alexandræ |
![]() ![]() | |||||||||||
|
boolean algebrasjanuary 1st, 2025happy new year everyone ! thought of making a video about boolean algebras at some point, and figured out these little results while messing with those : \(\textbf{Definition 1.1}~~\) A \(\textbf{partial order}\) \(\le\) on a set \(S\) is a binary relation, a subset of \(S\times S,\) which satisfies the following properties : - \(\textbf{Reflexivity}\) (i.e. includes equality) : for any \(a\in S,\) we have \(a\le a.\) - \(\textbf{Antisymmetry}:\) for any \(a,b\in S,\) if \(a\le b\) and \(b\le a,\) then \(a=b.\) - \(\textbf{Transitivity}:\) for any \(a,b,c\in S,\) if \(a\le b\) and \(b\le c,\) then \(a\le c.\) We call \((S,\le)\) a \(\textbf{poset}\) (partially ordered set). \(\textbf{Exemple 1.2}~~\) For any set \(S,\) we have \((S,=)\) a poset. \(\textit{Proof}~~\) \(=\) is reflexive by introduction of equality. If \(a=b\) and \(b=a,\) then \(a=b\) by conjunction elimination, so \(=\) is antisymmetric. Finally, if \(a=b\) and \(b=c,\) we can use elimination of equality to transform \(a=b\) into \(a=c\) by equality of \(b\) and \(c.\) \(\blacksquare\)\(\textbf{Definition 1.3}~~\) A partial order \(\le\) on a set \(S\) is a \(\textbf{total order}\) if it is \(\textbf{strongly connected}:\) in other words, for any \(a,b\in S,\) there's at least \(a\le b\) or \(b\le a.\) If \(\le\) is a total order on \(S,\) we call \((S,\le)\) a \(\textbf{toset}\) (totally ordered set). \(\textbf{Lemma 1.4}~~\) The \(\textbf{well-ordering principle}\) states that, given any set \(S,\) there exists a binary relation \(<\) on \(S\) which satisfies the following properties : - \(\textbf{Irreflexivity}:\) for all \(a\in S,\) we have \(\neg(a\lt a),\) which we can also write \(a\not\lt a.\) - \(\textbf{Asymmetry}:\) for all \(a,b\in S,\) if \(a\lt b,\) then \(b\not\lt a.\) - \(\textbf{Transitivity}:\) for any \(a,b,c\in S,\) if \(a\lt b\) and \(b\lt c,\) then \(a\lt c.\) - \(\textbf{Connected}:\) for any \(a,b\in S,\) either \(a=b,\) or \(a\lt b,\) or \(b\lt a.\) This lemma is equivalent to the axiom of choice in ZF set theory, and can therefore be used in ZFC set theory. \(\textbf{Theorem 1.5}\) Given any set \(S,\) there exists a total order \(\le\) on \(S.\) \(\textit{Proof}~~\) By applying the well-ordering principle on \(S,\) we get a binary relation \(\lt\) on \(S\) which is irreflexive, asymmetric, and transitive.\(\textbf{Definition 1.6}~~\) A poset \((S,\le)\) is \(\textbf{bounded}\) if there exist two elements \(\top,\bot\in S\) which satisfy the following : - \(\textbf{Top element}:\) for all \(x\in S,\) we have \(x\le\top.\) - \(\textbf{Bottom element}:\) for all \(x\in S,\) we have \(\bot\le x.\) \(\textbf{Example 1.7}~~\) \((\{0,1\},\le)\) and \((\{0,1,2,3\},\le)\) are bounded posets. \(\textit{Proof}~~\) For any \(x\in\{0,1\},\) either \(x=0\) and thus satisfies \(0\le x\le1,\) or \(x=1\) and thus satisfies \(0\le x\le1.\) Therefore, \((\{0,1\},\le)\) is a bounded poset : its top element is \(1,\) and its and bottom element is \(0.\)\(\textbf{Definition 1.8}~~\) Let \((S_1,\le_1)\) and \((S_2,\le_2)\) be posets. A \(\textbf{product poset}\) \((S_1,\le_1)\otimes(S_2,\le_2)\) is defined as \((S_1\times S_2,\le_1\otimes\le_2),\) where \(S_1\times S_2\) is simply the cartesian product of \(S_1\) with \(S_2,\) and \(\le_1\otimes\le_2\) is the binary relation such that, for all \((a_1,a_2),(b_1,b_2)\in S_1\times S_2,\) we have \((a_1,a_2)\le_1\otimes\le_2(b_1,b_2)\) if and only if both \(a_1\le_1b_1\) and \(a_2\le_2b_2.