The logic of quantum mechanics

The evaluation map \epsilon^{ \relax{S}} satisfies the following properties :

and

If (\sigma_1\sqcap_{{}_{{ \relax{S}}}} \sigma_2) and (\sigma'_1\sqcap_{{}_{{ \relax{S}}}} \sigma'_2) admit no common upper-bound in { \relax{S}}, then { \relax{l}}_{{}_{(\sigma_1,\sigma'_1)}} and { \relax{l}}_{{}_{(\sigma_2,\sigma'_2)}} admit a common upper-bound in { \relax{E}}_{ \relax{S}} which is { \relax{l}}_{{}_{(\sigma_1\sqcap_{{}_{{ \relax{S}}}} \sigma_2,\sigma'_1\sqcap_{{}_{{ \relax{S}}}} \sigma'_2)}}. As a consequence, the supremum ({ \relax{l}}_{{}_{(\sigma_1,\sigma'_1)}} \sqcup_{{}_{{ \relax{E}}_{ \relax{S}}}} { \relax{l}}_{{}_{(\sigma_2,\sigma'_2)}}) exists in { \relax{E}}_{ \relax{S}} and is given by

This result can be extended to the whole set of effects as long as we adopt the following conventions

The left-component f of a Chu morphism from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}) satisfies

The right-component f^{\ast} of a Chu morphism from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}) satisfies

The composition of a morphism (f ,f^{\ast}) from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}) by another morphism (g ,g^{\ast}) defined from ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}) to ({ \relax{S}}_C,{ \relax{E}}_{{ \relax{S}}_C},\epsilon^{{ \relax{S}}_C}) is given by ( g\circ f ,f^{\ast} \circ g^{\ast}) defining a valid morphism from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_C,{ \relax{E}}_{{ \relax{S}}_C},\epsilon^{{ \relax{S}}_C}).

Let us consider a states/effects Chu space ({ \relax{S}},{ \relax{E}}_{{ \relax{S}}},\epsilon^{{ \relax{S}}}). Let us consider a map (A : { \relax{S}}\longrightarrow { \relax{B}}, \sigma \mapsto { \relax{a}}_{\sigma}) satisfying

Then, we have

For any map f:{ \relax{S}}_A \longrightarrow { \relax{S}}_B satisfying

there exists a unique map f^\ast: { \relax{E}}_{ \relax{S}_B} \longrightarrow { \relax{E}}_{ \relax{S}_A} such that (f,f^{\ast}) is a morphism from the States/Effects Chu space ({ \relax{S}}_{A},{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to another States/Effects Chu space ({ \relax{S}}_{B},{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}).

Let us consider a states/effects Chu space ({ \relax{S}},{ \relax{E}}_{{ \relax{S}}},\epsilon^{{ \relax{S}}}). Let us consider a map (B : { \relax{E}}_{{ \relax{S}}}\longrightarrow { \relax{B}},{ \relax{l}}\mapsto { \relax{b}}_{{ \relax{l}}}) satisfying

Then, we have

For any map f^\ast:{ \relax{E}}_{ \relax{S}_B} \longrightarrow { \relax{E}}_{{ \relax{S}}_A} satisfying

there exists a unique map f:{ \relax{S}}_A \longrightarrow { \relax{S}}_B such that (f,f^{\ast}) is a morphism from the States/Effects Chu space ({ \relax{S}}_{A},{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to the States/Effects Chu space ({ \relax{S}}_{B},{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}).

The infimum of a morphism (f ,f^{\ast}) from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}) with another morphism (g ,g^{\ast}) defined from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}) defines a valid morphism from ({ \relax{S}}_A,{ \relax{E}}_{{ \relax{S}}_A},\epsilon^{{ \relax{S}}_A}) to ({ \relax{S}}_B,{ \relax{E}}_{{ \relax{S}}_B},\epsilon^{{ \relax{S}}_B}).

The space of measurements {\relax{M}}_{{}_{{ \relax{S}}}} is isomorphic to the space of effects { \relax{E}}_{ \relax{S}}.

