duplicate items. As usual, we’d like to define \(\unique\) as a fold. The signature needs to be \[\lists{A} \rightarrow \bool.\] How can we do this? Intuitively, we might say

  1. \(\nil\) is unique, and
  2. \(\cons(a,x)\) is unique if \(x\) is unique and \(a\) does not appear in \(x\).

Note that \(\unique\) will need to “have it’s cake and eat it too”; that is, when testing \(\cons(a,x)\) for uniqueness we have to check that \(x\) is unique (eat the cake) and check that \(a\) does not appear in \(x\) (have the cake). We had a similar problem when we defined \(\tails\); the solution there was to use consuming fold, so that’s what we’ll do here.

Let \(A\) be a set. Define \(\sigma : A \times \lists{A} \times \bool \rightarrow \bool\) by \[\sigma(a,x,p) = \band(\bnot(\elt(a,x)),p).\] We then define \(\unique : \lists{A} \rightarrow \bool\) by \[\unique = \cfoldr(\btrue)(\sigma).\]

In Haskell:

Since \(\unique\) is defined as a \(\cfoldr(\ast)(\ast)\), it can be characterized as the unique solution to a system of functional equations.

Let \(A\) be a set. \(\unique\) is the unique map \(f : \lists{A} \rightarrow \bool\) satisfying the following system of equations for all \(a \in A\) and \(x \in \lists{A}\). \[\left\{\begin{array}{l} f(\nil) = \btrue \\ f(\cons(a,x)) = \band(\bnot(\elt(a,x)),f(x)) \end{array}\right.\]

Special cases.

Let \(A\) be a set with \(a,b \in A\). Then we have the following.

  1. \(\unique(\cons(a,\nil)) = \btrue\).
  2. \(\unique(\cons(a,\cons(b,\nil))) = \bnot(\beq(a,b))\).
  1. Note that \[\begin{eqnarray*} & & \unique(\cons(a,\nil)) \\ & = & \band(\bnot(\elt(a,\nil)),\unique(\nil)) \\ & = & \band(\bnot(\bfalse),\btrue) \\ & \href{/posts/arithmetic-made-difficult/Not.html#thm-not-false} = & \band(\btrue,\btrue) \\ & = & \btrue \end{eqnarray*}\] as claimed.
  2. Note that \[\begin{eqnarray*} & & \unique(\cons(a,\cons(b,\nil))) \\ & = & \band(\bnot(\elt(a,\cons(b,\nil))),\unique(\cons(b,\nil))) \\ & = & \band(\bnot(\beq(a,b)),\btrue) \\ & \href{/posts/arithmetic-made-difficult/And.html#thm-and-true-right} = & \bnot(\beq(a,b)) \end{eqnarray*}\] as claimed.

\(\unique\) is “down closed” with respect to \(\sublist\).

Let \(A\) be a set with \(x,y \in \lists{A}\). If \(\sublist(x,y) = \btrue\) and \(\unique(y) = \btrue\), then \(\unique(x) = \btrue\).

We proceed by list induction on \(y\). For the base case \(y = \nil\), suppose \(\sublist(x,y) = \btrue\). Then we have \(x = \nil\), and thus \(\unique(x) = \btrue\). For the inductive step, suppose the implication holds for all \(x\) for some \(y\) and let \(b \in A\). Suppose further that \(\unique(\cons(b,y)) = \btrue\) and \(\sublist(x,\cons(b,y)) = \btrue\). We consider two possibilities for \(x\). If \(x = \nil\), then \(\unique(x) = \btrue\) as needed. Suppose instead that \(x = \cons(a,u)\) for some \(a \in A\) and \(u \in \lists{A}\). Note first that since \(\unique(\cons(b,y)) = \btrue\), we have \[\band(\bnot(\elt(a,x)),\unique(y)) = \btrue;\] in particular, \(\unique(y) = \btrue\). We now consider two possibilities. If \(a \neq b\), we have \[\begin{eqnarray*} & & \btrue \\ & = & \sublist(x,\cons(b,y)) \\ & = & \sublist(\cons(a,u),\cons(b,y)) \\ & = & \sublist(\cons(a,u),y) \\ & = & \sublist(x,y). \end{eqnarray*}\] By the inductive hypothesis we have \(\unique(x) = \btrue\) as needed. Suppose instead that \(a = b\). Note that \(\bnot(\elt(a,y))\), so \(\bnot(\elt(a,u))\). Now we have \[\begin{eqnarray*} & & \btrue \\ & = & \sublist(x,\cons(b,y)) \\ & = & \sublist(\cons(a,u),\cons(b,y)) \\ & = & \sublist(u,y). \end{eqnarray*}\] Again by the inductive hypothesis we have \(\unique(u) = \btrue\). So \[\begin{eqnarray*} & & \unique(x) \\ & = & \unique(\cons(a,u)) \\ & = & \band(\bnot(\elt(a,u)),\unique(u)) \\ & = & \band(\btrue,\btrue) \\ & \href{/posts/arithmetic-made-difficult/And.html#thm-and-eval-true-true} = & \btrue \end{eqnarray*}\] as needed.

