Classical and quantum logic

▶ Classical logic is the familiar logic used in math and computer

science.
▶ It has two truth values, TRUE and FALSE (usually written 1
and 0), and we can define the operations AND, OR and NOT
in terms of these truth values and nothing else.
▶ There are various nonclassical logics which obey different laws
than classical logic: for instance, intuitionistic, fuzzy and
modal logic.
▶ Quantum logic is a nonclassical logic designed to model
observations of quantum systems.
▶ It was originally proposed by Garrett Birkhoff and John von
Neumann in their 1936 paper “The Logic of Quantum
Mechanics”, and revised by later logicians and physicists.
Classical logic

▶ Here are the definitions of AND (∧), OR (∨) and NOT (¬) in
classical logic:
∧ 1 0 ∨ 1 0 ¬
1 1 0 1 1 1 1 0
0 0 0 0 1 0 0 1
▶ Given these definitions, we can prove theorems of classical
logic, which we can write as equations:

α ∨ ¬α = 1 (Excluded middle)
α ∧ ¬α = 0 (Noncontradiction)
α = ¬¬α (Double negation)
α ∧ (β ∨ γ) = (α ∧ β) ∨ (α ∧ γ) (Distribution)
Set algebra

▶ Now consider the set {a, b, c}. Its subsets are {a, b, c}, {a, b},
{a, c}, {b, c}, {a}, {b}, {c} and the empty set ∅.
▶ It’s straightforward to define operations on any of these
subsets:
▶ The intersection of α and β, α ∩ β, is the set of all things
in both α and β.
▶ Their union, α ∪ β, is the set of all things in either α or β.
▶ The complement of α, α∁ , is the set of all members of
{a, b, c} which are not in α.
Set algebra

▶ But there’s another way of defining these operations which is

very useful! We use the idea of an order on the subsets of
{a, b, c}.
▶ The most natural order is the one we get by using the relation
‘is a subset of’ (⊆).
▶ For instance, {a} ⊆ {a, b}, {a, b} ⊆ {a, b, c}, every set is a
subset of itself and ∅ is a subset of every set.
Set algebra
▶ We can draw this as follows:

{a, b, c}

{a, b} {a, c} {b, c}

{a} {b} {c}

∅
▶ In this picture, there is a path going upwards from a set α to
another set β if and only if α ⊆ β.
Set algebra

▶ Recall that a lower bound of α and β, given an order ⊆, is any

γ with γ ⊆ α and γ ⊆ β.
▶ Their greatest lower bound glb {α, β} is a lower bound such
that if φ is any other lower bound, φ ⊆ glb {α, β}.
▶ An upper bound of α and β is any γ with α ⊆ γ and β ⊆ γ.
▶ Their least upper bound lub {α, β} is an upper bound such
that if φ is any other upper bound, lub {α, β} ⊆ φ.
Set algebra

{a, b, c}

{a, b} {a, c} {b, c}

{a} {b} {c}

∅
▶ We can now define the set operations in terms of our order:
▶ α ∩ β is the greatest lower bound of α and β.
▶ α ∪ β is their least upper bound.
▶ α∁ is the subset of {a, b, c} such that α ∩ α∁ = ∅ (α and
α∁ are mutually exclusive) and α ∪ α∁ = {a, b, c} (they
are jointly exhaustive).
Boolean algebra
▶ Now look at this picture:
1

0
▶ Let’s compare it to our truth tables for classical logic:

∧ 1 0 ∨ 1 0 ¬
1 1 0 1 1 1 1 0
0 0 0 0 1 0 0 1
▶ If we read
▶ α ∧ β as the greatest lower bound of α and β,
▶ α ∨ β as their least upper bound,
▶ and ¬α as the member of {1, 0} such that α ∧ ¬α = 0
and α ∨ ¬α = 1,
▶ we can read off the truth tables from the picture.
Boolean algebra

▶ So there’s some kind of structure in common between classical

logic and our algebra of subsets of a set!
▶ This structure is known as a Boolean algebra. The two
Boolean algebras we’ve seen are the subset algebra for a set
with three members, and the two-element Boolean algebra,
which is a representation of classical logic.
▶ Any finite Boolean algebra can be drawn as a picture like the
ones we’ve seen.
Boolean algebra
▶ A Boolean algebra is defined by these properties:
(Lattice) It’s a set B with an order ⊆ defined on it so that
each pair of elements has both a greatest lower bound
and a least upper bound (as we saw, this is all we need to
define ∩ and ∪).
(Complementation) There is a greatest element G and a
least element L, and for any member α of B, there’s a
member α∁ of B with α ∩ α∁ = L and α ∪ α∁ = G .
(In the set algebra we saw, G is {a, b, c} and L is ∅; in
the two-element Boolean algebra, G is 1 and L is 0.)
(Distributivity) For each α, β and γ in B,
α ∩ (β ∪ γ) = (α ∩ β) ∪ (α ∩ γ).
▶ So the law of distribution we saw from classical logic
[α ∧ (β ∨ γ) = (α ∧ β) ∨ (α ∧ γ)] is actually one of the
defining properties of a Boolean algebra!
Theorem