The operator defined on {\mathcal{P}}({}{\relax{S}}) which maps S to (S^{\underline{\perp}})^{\underline{\perp}} (which will eventually be denoted by S^{{\underline{\perp}}{\underline{\perp}}}) owns the following properties

We have moreover

and then

\varnothing and (\uparrow^{{}^{{}{\relax{S}}}}\!\!\!\bot_{{}_{{\relax{S}}}}) are both elements of { \relax{H}}({}{\relax{S}}, {\underline{\perp}}).

Equipped with the inclusion \subseteq, { \relax{H}}({}{\relax{S}}, {\underline{\perp}}) is a partially ordered set. With the intersection \cap, { \relax{H}}({}{\relax{S}}, {\underline{\perp}}) is a down complete Inf semi-lattice.

As a consequence of property (r@ortho3=orthoortho3=ortho), we have that, for any S orthoclosed subset of {}{\relax{S}}, there exists a subset S' of {}{\relax{S}} such that (S')^{\underline{\perp}}=S, and conversely, any subset of the form S^{\underline{\perp}} is orthoclosed.
As a conclusion, the image of { \mathcal{P}}({}{\relax{S}}) by the orthogonality operator is exactly { \relax{H}}({}{\relax{S}}, {\underline{\perp}}).

The set of orthoclosed subsets satisfies the following properties :

For any S\subseteq {}{\relax{S}}, we define U_S as the set of maps from S to {\overline{{ \relax{S}}}} mapping any \sigma\in S to an element of \Theta^{\overline{{ \relax{S}}}}(\sigma). In other words,

We then define, for any S\subseteq {}{\relax{S}}, an element V_S in \widecheck{ \relax{S}}\cup \{\centerdot\} by

Endly, we have

with the obvious convention (\uparrow^{{}^{{}{\relax{S}}}}\!\!\centerdot) :=\varnothing.

For any S\subseteq {}{\relax{S}}, we define U_S as in r@defUSdefUS, and we have

We have the following ortho-complementation property :

{ \relax{H}}({}{\relax{S}}, {\underline{\perp}}) is a complete ortho-lattice.

The set { \relax{H}}({}{\relax{S}}, {\underline{\perp}}) of orthoclosed subsets of {}{\relax{S}} is explicitly described in terms of \langle {}{\relax{S}}\rangle, as follows :

The action of the orthogonality operator on the elements of { \relax{H}}({}{\relax{S}}, {\underline{\perp}}) is explicitly given by

Real measurements (as defined in previous definition) are obviously real morphisms from { \relax{S}} to { \relax{B}}.

The space of states { \relax{S}} equipped with its real structure (\overline{ \relax{S}},\star) is a deterministic space of states iff

(we note that necessarily U_\sigma=\underline{\sigma}_{{}_{\overline{ \relax{S}}}})

(\overline{ \relax{S}},\star) is deterministic iff (\overline{ \relax{S}}{}^{{}^{pure}},\underline{\perp}) is completely reducible, i.e. iff

Here, the operator (\cdot)^{\underline{\perp}} designates the operator from \mathcal{P}(\overline{ \relax{S}}{}^{{}^{pure}}) to itself which maps A to \{\sigma\in \overline{ \relax{S}}{}^{{}^{pure}}\;\vert\; \forall \sigma'\in A, \sigma\underline{\perp}\sigma'\}.

We have also obtained the following characterization

Let us suppose that \overline{ \relax{S}} is a simplex space of states and let \star be defined by

Then, (\overline{ \relax{S}}, \star) is a real structure of \overline{ \relax{S}}.
Conversely, let us consider { \relax{S}} a space of states which admits a real structure (\overline{ \relax{S}}, \star). Let us suppose that \overline{ \relax{S}} is a simplex space of states. Then, we have necessarily { \relax{S}}=\overline{ \relax{S}}.

A simplex space of states is necessarily distributive as an Inf semi-lattice.

If (\overline{ \relax{S}},\star) is completely indeterministic then (\overline{ \relax{S}}{}^{{}^{pure}},\underline{\perp}) is irreducible, i.e.

Here again, the operator (\cdot)^{\underline{\perp}} designates the operator from \mathcal{P}(\overline{ \relax{S}}{}^{{}^{pure}}) to itself which maps A to \{\sigma\in \overline{ \relax{S}}{}^{{}^{pure}}\;\vert\; \forall \sigma'\in A, \sigma\underline{\perp}\sigma'\}.