\) \(\textbf{Property 1.9}~~\) Given posets \((S_1,\le_1)\) and \((S_2,\le_2),\) the product poset \((S_1,\le_1)\otimes(S_2,\le_2)=:(S_1\times S_2,\le)\) is a poset. \(\textit{Proof}~~\) Let \(a\in S_1\times S_2.\) We have \(a_1\le_1a_1\) from reflexivity of \(\le_1,\) and \(a_2\le_2a_2\) from reflexivity of \(\le_2.\) Therefore, \(a\le a.\)\(\textbf{Proposition 1.10}~~\) The product of two bounded posets is bounded. \(\textit{Proof}~~\) Let \((S_1,\le_1)\) and \((S_2,\le_2)\) be two posets. Then, there exists \(\bot_1,\top_1\in S_1\) and \(\bot_2,\top_2\in S_2\) such that for all \(x\in S_1\times S_2,\) we have \(\bot_1\le_1x_1\le\top_1\) and \(\bot_2\le_2x_2\le\top_2,\) thus \((\bot_1,\bot_2)\le x\le(\top_1,\top_2).\) \(\blacksquare\)\(\textbf{Proposition 1.11}~~\) In a bounded poset \((S,\le),\) top and bottom elements are unique. \(\textit{Proof}~~\) Let \(\bot,\bot'\) be bottom elements of \((S,\le).\) Since \(\bot\) is bottom, we have \(\bot\le\bot'\;;\) however, \(\bot'\) is also bottom, so \(\bot'\le\bot\;;\) by antisymmetry, we conclude that \(\bot=\bot'.\)\(\textbf{Definition 2.1.a}~~\) Given a poset \((S,\le),\) let \(a,b\in S\) such that there exists \(x\in S\) which universally satisfies \(a\le x\) and \(b\le x,\) which means : - \(\textbf{Upper bound}\text{ (}\textit{cocone}\text):\) \(a\le x\) and \(b\le x.\) - \(\textbf{Colimit}\text{ (}\textit{least}\text{ness)}:\) for all \(y\in S\) such that also \(a\le y\) and \(b\le y,\) we have \(x\le y.\) Such a \(x\) is called a \(\textbf{join}\) of \(a\) and \(b.\) A join is also called a least upper bound, or supremum. \(\textbf{Definition 2.1.b}~~\) Given a poset \((S,\le),\) let \(a,b\in S\) such that there exists \(x\in S\) which universally satisfies \(x\le a\) and \(x\le b,\) which means : - \(\textbf{Lower bound}\text{ (}\textit{cone}\text):\) \(x\le a\) and \(x\le b.\) - \(\textbf{Limit}\text{ (}\textit{greatest}\text{ness)}:\) for all \(y\in S\) such that also \(y\le a\) and \(y\le a,\) we have \(y\le x.\) Such a \(x\) is called a \(\textbf{meet}\) of \(a\) and \(b.\) A meet is also called a greatest lower bound, or infimum. \(\textbf{Proposition 2.2}~~\) Given a poset \((S,\le),\) for any \(a,b\in S,\) there is at most one join \(a\lor b\in S,\) and at most one meet \(a\land b\in S,\) of \(a\) and \(b.\) \(\textit{Proof}~~\) Let \(a,b\in S\) which have a meet. If \(c_1\) and \(c_2\) are meets of \(a\) and \(b,\) we get \(c_1\le a\) and \(c_1\le b\) by lower bound property so \(c_1\le c_2\) by limit property, as well as \(c_2\le a\) and \(c_2\le b\) by lower bound property so \(c_2\le c_1\) by limit property ; therefore, \(c_1=c_2\) by antisymmetry.\(\textbf{Definition 2.3}~~\) Given a poset \((S,\le),\) if, for all \(a,b\in S,\) there exists a join between \(a\) and \(b,\) and a meet between \(a\) and \(b,\) we can unambiguously define \(\vee:S\times S\to S\) and \(\land:S\times S\to S\) such that \(\vee:(a,b)\mapsto a\vee b\) and \(\land:(a,b)\mapsto a\land b.\) Both \((S,\le)\) and \((S,\wedge,\vee)\) are called a \(\textbf{lattice}.\) \(\textbf{Proposition 2.4}~~\) Given a \((L,\le)\) lattice, if we define \(\mathcal R_1,\mathcal R_2\) binary relations on \(L\) such that : - for all \(a,b\in L,\) we have \(a~\mathcal R_1~b\) if and only if \(a\lor b=b,\) - for all \(a,b\in L,\) we have \(a~\mathcal R_2~b\) if and only if \(a\land b=a,\) then \(\mathcal R_1=\mathcal R_2={\le}.