If \(f\) is injective, \(\map(f)\) preserves uniqueness.

Let \(A\) and \(B\) be sets with \(x \in \lists{A}\) and \(f : A \rightarrow B\). If \(f\) is injective then \(\unique(x) = \unique(\map(f)(x))\).

We proceed by list induction on \(x\). For the base case \(x = \nil\), note that \[\begin{eqnarray*} & & \unique(\map(f)(x)) \\ & = & \unique(\map(f)(\nil)) \\ & \href{/posts/arithmetic-made-difficult/Map.html#cor-map-nil} = & \unique(\nil) \\ & = & \unique(x) \end{eqnarray*}\] as needed. For the inductive step, suppose the implication holds for all \(f\) for some \(x\) and let \(a \in A\). Suppose further that \(\unique(\cons(a,x)) = \btrue\). Now if \(f\) is injective, we have \(\elt(f(a),\map(f)(x)) = \elt(a,x)\). Using the inductive hypothesis, we have \[\begin{eqnarray*} & & \unique(\map(f)(\cons(a,x))) \\ & \href{/posts/arithmetic-made-difficult/Map.html#cor-map-cons} = & \unique(\cons(f(a),\map(f)(x))) \\ & = & \band(\bnot(\elt(f(a),\map(f)(x))),\unique(\map(f)(x))) \\ & = & \band(\bnot(\elt(a,x)),\unique(x)) \\ & = & \unique(\cons(a,x)) \\ & = & \btrue \end{eqnarray*}\] as needed.

\(\unique\) and \(\filter\).

Let \(A\) be a set with \(x \in \lists{A}\) and \(p : A \rightarrow \bool\). If \(\unique(x) = \btrue\), then \(\unique(\filter(p,x)) = \btrue\).

Note that \(\sublist(\filter(p,x),x) = \btrue\); so we have \(\unique(\filter(p,x)) = \btrue\) as claimed.

\(\unique\) interacts with \(\snoc\).

Let \(A\) be a set with \(a \in A\) and \(x \in \lists{A}\). Then \[\unique(\snoc(a,x)) = \band(\bnot(\elt(a,x)),\unique(x)).\]

We proceed by list induction on \(x\). For the base case \(x = \nil\), note that \[\begin{eqnarray*} & & \unique(\snoc(a,\nil)) \\ & \href{/posts/arithmetic-made-difficult/Snoc.html#cor-snoc-nil} = & \unique(\cons(a,\nil)) \\ & = & \band(\bnot(\elt(a,\nil)),\unique(\nil)) \end{eqnarray*}\] as needed. For the inductive step, suppose the equality holds for all \(a\) for some \(x\) and let \(b \in A\). Using the inductive hypothesis, we have \[\begin{eqnarray*} & & \unique(\snoc(a,\cons(b,x))) \\ & \href{/posts/arithmetic-made-difficult/Snoc.html#cor-snoc-cons} = & \unique(\cons(b,\snoc(a,x))) \\ & = & \band(\bnot(\elt(b,\snoc(a,x))),\unique(\snoc(a,x))) \\ & \href{/posts/arithmetic-made-difficult/Elt.html#thm-elt-snoc} = & \band(\bnot(\bif{\beq(b,a)}{\btrue}{\elt(b,x)}),\unique(\snoc(a,x))) \\ & = & \band(\bnot(\elt(b,\cons(a,x))),\band(\bnot(\elt(a,x))),\unique(x)) \\ & = & \band(\bnot(\bif{\beq(b,a)}{\btrue}{\elt(b,x)}),\band(\bnot(\elt(a,x)),\unique(x))) \\ & = & \band(\bif{\beq(b,a)}{\bfalse}{\bnot(\elt(b,x))},\band(\bnot(\elt(a,x)),\unique(x))) \\ & = & \bif{\beq(a,b)}{\band(\bfalse,\band(\bnot(\elt(a,x)),\unique(x)))}{\band(\bnot(\elt(b,x)),\band(\bnot(\elt(a,x)),\unique(x)))} \\ & = & \bif{\beq(a,b)}{\bfalse}{\band(\band(\bnot(\elt(b,x)),\bnot(\elt(a,x))),\unique(x))} \\ & \href{/posts/arithmetic-made-difficult/And.html#thm-and-commutative} = & \bif{\beq(a,b)}{\bfalse}{\band(\band(\bnot(\elt(a,x)),\bnot(\elt(b,x))),\unique(x))} \\ & \href{/posts/arithmetic-made-difficult/And.html#thm-and-associative} = & \bif{\beq(a,b)}{\bfalse}{\band(\bnot(\elt(a,x)),\band(\bnot(\elt(b,x)),\unique(x)))} \\ & \href{/posts/arithmetic-made-difficult/And.html#thm-and-false-left} = & \bif{\beq(a,b)}{\band(\bfalse,\band(\bnot(\elt(b,x)),\unique(x)))}{\band(\bnot(\elt(a,x)),\band(\bnot(\elt(b,x)),\unique(x)))} \\ & = & \band(\bif{\beq(a,b)}{\bfalse}{\bnot(\elt(a,x))},\band(\bnot(\elt(b,x)),\unique(x))) \\ & = & \band(\bnot(\bif{\beq(a,b)}{\btrue}{\elt(a,x)}),\band(\bnot(\elt(b,x)),\unique(x))) \\ & \href{/posts/arithmetic-made-difficult/Elt.html#cor-elt-cons} = & \band(\bnot(\elt(a,\cons(b,x))),\band(\bnot(\elt(b,x)),\unique(x))) \\ & = & \band(\bnot(\elt(a,\cons(b,x))),\unique(\cons(b,x))) \end{eqnarray*}\] as needed.