{\widehat{{\mathcal{P}}(\overline{{ \relax{S}}})}} is a lower subset (i.e. it is downward closed) of ({\mathcal{P}}(\overline{{ \relax{S}}}),\sqsubseteq) containing every elements of the form \{\sigma\} for \sigma\in \overline{{ \relax{S}}}.

The pair (\Theta^{\overline{{ \relax{S}}}},\Lambda^{\overline{{ \relax{S}}}}) is a Galois connection, i.e.

We have as usual

Note that this condition is equivalent to the following one

The pre-order \sqsubseteq defined on { \mathcal{P}}(\overline{{ \relax{S}}}) is in fact a partial order on { \relax{J}}_{\overline{{ \relax{S}}}}.

The infimum of two elements J and J' of { \relax{J}}_{\overline{{ \relax{S}}}}, denoted J\sqcap_{{}_{{ \relax{J}}_{\overline{{ \relax{S}}}}}} J' is given as follows

Let I and I' be two elements of { \relax{J}}_{\overline{{ \relax{S}}}} such that there exists I''\in { \relax{J}}_{\overline{{ \relax{S}}}} with I\sqsubseteq I'' and I'\sqsubseteq I''.
The supremum of I and I' in {{ \relax{J}}_{\overline{{ \relax{S}}}}} (denoted I\sqcup_{{}_{{ \relax{J}}_{\overline{{ \relax{S}}}}}} I') exists and is simply given by

The map \Theta^{\overline{{ \relax{S}}}} is an isomorphism from { \relax{S}} to { \relax{J}}_{\overline{{ \relax{S}}}}.

\{\,J\in { \relax{J}}_{\overline{{ \relax{S}}}} \;\vert\; Card(J)=1\,\} is isomorphic to \overline{{ \relax{S}}}. In other words,

and for any \sigma,\sigma'\in {\overline{{ \relax{S}}}}, we have

We have necessarily, for any I subset of \overline{{ \relax{S}}}

Let us denote by U:=\Theta^{\overline{ \relax{S}}}(\xi)\in { \relax{J}}_{\overline{{ \relax{S}}}}\;\; for any \xi\in { \relax{S}}. We then have

We will assume that ({\relax{S}}',{ \mathcal{Q}}({ \relax{S}}'),cl^{{{ \relax{S}}'}},\star) is a real space of states.
We define { \relax{J}}_{{{ \relax{S}}'}} to be the set cl^{{{ \relax{S}}'}}({ \mathcal{Q}}({ \relax{S}}')). The Inf semi-lattice structure being given by (r@RECdefJsqbarJRECdefJsqbarJ). We define the generalized space of effects as in Definition r@defgeneralizedeffectspacedefgeneralizedeffectspace (it is necessary to use the formula for the suprema established in (r@supremaJsupremaJ) ). The reduced space of effects is defined as in (r@reducedspaceofeffectsprimereducedspaceofeffectsprime). Then, we define an evaluation map { \epsilon}^{{ \relax{J}}_{{{ \relax{S}}'}}} on the whole { \relax{J}}_{{{ \relax{S}}'}} as in (r@defepsilonSdefepsilonS).
As a result, { \relax{J}}_{{{ \relax{S}}'}} is a well defined space of states which admits the real structure ({{{ \relax{S}}'}},\star). The elements of {{{ \relax{S}}'}} are the real states, whereas the elements of { \relax{J}}_{{{ \relax{S}}'}}\smallsetminus {{{ \relax{S}}'}} are the hidden states.

Let ({\relax{S}}',{ \mathcal{Q}}({ \relax{S}}'),cl^{{{ \relax{S}}'}},\star) be a real space of states. We will assume that {\relax{S}}' is a simplex space of states. Then, { \relax{J}}_{{{ \relax{S}}'}}={\relax{S}}'.

{\relax{c}}^{{ \relax{S}}'} is a pre-closure. In other words,

We have

cl_c^{{ \relax{S}}'} satisfies

Let us now suppose that { \relax{S}}' satisfies the Finite Rank Condition (Definition r@FiniteRankConditionFiniteRankCondition).
Then, cl_c^{{ \relax{S}}'} satisfies the idempotency property

and the following algebraicity property

{\mathcal{Q}}_c({ \relax{S}}') satisfies the property :

{\mathcal{Q}}_c({ \relax{S}}') satisfies

{\mathcal{Q}}_c({ \relax{S}}') satisfies

In other words, we have

({\relax{S}}',{ \mathcal{Q}}_c({ \relax{S}}'),cl_c^{{{ \relax{S}}'}},\star) satisfies the axiomatic of a real space of states.

