Theory of conditionally complete lattices #
A conditionally complete lattice is a lattice in which every non-empty bounded subset s
has a least upper bound and a greatest lower bound, denoted below by sSup s
and sInf s
.
Typical examples are ℝ
, ℕ
, and ℤ
with their usual orders.
The theory is very comparable to the theory of complete lattices, except that suitable
boundedness and nonemptiness assumptions have to be added to most statements.
We express these using the BddAbove
and BddBelow
predicates, which we use to prove
most useful properties of sSup
and sInf
in conditionally complete lattices.
To differentiate the statements between complete lattices and conditionally complete
lattices, we prefix sInf
and sSup
in the statements by c
, giving csInf
and csSup
.
For instance, sInf_le
is a statement in complete lattices ensuring sInf s ≤ x
,
while csInf_le
is the same statement in conditionally complete lattices
with an additional assumption that s
is bounded below.
Alias of WithBot.sSup_empty
.
Equations
- One or more equations did not get rendered due to their size.
A complete lattice is a conditionally complete lattice, as there are no restrictions on the properties of sInf and sSup in a complete lattice.
Equations
- CompleteLattice.toConditionallyCompleteLattice = ConditionallyCompleteLattice.mk ⋯ ⋯ ⋯ ⋯
Equations
- CompleteLinearOrder.toConditionallyCompleteLinearOrderBot = ConditionallyCompleteLinearOrderBot.mk ⋯
Equations
Equations
- OrderDual.instConditionallyCompleteLinearOrder α = ConditionallyCompleteLinearOrder.mk ⋯ LinearOrder.decidableLE LinearOrder.decidableEq LinearOrder.decidableLT ⋯ ⋯
A greatest element of a set is the supremum of this set.
A least element of a set is the infimum of this set.
Alias of csSup_lowerBounds_eq_csInf
.
Alias of csInf_upperBounds_eq_csSup
.
Introduction rule to prove that b
is the supremum of s
: it suffices to check that b
is larger than all elements of s
, and that this is not the case of any w<b
.
See sSup_eq_of_forall_le_of_forall_lt_exists_gt
for a version in complete lattices.
Introduction rule to prove that b
is the infimum of s
: it suffices to check that b
is smaller than all elements of s
, and that this is not the case of any w>b
.
See sInf_eq_of_forall_ge_of_forall_gt_exists_lt
for a version in complete lattices.
b < sSup s
when there is an element a
in s
with b < a
, when s
is bounded above.
This is essentially an iff, except that the assumptions for the two implications are
slightly different (one needs boundedness above for one direction, nonemptiness and linear
order for the other one), so we formulate separately the two implications, contrary to
the CompleteLattice
case.
sInf s < b
when there is an element a
in s
with a < b
, when s
is bounded below.
This is essentially an iff, except that the assumptions for the two implications are
slightly different (one needs boundedness below for one direction, nonemptiness and linear
order for the other one), so we formulate separately the two implications, contrary to
the CompleteLattice
case.
If all elements of a nonempty set s
are less than or equal to all elements
of a nonempty set t
, then there exists an element between these sets.
The supremum of a singleton is the element of the singleton
The infimum of a singleton is the element of the singleton
If a set is bounded below and above, and nonempty, its infimum is less than or equal to its supremum.
The sSup
of a union of two sets is the max of the suprema of each subset, under the
assumptions that all sets are bounded above and nonempty.
The sInf
of a union of two sets is the min of the infima of each subset, under the assumptions
that all sets are bounded below and nonempty.
The supremum of an intersection of two sets is bounded by the minimum of the suprema of each set, if all sets are bounded above and nonempty.
The infimum of an intersection of two sets is bounded below by the maximum of the infima of each set, if all sets are bounded below and nonempty.
Introduction rule to prove that b
is the supremum of s
: it suffices to check that
b
is an upper bound- every other upper bound
b'
satisfiesb ≤ b'
.
Equations
- Pi.conditionallyCompleteLattice = ConditionallyCompleteLattice.mk ⋯ ⋯ ⋯ ⋯
When b < sSup s
, there is an element a
in s
with b < a
, if s
is nonempty and the order
is a linear order.
When sInf s < b
, there is an element a
in s
with a < b
, if s
is nonempty and the order
is a linear order.
When every element of a set s
is bounded by an element of a set t
, and conversely, then
s
and t
have the same supremum. This holds even when the sets may be empty or unbounded.
When every element of a set s
is bounded by an element of a set t
, and conversely, then
s
and t
have the same infimum. This holds even when the sets may be empty or unbounded.
Lemmas about a conditionally complete linear order with bottom element #
In this case we have Sup ∅ = ⊥
, so we can drop some Nonempty
/Set.Nonempty
assumptions.
In conditionally complete orders with a bottom element, the nonempty condition can be omitted
from csSup_le_iff
.
In conditionally complete orders with a bottom element, the nonempty condition can be omitted
from lt_csSup_iff
.
The sSup
of a non-empty set is its least upper bound for a conditionally
complete lattice with a top.
The sInf
of a bounded-below set is its greatest lower bound for a conditionally
complete lattice with a top.
Equations
- WithTop.instCompleteLinearOrder = CompleteLinearOrder.mk ⋯ ⋯ ⋯ ⋯ ⋯ LinearOrder.decidableLE LinearOrder.decidableEq LinearOrder.decidableLT
A version of WithTop.coe_sSup'
with a more convenient but less general statement.
A version of WithTop.coe_sInf'
with a more convenient but less general statement.
A monotone function into a conditionally complete lattice preserves the ordering properties of
sSup
and sInf
.
Supremum/infimum of Set.image2
#
A collection of lemmas showing what happens to the suprema/infima of s
and t
when mapped under
a binary function whose partial evaluations are lower/upper adjoints of Galois connections.
Complete lattice structure on WithTop (WithBot α)
#
If α
is a ConditionallyCompleteLattice
, then we show that WithTop α
and WithBot α
also inherit the structure of conditionally complete lattices. Furthermore, we show
that WithTop (WithBot α)
and WithBot (WithTop α)
naturally inherit the structure of a
complete lattice. Note that for α
a conditionally complete lattice, sSup
and sInf
both return
junk values for sets which are empty or unbounded. The extension of sSup
to WithTop α
fixes
the unboundedness problem and the extension to WithBot α
fixes the problem with
the empty set.
This result can be used to show that the extended reals [-∞, ∞]
are a complete linear order.
Adding a top element to a conditionally complete lattice gives a conditionally complete lattice
Equations
- WithTop.conditionallyCompleteLattice = ConditionallyCompleteLattice.mk ⋯ ⋯ ⋯ ⋯
Adding a bottom element to a conditionally complete lattice gives a conditionally complete lattice
Equations
- WithBot.conditionallyCompleteLattice = ConditionallyCompleteLattice.mk ⋯ ⋯ ⋯ ⋯
Equations
- WithTop.WithBot.completeLattice = CompleteLattice.mk ⋯ ⋯ ⋯ ⋯ ⋯ ⋯
Equations
- WithTop.WithBot.completeLinearOrder = CompleteLinearOrder.mk ⋯ ⋯ ⋯ ⋯ ⋯ LinearOrder.decidableLE LinearOrder.decidableEq LinearOrder.decidableLT
Equations
- WithBot.WithTop.completeLattice = CompleteLattice.mk ⋯ ⋯ ⋯ ⋯ ⋯ ⋯
Equations
- WithBot.WithTop.completeLinearOrder = CompleteLinearOrder.mk ⋯ ⋯ ⋯ ⋯ ⋯ LinearOrder.decidableLE LinearOrder.decidableEq LinearOrder.decidableLT