\(\unique\) interacts with \(\rev\).

Let \(A\) be a set with \(x \in \lists{A}\). Then \(\unique(x) = \unique(\rev(x))\).

We proceed by list induction on \(x\). For the base case \(x = \nil\), note that \[\begin{eqnarray*} & & \unique(\rev(x)) \\ & = & \unique(\rev(\nil)) \\ & \href{/posts/arithmetic-made-difficult/Reverse.html#cor-rev-nil} = & \unique(\nil) \\ & = & \unique(x) \end{eqnarray*}\] as needed. For the inductive step, suppose the equality holds for some \(x\) and let \(a \in A\). Now \[\begin{eqnarray*} & & \unique(\rev(\cons(a,x))) \\ & \href{/posts/arithmetic-made-difficult/Reverse.html#cor-rev-cons} = & \unique(\snoc(a,\rev(x))) \\ & = & \band(\bnot(\elt(a,\rev(x))),\unique(\rev(x))) \\ & = & \band(\bnot(\elt(a,x)),\unique(x)) \\ & = & \unique(\cons(a,x)) \end{eqnarray*}\] as needed.

\(\range\)s are unique.

Let \(a,b \in \nats\). We have \[\unique(\range(a,b)) = \btrue.\]

We proceed by induction on \(b\). For the base case \(b = \zero\), we have \[\begin{eqnarray*} & & \unique(\range(a,\zero)) \\ & = & \unique(\nil) \\ & \href{/posts/arithmetic-made-difficult/Unique.html#cor-unique-nil} = & \btrue \end{eqnarray*}\] as needed. For the inductive step, suppose the equality holds for all \(a\) for some \(b\). Now \[\begin{eqnarray*} & & \unique(a,\next(b)) \\ & = & \unique(\cons(a,\range(\next(a),b))) \\ & = & \band(\bnot(\elt(a,\range(\next(a),b))),\unique(\range(\next(a),b))) \\ & \href{/posts/arithmetic-made-difficult/Unique.html#thm-unique-range} = & \band(\bnot(\elt(a,\range(\next(a),b))),\btrue) \\ & \href{/posts/arithmetic-made-difficult/And.html#thm-and-true-right} = & \bnot(\elt(a,\range(\next(a),b))) \\ & = & \bnot(\bfalse) \\ & \href{/posts/arithmetic-made-difficult/Not.html#thm-not-false} = & \btrue \end{eqnarray*}\]

\(\unique\) can detect \(\elt\).