We will assume the following basic property for {\relax{S}}' :

Let us now consider two distinct pure states \sigma_1,\sigma_2\in {{\relax{S}}'}{}^{{}^{pure}} such that (\sigma_1\sqcap_{{}_{{{\relax{S}}'}}}\sigma_2)\not= \bot_{{}_{{{\relax{S}}'}}} and \sigma_1^\star \not\sqsubseteq_{{}_{{{\relax{S}}'}}}\sigma_2, we have then \{\sigma_1,\sigma_2\}\in \widehat{\mathcal{K}}({{\relax{S}}'}) but cl_c^{{{\relax{S}}'}}(\{\sigma_1,\sigma_2\})\notin {\mathcal{K}}({{\relax{S}}'}), i.e. \{\sigma_1,\sigma_2\}\notin {\mathcal{Q}}_c({{\relax{S}}'}).
As a conclusion, if {\relax{S}}' satisfies the assumed property (r@assumedcoveringassumedcovering) and if there exists \sigma_1,\sigma_2\in {{\relax{S}}'}{}^{{}^{pure}} such that (\sigma_1\sqcap_{{}_{{{\relax{S}}'}}}\sigma_2)\not= \bot_{{}_{{{\relax{S}}'}}}, then the ontic completion of ({{\relax{S}}'},\star) cannot be complete.

There exists a map denoted \iota^{{{\overline{S}}}_{AB}} from \overline{{ \relax{S}}_{A}} {\times} \overline{{ \relax{S}}_{B}} to {{\overline{S}}}_{AB}. We have explicitly

We will adopt the basic notation \sigma_A\widetilde{\otimes}\sigma_B:=\iota^{{{\overline{S}}}_{AB}}(\sigma_A,\sigma_B)

By construction, we have

For any \{\,\sigma_{i,A}\;\vert\; i\in I\,\}\subseteq \overline{{ \relax{S}}_{A}}, \{\,\sigma_{j,B}\;\vert\; j\in J\,\}\subseteq \overline{{ \relax{S}}_{B}}, \sigma_A\in \overline{{ \relax{S}}_{A}} and \sigma_B\in \overline{{ \relax{S}}_{B}}, we have the two following fundamental properties

Let us consider u_{AB}:=\{\, (\sigma_{i,A},\sigma_{i,B})\;\vert\; i\in I\,\} an element of { \mathcal{P}}(\overline{{ \relax{S}}_{A}} \times \overline{{ \relax{S}}_{B}}). We have explicitly, for any \sigma_{A}\in { \relax{S}}_{A} and \sigma_{B}\in { \relax{S}}_{B}, the following equivalence

It is recalled that { \relax{S}}_{A} and { \relax{S}}_{B} are down-complete Inf semi-lattice and then the infima in this formula are well-defined.

{{\overline{S} }}_{AB} is a down-complete Inf semi-lattice with

If { \relax{S}}_{A} and { \relax{S}}_{B} admit \bot_{{}_{{\relax{S}}_A}} and \bot_{{}_{{\relax{S}}_B}} respectively as bottom elements, then {{\overline{S} }}_{AB} admits a bottom element explicitly given by \bot_{{}_{{{\overline{S} }}_{AB}}}=\bot_{{}_{{\relax{S}}_A}}\widetilde{\otimes}\bot_{{}_{{\relax{S}}_B}}.