\) This establishes that \((L,\le)\) and \((L,\wedge,\vee)\) are essentially interchangeable. \(\textit{Proof}~~\) Let \(a,b\in L\) such that \(a~\mathcal R_1~b.\) By upper boundedness property, we have \(a\vee b\le b.\) By colimit property, \(a\le b.\) Therefore, \(\mathcal R_1\subset{\le}.\)\(\textbf{Proposition 2.5}~~\) The product of two lattices is a lattice. \(\textit{Proof}~~\) Let \((L_1,\land_1,\lor_1)\) and \((L_2,\land_1,\lor_2)\) be lattices, \(L=L_1\times L_2,\) maps \(\land:L\times L\to L\) and \(\lor:L\times L\to L\) such that, for all \(a,b\in L,\) we have \(a\land b=(a_1\land_1b_1,a_2\land_2b_2)\) and \(a\lor b=(a_1\lor_1b_1,a_2\lor_2b_2).\)\(\textbf{Definition 2.6}~~\) A \(\textbf{lattice homomorphism}\) \((L_1,\le_1)\) to \((L_2,\le_2)\) is a map \(\varphi:L_1\to L_2\) which : - Preserves joins : if \(a,b\in L_1,\) we have \(\varphi(a\lor b)=\varphi(a)\lor\varphi(b),\) - Preserves meets : if \(a,b\in L_1,\) we have \(\varphi(a\land b)=\varphi(a)\land\varphi(b).\) \(\textbf{Definition 2.7}~~\) We say two lattices \((L_1,\le_1)\) and \((L_2,\le_2)\) are \(\textbf{isomorphic}\) if there exist lattice homomorphisms \(\varphi:L_1\to L_2\) and \(\varphi':L_2\to L_1\) such that \(\varphi'(\varphi(x))=x\) for all \(x\in L_1,\) and \(\varphi(\varphi'(y))\) for all \(y\in L_2.\) More condensedly, this means the following diagram commutes : $$\begin{matrix} L_1&\!\!\!\!\!\!\!\!\begin{matrix}\overset\varphi\longrightarrow\\[-5pt]\underset{\varphi'}\longleftarrow\end{matrix}\!\!\!\!\!\!\!\!&L_2\\[-10pt] \underset{\text{id}_{L_1}}{\Large\circlearrowright}&&\underset{\text{id}_{L_2}}{\Large\circlearrowright} \end{matrix}$$ \(\textbf{Definition 3.1}~~\) Let \((L,\le)\) be a \(\underline{\rm bounded}\) lattice in which we can construct \({\Rightarrow}:L\times L\to L\) which, for all \(a,b\in L,\) satisfies : - \(\textbf{Modus ponens}\text{ (}\textit{cone}\text):\) \((a{\Rightarrow}b)\land a\le b.\) - \(\textbf{Limit}\text{ (}\textit{greatest}\text{ness)}:\) for all \(x\in L\) such that \(x\land a\le b,\) we have \(x\le a{\Rightarrow}b.\) We say \((L,\le)\) is a \(\textbf{Heyting algebra}.\) \(\textbf{Property 3.2}~~\) Given a \(\underline{\rm total}\) Heyting algebra \((H,\le),\) for all \(a,b\in H,\) we have : $$a{\Rightarrow}b=\left\{\begin{array}{ll}\top&\text{if }a\le b\\b&\text{if }b\lt a.\end{array}\right.$$ \(\textit{Proof}~~\) Let \(a,b\in H.\) Since \((H,\le)\) is a toset, we either have \(a\le b\) or \(b\lt a.\) If \(a\le b,\) \(\top\land a\le a\) by lower boundedness property, thus \(\top\land a\le b\) by transitivity. For all \(x\in H,\) we have \(x\le\top\) by \(\top\) being the top element. In conclusion, \(a{\Rightarrow}b=\top.\)\(\textbf{Proposition 3.3}~~\) The product of two Heyting algebras is Heyting. \(\textit{Proof}~~\) Let \((H_1,\le_1)\) and \((H_2,\le_2)\) be two Heyting algberas, with conditionals \({\Rightarrow}_1:H_1\times H_1\to H_1\) and \({\Rightarrow}_2:H_2\times H_2\to H_2.\) We define \((H,\le)\) as the product lattice \((H_1,\le_1)\otimes(H_2,\le_2),\) as well as \({\Rightarrow}:H\times H\to H\) such that, for all \(a,b\in H,\) we have \(a{\Rightarrow}b=(a_1{\Rightarrow_1}b_1,a_2{\Rightarrow_2}b_2).\)\(\textbf{Proposition 3.4}~~\) Given a Heyting algebra \((H,\le),\) the conditional \(\Rightarrow\) is unique. \(\textit{Proof}~~\) Let \({\Rightarrow}:H\times H\to H\) and \({\Rightarrow}':H\times H\to H\) be conditionals, and \(a,b\in H.