Let \(A\) be a set. For all \(a \in A\) and \(x \in X\), we have \[\all(\unique)(\map(\cons(a,\cons(-,\nil)))(x)) = \bnot(\elt(a,x)).\]

We proceed by list induction on \(x\). For the base case \(x = \nil\), we have \[\begin{eqnarray*} & & \all(\unique)(\map(\cons(a,\cons(-,x)))(\nil)) \\ & = & \all(\unique)(\nil) \\ & = & \btrue \\ & \href{/posts/arithmetic-made-difficult/Not.html#thm-not-false} = & \bnot(\bfalse) \\ & = & \bnot(\elt(a,\nil)) \end{eqnarray*}\] as needed. For the inductive step, suppose the equality holds for all \(a\) for some \(x\), and let \(b \in A\). Now \[\begin{eqnarray*} & & \all(\unique)(\map(\cons(a,\cons(-,\nil)))(\cons(b,x))) \\ & = & \all(\unique)(\cons(\cons(a,\cons(b,\nil)),\map(\cons(a,\cons(-,\nil)))(x))) \\ & = & \band(\unique(\cons(a,\cons(b,\nil))),\all(\unique)(\map(\cons(a,\cons(-,\nil)))(x))) \\ & = & \band(\unique(\cons(a,\cons(b,\nil))),\bnot(\elt(a,x))) \\ & \href{/posts/arithmetic-made-difficult/Unique.html#thm-unique-two} = & \band(\bnot(\beq(a,b)),\bnot(\elt(a,x))) \\ & \href{/posts/arithmetic-made-difficult/Or.html#thm-demorgan-not-or} = & \bnot(\bor(\beq(a,b),\elt(a,x))) \\ & = & \bnot(\elt(a,\cons(b,x))) \end{eqnarray*}\] as needed.

\(\unique\) and \(\select\).

Let \(A\) be a set. For all \(x \in \lists{A}\) we have \[\unique(x) = \all(\unique)(\select(\next(\next(\zero)),x)).\]

We proceed by list induction on \(x\). For the base case \(x = \nil\), we have \[\begin{eqnarray*} & & \all(\unique)(\select(\next(\next(\zero)),x)) \\ & = & \all(\unique)(\select(\next(\next(\zero)),\nil)) \\ & = & \all(\unique)(\nil) \\ & = & \btrue \\ & \href{/posts/arithmetic-made-difficult/Unique.html#cor-unique-nil} = & \unique(\nil) \\ & = & \unique(x) \end{eqnarray*}\] as needed. For the inductive step, suppose the equality holds for some \(x\) and let \(a \in A\). Using the inductive hypothesis, we have \[\begin{eqnarray*} & & \all(\unique)(\select(\next(\next(\zero)),\cons(a,x))) \\ & = & \all(\unique)(\cat(\map(\cons(a,-))(\select(\next(\zero),x)),\select(\next(\next(\zero)),x))) \\ & = & \all(\unique)(\cat(\map(\cons(a,-))(\map(\cons(-,\nil))(x)),\select(\next(\next(\zero)),x))) \\ & = & \all(\unique)(\cat(\map(\cons(a,\cons(-,\nil)))(x),\select(\next(\next(\zero)),x))) \\ & = & \band(\all(\unique)(\map(\cons(a,\cons(-,\nil)))(x)),\all(\unique)(\select(\next(\next(\zero)),x))) \\ & = & \band(\all(\unique)(\map(\cons(a,\cons(-,\nil)))(x)),\unique(x)) \\ & = & \band(\bnot(\elt(a,x)),\unique(x)) \\ & = & \unique(\cons(a,x)) \end{eqnarray*}\] as needed.

Testing

Suite:

_test_unique ::
  ( TypeName a, Equal a, Show a, Arbitrary a, CoArbitrary a
  , TypeName (t a), List t, Arbitrary (t n), Show (t n), Equal (t n)
  , TypeName n, Natural n, Show n, Arbitrary n, Equal n
  , Equal (t a), Show (t a), Arbitrary (t a), Equal (t (t a))
  ) => Int -> Int -> t a -> n -> IO ()
_test_unique size cases t n = do
  testLabel1 "unique" t

  let args = testArgs size cases

  runTest args (_test_unique_nil t)
  runTest args (_test_unique_cons t)
  runTest args (_test_unique_single t)
  runTest args (_test_unique_double t)
  runTest args (_test_unique_sublist t)
  runTest args (_test_unique_filter t)
  runTest args (_test_unique_snoc t)
  runTest args (_test_unique_rev t)
  runTest args (_test_unique_range t n)
  runTest args (_test_unique_elt t)
  runTest args (_test_unique_select_two t n)

Main:

main_unique :: IO ()
main_unique = do
  _test_unique 20 100 (nil :: ConsList Bool)  (zero :: Unary)
  _test_unique 20 100 (nil :: ConsList Unary) (zero :: Unary)