We have the two following homomorphisms

If F is a filter of {S}_{AB} then the set \alpha(F):=\{\,(\sigma_{A},\sigma_{B})\in \overline{{ \relax{S}}_{A}} \times \overline{{ \relax{S}}_{B}}\;\vert\; \sigma_{A}\otimes \sigma_{B} \in F\;\} is a bi-filter of \overline{{ \relax{S}}_{A}} \times \overline{{ \relax{S}}_{B}}.

cite Fraser1978 Let us choose \sigma_A,\sigma_{1,A},\cdots,\sigma_{n,A}\in \overline{{ \relax{S}}_{A}} and \sigma_B,\sigma_{1,B},\cdots,\sigma_{n,B}\in \overline{{ \relax{S}}_{B}}. Then,

cite Fraser1978
Let us choose \sigma_A,\sigma_{1,A},\cdots,\sigma_{n,A}\in \overline{{ \relax{S}}_{A}} and \sigma_B,\sigma_{1,B},\cdots,\sigma_{n,B}\in \overline{{ \relax{S}}_{B}}. Then,

where p^\ast denotes the lattice polynomial obtained from p by dualizing the lattice operations.

We have the following obvious property relating the partial orders of {{\overline{S} }}{}^{fin}_{AB} and {S}_{AB}. For any \{(\sigma_{i,A},\sigma_{i,B})\;\vert\; i\in I\}\subseteq_{fin} S_{AB},

If \overline{{ \relax{S}}_{A}} or \overline{{ \relax{S}}_{B}} are distributive (for example as simplex spaces of states), then {{\overline{S} }}{}^{fin}_{AB} and {S}_{AB} are isomorphic posets.

If { \relax{S}}_A and {{ \relax{S}}_B} are simplex spaces of states, then {{\overline{S} }}_{AB} is also a simplex space of states. In other words, the bipartite experiments based on a pair { \relax{S}}_A,{ \relax{S}}_B of deterministic spaces of states are described by a deterministic space of states obtained as the canonical tensor product { \relax{S}}_A {\otimes}{ \relax{S}}_B (Indeed, using Theorem r@theoremsqsubseteqPAB=SABtheoremsqsubseteqPAB=SAB, we have also {{\overline{S} }}{}^{fin}_{AB}={S}_{AB} in this situation).
In that case, the explicit expression for the supremum of two elements in {{\overline{S} }}{}^{fin}_{AB} is given by

Let \widetilde{\sigma}_{AB} and \widetilde{\sigma}'_{AB} be two elements of {{{ \overline{S}}}}_{AB} having a common upper-bound. Then the supremum of \{\widetilde{\sigma}_{AB},\widetilde{\sigma}'_{AB}\} exists in {{{ \overline{S}}}}_{AB} and its expression is given by

\overline{ \relax{S}}_{A} and \overline{ \relax{S}}_{B} being atomic, then {{{ \overline{S}}}}_{AB} is also atomic, i.e.

The set of atoms of {{{ \overline{S}}}}_{AB} is indeed defined by

The map \star is involutive and order-reversing and satisfies \forall \sigma\in {{{ \overline{S}}}}_{AB}\smallsetminus \{\bot_{{}_{{{{ \overline{S}}}}_{AB}}}\},\;\;\;\; \neg \;\widehat{\sigma^{\star}\sigma}{}^{{}^{{{{{ \overline{S}}}}_{AB}}}}.

Let us suppose that \overline{ \relax{S}}_A and \overline{ \relax{S}}_B both satisfy the following basic property

Then {{{ \overline{S}}}}_{AB} satisfies the same property.
In other words, for any \sigma,\omega\in {{{ \overline{S}}}}{}_{AB}^{{}^{pure}} such that \omega\not=\sigma we have then (\omega\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\sigma)\raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{{{{ \overline{S}}}}_{AB}}}\sigma,\omega.

Let us suppose that \overline{ \relax{S}}_A and \overline{ \relax{S}}_B both satisfy the following basic property

Then {{{ \overline{S}}}}_{AB} satisfies the same property.
In other words, if we consider \lambda,\alpha,\beta,\gamma,\delta\in {{{ \overline{S}}}}{}_{AB}^{{}^{pure}} distinct, such that (\,(\alpha\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\beta)\not=(\gamma\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\delta) and (\alpha\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\beta),(\gamma\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\delta)\raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{{{{ \overline{S}}}}_{AB}}}\lambda \,), we have then (\alpha\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\beta\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\gamma\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\delta) \raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{{{{ \overline{S}}}}_{AB}}}(\alpha\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\beta),(\gamma\sqcap_{{}_{{{{ \overline{S}}}}_{AB}}}\delta).

The minimal tensor product is adequate for the description of bipartite deterministic experiments.