\) By modus ponens, \((a{\Rightarrow}b)\land a\le b,\) therefore \(a{\Rightarrow}b\le a{\Rightarrow}'b\) by limit property. By modus ponens, \((a{\Rightarrow}'b)\land a\le b,\) therefore \(a{\Rightarrow}'b\le a{\Rightarrow}b\) by limit property. By asymmetry, we have \(a{\Rightarrow}b=a{\Rightarrow}'b,\) which demonstrates \({\Rightarrow}={\Rightarrow}'.\) \(\blacksquare\)\(\textbf{Definition 3.5}~~\) Given a Heyting algebra \((H,\le),\) we define negation as \(\neg:H\to H\) such that \(\neg:a\mapsto a{\Rightarrow}\bot.\) \(\textbf{Property 3.6}~~\) Given a \(\underline{\rm total}\) Heyting algebra \((H,\le),\) for all \(x\in H,\) we have : $$\neg x=\left\{\begin{array}{ll}\top&\text{if }x=\bot\\\bot&\rm otherwise.\end{array}\right.$$ \(\textit{Proof}~~\) Follows directly from property 3.2. \(\blacksquare\)\(\textbf{Definition 4.1}~~\) Let \((H,\le)\) be a Heyting algebra. We say \(\neg\) is \(\textbf{involutive}\) if, for all \(x\in H,\) we have \(\neg\neg x=x.\) A Heyting algebra whose negation is involutive is called a \(\textbf{Boolean algebra}.\) \(\textbf{Proposition 4.2}~~\) The product of two Boolean algebras is a Boolean algebra. \(\textit{Proof}~~\) Let \((B_1,\le_1)\) and \((B_2,\le_2)\) be Boolean algebras, and \((B,\le)\) the product Heyting algebra \((B_1,\le_1)\otimes(B_2,\le_2).\) For all \(x\in B,\) we have \(\neg_1\neg_1x_1=x_1\) and \(\neg_2\neg_2x_2=x_2,\) so by definition, \(\neg\neg x=(\neg_1\neg_1x_1,\neg_2\neg_2x_2)=(x_1,x_2)=x.\) This proves negation is involutive in \(B,\) which demonstrates that \((B,\le)\) is Boolean.\(\textbf{Proposition 4.3}~~\) A total Boolean algebra either has one or two elements. \(\textit{Proof}~~\) Let \((B,\le)\) be a Boolean algebra. Since it is Heyting, it's also bounded, so it has at least one element, which is its top/bottom element. For all \(x\in B,\) \(x\) belongs in the image of \(\neg\) as it can be expressed as a negation, that is, \(x=\neg(\neg x),\) thanks to \(\neg\) being involutive. However, we saw in property 3.5 that, in total Heyting algebras, negation's image is \(\{\bot,\top\}\). In summary, \(\varnothing\ne B\subset\{\bot,\top\},\) hence \(0\lt|B|\le2,\) so \(|B|=1\) or \(|B|=2,\) which proves our proposition. \(\blacksquare\)\(\textbf{Proposition 4.4}~~\) Finite Boolean algebras have cardinals \(1\) (trivial) or some positive even number. \(\textit{Proof}~~\) Let \((B,\le)\) be a Boolean algebra. By law of excluded middle, there either exists \(x\in B\) such that \(x=\neg x,\) or there doesn't. Let's assume there does. In that case, \(\neg x\land x=(x{\Rightarrow}\bot)\land x\le\bot\) by modus ponens, but since \(\bot\) is bottom, \(\neg x\land x=\bot.\) However, \(\neg x=x,\) so \(x\land x=\bot.\) By reflexivity, \(x\le x,\) so by property 2.4, \(x\land x=x.\) By transitivity of equality, \(x=\bot.\) Since \(\bot\le\bot,\) we have \(\neg\bot=\bot{\Rightarrow}\bot=\top,\) but from elimination of equality with \(x=\bot\) and \(\neg x=x,\) we obtain \(\neg\bot=\bot.\) By transitivity of equality, this yields \(\top=\bot=:0,\) therefore every \(x\in B\) is such that \(x\le0\) by top property, and \(0\le x\) by bottom property, which yields \(x=0\) by asymmetry. This shows us that \(B\subset\{0\}.\) Since \((B,\le)\) is Boolean, it's also bounded, and therefore non-empty, which yields \(B=\{0\},\) thus \(|B|=1.\)
| |||||||||||