▶ An equation of Boolean algebra (such as α ∪ α∁ = G ) is true

in all Boolean algebras if and only if it is true in the
two-element Boolean algebra.
▶ (The proof is in Givant and Halmos, Introduction to Boolean
Algebras, pp. 129-132.)
▶ This means that classical logic characterizes Boolean algebras
in general.
Quantum logic

▶ Quantum theory represents physical systems using the

mathematical structure of Hilbert spaces.
▶ Can we similarly find a logic to characterize the Hilbert space
formalism?
▶ Birkhoff and von Neumann said that we can!
▶ To explain how this works, we need to review some concepts
from linear algebra.
Vector spaces
▶ A vector space is a set V of objects, called vectors, together
with a set F of numbers, called scalars (technically a field :
usually R or C), with operations of vector addition and
scalar multiplication defined on V such that for all u, v , w in
V and all a, b in F:
Commutativity u + v = v + u
Associativity (u + v ) + w = u + (v + w ) and (ab)v = a(bv )
Distributivity a(u + v ) = au + av and (a + b)v = av + bv
Multiplicative identity 1v = v , where 1 is the multiplicative
identity of F (i.e., the number 1 in R and C)
▶ and such that
Additive identity There’s a member 0 of V so that
v + 0 = v for any v in V
Additive inverse For each v in V , there’s a member −v of V
with v + (−v ) = 0
Vector spaces
▶ Familiar examples of vector spaces:
▶ The vector spaces Rn and Cn of lists of real or complex
numbers of length n (e.g. the plane R2 , three-dimensional
space R3 , etc.)
▶ The sets R∞ and C∞ of infinite sequences of real or
complex numbers
▶ The set of differentiable real-valued functions on R
▶ The set Fm×n of m × n matrices with entries in F
▶ A subspace of a vector space V is a subset of V which is itself
a vector space.
▶ The span of a set of vectors α in a vector space V , span(α), is
the smallest subspace of V which contains every vector in α.
▶ (Equivalently, span(α) is the set of all linear combinations of
vectors in α, i.e. the set of all vectors which can be obtained
by adding multiples of vectors in α.)
Inner product spaces
▶ An inner product space is a vector space V where F is either
R or C and where there is a function taking ordered pairs
(u, v ) of vectors in V to a scalar ⟨u, v ⟩, called the inner
product of u and v .
▶ In addition, f satisfies the following properties for all u, v , w in
V and all a in F (I’m including these for the sake of
completeness but don’t worry too much about them!):
Positivity ⟨v , v ⟩ ≥ 0
Definiteness ⟨v , v ⟩ = 0 if and only if v = 0
Additivity ⟨u + v , w ⟩ = ⟨u, w ⟩ + ⟨v , w ⟩
Homogeneity ⟨au, v ⟩ = a⟨u, v ⟩
Conjugate symmetry ⟨u, v ⟩ equals the complex conjugate of
⟨v , u⟩
▶ Using the inner product, we can define the norm ∥v ∥ of a
p
vector v by ∥v ∥ = ⟨v , v ⟩.
Inner product spaces

▶ The most familiar inner product is the Euclidean dot product

in Rn , where if u = (u1 , . . . , un ) and v = (v1 , . . . , vn ),
⟨u, v ⟩ = u1 v1q
+ · · · + un vn . This gives rise to the Euclidean
norm ∥v ∥ = v12 + · · · + vn2 .
▶ The Euclidean norm intuitively represents the length of a
vector, and the Euclidean norm of a difference ∥v − u∥
represents the distance between the vectors v and u.
▶ A pair of vectors u, v in V are orthogonal if ⟨u, v ⟩ = 0.
Hilbert space

▶ A Hilbert space is an inner product space V where every

Cauchy sequence converges, given the norm defined on V .
▶ That is, if there’s a sequence of vectors v1 , v2 , . . . in V where
if you go far enough in the sequence, the vectors get arbitrarily
close to each other (∥vm − vn ∥ → 0), then the sequence has a
limit u in V (∥vn − u∥ → 0).
▶ Don’t worry too much about this if you haven’t studied real
analysis: this is just a way of saying that the space is
“complete” in the way R is.
Back to quantum logic