{{{ \relax{S}}_A}\widehat{\otimes} {{ \relax{S}}_{B}}} is a down complete Inf semi-lattice. The lowest bound is given by \bot_{{}_{{{{ \relax{S}}_A}\widehat{\otimes} {{ \relax{S}}_{B}}}}}=\bot_{{}_{{{\overline{ \relax{S}}_A}\widetilde{\otimes} {\overline{ \relax{S}}_{B}}}}}.

{{{ \relax{S}}_A}\widehat{\otimes} {{ \relax{S}}_{B}}} admits a real structure given by ({\overline{ \relax{S}}_A}\widetilde{\otimes} {\overline{ \relax{S}}_B},\star).

There exists a map \iota^{{{ \relax{S}}_A}\widehat{\otimes} {{ \relax{S}}_{B}}} from \overline{{ \relax{S}}_A} {\times} \overline{{ \relax{S}}_{B}} to {{{ \relax{S}}_A}\widehat{\otimes} {{ \relax{S}}_{B}}}.

We have the two required bimorphic properties for the tensor product. Explicitly, for any \{\,\sigma_{i,A}\;\vert\; i\in I\,\}\subseteq \overline{{ \relax{S}}_{A}}, \{\,\sigma_{j,B}\;\vert\; j\in J\,\}\subseteq \overline{{ \relax{S}}_{B}}, \sigma_A\in \overline{{ \relax{S}}_{A}} and \sigma_B\in \overline{{ \relax{S}}_{B}}, we have

We have the two following partial traces

The tensor product { \relax{S}}_{A}\widehat{\otimes} { \relax{S}}_{B} defined in Definition r@definitiongenerictensorproductdefinitiongenerictensorproduct satisfies the requirements addressed in subsection r@subsectioncompoundfirstremarkssubsectioncompoundfirstremarks for the description of generic bipartite experiments.

If (\overline{{ \relax{S}}_{A}},\star) and (\overline{{ \relax{S}}_{B}},\star) are completely indeterministic spaces of states, then (\overline{{ \relax{S}}_{A}}\widetilde{\otimes} \overline{{ \relax{S}}_{B}},\star) is also a completely indeterministic space of states.

Let { \relax{S}} be a simplex.
We note that { \relax{S}}=\overline{ \relax{S}}, \overline{ \relax{E}}_{ \relax{S}}={ \relax{E}}_{ \relax{S}} and all measurements are real.
Let us consider any two measurement morphisms { \relax{m}}_{{ \relax{l}}_{(\sigma_1,\sigma'_1)}} and { \relax{m}}_{{ \relax{l}}_{(\sigma_2,\sigma'_2)}}. These morphisms are necessarily compatible.

If { \relax{S}} is a simplex then the whole set of elements of \{\,{ \relax{m}}_{{ \relax{l}}} \;\vert\; {{ \relax{l}}}\in { \relax{E}}_{ \relax{S}}\,\} are jointly compatible measurement maps. In other words, the compatibility cover { \mathcal{C}} is then reduced to the singleton \{\overline{ \relax{E}}_{ \relax{S}}\}.

If the N-uple of real measurement maps ({ \relax{m}}_{{ \relax{l}}_1},\cdots, { \relax{m}}_{{ \relax{l}}_N}) are jointly compatible, then the (N+1)-uple of real measurement maps ({ \relax{m}}_{{ \relax{l}}_1},{ \relax{m}}_{{ \relax{l}}_1},\cdots, { \relax{m}}_{{ \relax{l}}_N}) are jointly-compatible, for any { \relax{l}}_1,\cdots,{ \relax{l}}_N\in { \relax{E}}.

If the N-uple of real measurement maps ({ \relax{m}}_{{ \relax{l}}_1},\cdots, { \relax{m}}_{{ \relax{l}}_{N}}) are jointly compatible, then the (N+1)-uple of real measurement maps ({ \relax{m}}_{\overline{{ \relax{l}}_{1}}},{ \relax{m}}_{{ \relax{l}}_{1}},\cdots, { \relax{m}}_{{ \relax{l}}_{N}}) are also jointly-compatible.

For any \eta\in { \relax{B}}, we define the morphism \tau_{\eta} as the constant map from { \relax{S}} to { \relax{B}} sending any state to \eta.
If the N-uple of real measurement maps ({ \relax{m}}_{{ \relax{l}}_{1}},\cdots, { \relax{m}}_{{ \relax{l}}_{N}}) are jointly compatible, then the (N+1)-uple of real measurement maps (\tau_{\eta},{ \relax{m}}_{{ \relax{l}}_{1}},\cdots, { \relax{m}}_{{ \relax{l}}_{N}}) are jointly compatible.

For any \eta\in { \relax{B}}, if the N-uple of real measurement maps ({ \relax{m}}_{{ \relax{l}}_{1}},\cdots, { \relax{m}}_{{ \relax{l}}_{N}}) are jointly compatible (N\geq 2), then the (N+1)- uple of real measurement maps ({ \relax{m}}_{({ \relax{l}}_{1} \sqcap_{{}_{{ \relax{E}}_{ \relax{S}}}} { \relax{l}}_{2})},{ \relax{m}}_{{ \relax{l}}_{1}},\cdots, { \relax{m}}_{{ \relax{l}}_{N}}) are jointly compatible.

Let C \in { \mathcal{C}} designate a maximal compatibility context, we have

We note the following fact, implicit in (r@contextwedgecontextwedge) and clarified in Lemma r@genmeasurementwedgecompatiblegenmeasurementwedgecompatible :

If { \relax{S}} is a simplex then every coherent operational description { \relax{d}} are defined globally. The empirical model associated to a simplex space of states is then non-contextual.

Let us fix a coherent operational description { \relax{D}}:=({ \relax{d}}^{C})_{C \in { \mathcal{C}}} \in {{ \mathcal{D}}} associated to the compatibility cover { \mathcal{C}}. Let us also fix a compatibility context C. Then,

More explicitly, we will fix explicitly the solution \sigma^C_{ \relax{D}} of the equation (r@descriptionstatedescriptionstate) defined by

The family (\sigma^C_{ \relax{D}})_{C \in { \mathcal{C}}}\in \overline{ \relax{S}}{}^{\, \mathcal{C}} is the ontic state associated to { \relax{D}}.

Let us consider a compatibility context denoted C such that C\subseteq { \relax{U}}_{\overline{ \relax{S}}}. We have necessarily

In the following, we will adopt the following notation

Let C be a maximal compatibility context of type 1. Then,

Let C be a maximal compatibility context of type 2 and let \gamma,\gamma'\in \overline{ \relax{S}} \smallsetminus \{\bot_{{}_{\overline{ \relax{S}}}}\} with \gamma'\sqsupseteq_{{}_{\overline{ \relax{S}}}}\gamma^\star and such that { \relax{l}}_{(\gamma,\gamma')}\in C. Then,

Let C be a maximal compatibility context of type 2 such that there exists \gamma\in \overline{ \relax{S}} \smallsetminus \{\bot_{{}_{\overline{ \relax{S}}}}\} with { \relax{l}}_{(\gamma,\gamma^\star)}\in C. Then,

For any pair C^{(1)}_{\omega},C^{(1)}_{\omega'} of maximal compatibility contexts of type 1 associated to \omega,\omega'\in \overline{ \relax{S}}^{{}^{pure}} (see Lemma r@lemmacontexttype1lemmacontexttype1), we have

If there exists \omega,\omega'\in \overline{ \relax{S}}^{{}^{pure}} such that \Sigma^{C^{(1)}_{\omega}}_{ \relax{D}} and \Sigma^{C^{(1)}_{\omega'}}_{ \relax{D}} admit a common upper-bound in \overline{ \relax{S}}, then for any \omega''\in \underline{\Sigma^{C^{(1)}_{\omega}}_{ \relax{D}} \sqcup_{{}_{\overline{ \relax{S}}}} \Sigma^{C^{(1)}_{\omega'}}_{ \relax{D}}}_{{}_{\overline{ \relax{S}}}} we have \Sigma^{C^{(1)}_{\omega''}}_{ \relax{D}} \sqsupseteq_{{}_{\overline{ \relax{S}}}} (\Sigma^{C^{(1)}_{\omega}}_{ \relax{D}} \sqcup_{{}_{\overline{ \relax{S}}}} \Sigma^{C^{(1)}_{\omega'}}_{ \relax{D}})