▶ The basic mathematical idea of quantum mechanics is to

associate a Hilbert space H(S) with the set of possible states
of a physical system S.
▶ Birkhoff and von Neumann’s idea was to construct an algebra
similar to a set algebra which could characterize these Hilbert
spaces.
▶ Instead of constructing our algebra from the set of subsets of
the set H(S), we’ll use the set of closed subspaces of H(S).
▶ A subspace α of H(S) is closed if it contains its limit points
(i.e., if a sequence of vectors in α converges to a limit v in
H(S), v is in α).
▶ Our greatest element G will be H(S) itself, and the least
element L will be the space {0} containing only the zero
vector.
Quantum logic
▶ We now associate each statement about the quantum system
(such as ‘the magnitude m has the value r in the system S’)
with a subspace of H(S).
▶ Let’s define the operation ∧ (greatest lower bound) as
intersection, just as we did for sets. This is possible because
the intersection of closed subspaces α, β of H(S) is also a
closed subspace.
▶ But we can’t define ∨ (least upper bound) as union, since the
union of two subspaces α and β of a vector space V is itself a
vector space if and only if α ⊆ β or β ⊆ α (this can easily be
proved from the vector space axioms).
▶ Instead, we’ll define α ∨ β as span(α ∪ β).
▶ Finally, instead of defining ¬α as α∁ , we’ll use the orthogonal
complement α⊥ : the set of all vectors in H(S) which are
orthogonal to every vector in α.
Distribution fails

▶ Consider a one-particle system S where we have measured the

position of the particle. Let α be the subspace of H(S)
corresponding to the statement ‘the position of the particle is
r ’.
▶ Let β1 , . . . , βn be subspaces corresponding to the statements
‘the momentum of the particle is q’ for each possible
momentum q.
▶ Then the statement corresponding to α ∧ (β1 ∨ · · · ∨ βn ) is
true, but each of (α ∧ β1 ), . . . , (α ∧ βn ) is false, due to
Heisenberg uncertainty, so that (α ∧ β1 ) ∨ · · · ∨ (α ∧ βn ) is
also false.
▶ This shows that the law of distribution
[α ∧ (β ∨ γ) = (α ∧ β) ∨ (α ∧ γ)] fails in our algebra.
Characterizing quantum logic
▶ We can characterize the algebraic system corresponding to
H(S) as an ortholattice O:
▶ O is a lattice with a greatest and least element.
▶ Each element α of O has an orthocomplement α⊥
satisfying:
Double negation α = α⊥⊥
Contraposition If α ⊆ β, then β ⊥ ⊆ α⊥
Noncontradiction α ∧ α⊥ = L.
▶ If we add the additional law
Orthomodularity α ∨ (α⊥ ∧ (α ∨ β)) = α ∨ β,
▶ which is a weakened version of distributivity, we obtain an
orthomodular lattice.
▶ Our system of closed subspaces of a Hilbert space is both an
ortholattice and an orthomodular lattice.
▶ An ortholattice satisfying distributivity is a Boolean algebra.
Why care?
Mathematical reasons: The algebraic structure of ortholattices is
interesting in itself, and fun to prove theorems about.
Philosophical reasons: The philosopher Hilary Putnam thought
that the failure of the law of distribution in quantum logic
shows that it’s not really a logical law. So instead of holding
onto classical logic, we should switch to quantum logic.
He argued that replacing classical with quantum logic would
explain puzzling features of quantum systems, like the
uncertainty principle and other complementarity phenomena.
Physical reasons: Quantum logic is a formalism that lets us
describe features of physical systems in a very general way.
Thus, it gives us a new setting for proving physical theorems.
A major theorem which arose from research in quantum logic
is Gleason’s theorem: “A quantum state is completely
determined by only knowing the answers to all of the possible
yes/no questions” (nLab).
References

Boolean algebra
Arnold - Logic and Boolean Algebra (1962) – this is a friendly but
rigorous introduction to BAs.
Givant and Halmos - Introduction to Boolean Algebras (2010) –
more in-depth but still accessible.
Hilbert spaces
Berberian - Introduction to Hilbert Space (1961) – I used this
book as a reference, it’s accessible and only requires linear
algebra.
Axler - Linear Algebra Done Right (2015) – this was my reference
for linear algebra material, but everything I discussed will be in
any standard text.
References

Quantum logic
Megill, Quantum Logic Explorer,
https://us.metamath.org/qleuni/mmql.html – the best
intro to quantum logic. This page introduces quantum logic
and hosts a huge collection of proofs about orthomodular
lattices, none of which require any knowledge of advanced
math!
Dalla Chiara and Giuntini - ‘Quantum Logics’ (2002) – an article
giving a general overview of the history of quantum logic and
some of its main results.
Birkhoff and von Neumann - ‘The Logic of Quantum Mechanics’
(1936) – the original paper that established the field.
References

Putnam - ‘Is Logic Empirical?’ (1968) – influential philosophical

discussion of the interpretation of quantum logic.
Dummett - ‘Is Logic Empirical?’ (1976) – a reply to Putnam.
Beran – Orthomodular Lattices: Algebraic Approach (1985) – an
introduction to the theory of quantum logic and orthomodular
lattices from an algebraic perspective.