For any coherent operational description { \relax{D}} the subset U_{ \relax{D}} of \overline{ \relax{S}} defined by U_{ \relax{D}}:=\{\, \Sigma^{C^{(1)}_{\omega}}_{ \relax{D}}\;\vert\; \omega\in \overline{ \relax{S}}{}^{{}^{pure}}\,\} satisfies

The subset of \overline{ \relax{S}} defined by \{\, \Sigma^{C^{(1)}_{\omega}}_{ \relax{D}}\;\vert\; \omega\in \overline{ \relax{S}}^{{}^{pure}}\,\} satisfies

Every operational description { \relax{D}} are admissible.

As long as the operational description { \relax{D}} is admissible (see Theorem r@theoremoperationaldescriptionadmissibletheoremoperationaldescriptionadmissible), the supremum

is well defined as an element of { \relax{S}}. Then,we have

Let us consider a maximal compatibility context of type 1 denoted C^{(1)}_{\omega}. We have

Let us consider a maximal compatibility context of type 2 denoted by C. We have

Let us consider an operational description { \relax{D}}. We have then necessarily the following property :

The map from { \mathcal{D}} to { \relax{S}}, sending { \relax{D}} to \Sigma_{ \relax{D}}:= (\bigsqcup{}^{{}^{{ \relax{S}}}}_{\omega\in \overline{ \relax{S}}^{{}^{pure}}}\; \Sigma^{C^{(1)}_{\omega}}_{ \relax{D}}) is an isomorphism.

Let us consider \mu,\nu,\phi\in \overline{ \relax{S}}{}^{{}^{pure}} and \gamma\in \overline{ \relax{S}} such that \gamma=(\phi\sqcap_{{}_{\overline{\relax{S}}}}\nu)\raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{\overline{ \relax{S}}}} \phi,\nu and \mu^\star\not\sqsubseteq_{{}_{\overline{ \relax{S}}}}\gamma, we have

We have then either \chi\in \overline{ \relax{S}}{}^{{}^{pure}} or \chi\in { \relax{S}}\smallsetminus \overline{ \relax{S}} and then, in the last case, denoting as usual \Theta^{\overline{ \relax{S}}}(\chi):=Max((\downarrow_{{}_{{ \relax{S}}}}\chi)\cap \overline{ \relax{S}}), we have

Let us once again consider \mu\in \overline{ \relax{S}}{}^{{}^{pure}} and \gamma\in \overline{ \relax{S}} such that \mu^\star\not\sqsubseteq_{{}_{\overline{ \relax{S}}}}\gamma. Let us assume that there exists \nu\in \overline{ \relax{S}}{}^{{}^{pure}} such that \gamma=(\mu\sqcap_{{}_{\overline{\relax{S}}}}\nu)\raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{\overline{ \relax{S}}}} \mu,\nu then let us here again denote \chi=(\mu^\star\sqcup_{{}_{{ \relax{S}}}}\gamma).
If we have \chi\in \overline{ \relax{S}}, then necessarily \chi\in \overline{ \relax{S}}{}^{{}^{pure}}, and if we have \chi\in { \relax{S}}\smallsetminus \overline{ \relax{S}} then

The collections (\varphi_{\epsilon})_{\epsilon},(\psi_{\epsilon})_{\epsilon} \in (\overline{ \relax{S}}{}^{{}^{pure}})^{\Theta^{\overline{ \relax{S}}}(\chi)} satisfying the properties (r@thirdcoveringpropertySbar0thirdcoveringpropertySbar0)(r@thirdcoveringpropertySbarviicthirdcoveringpropertySbarviic)(r@thirdcoveringpropertySbarviibthirdcoveringpropertySbarviib) are unique up to a permutation on the elements of \Theta^{\overline{ \relax{S}}}(\chi)\smallsetminus \{\gamma\}.

Let us once again consider \mu\in \overline{ \relax{S}}{}^{{}^{pure}} and \gamma\in \overline{ \relax{S}} such that \mu^\star\not\sqsubseteq_{{}_{\overline{ \relax{S}}}}\gamma. If there exists \nu\in \overline{ \relax{S}}{}^{{}^{pure}} such that \gamma=(\mu\sqcap_{{}_{\overline{\relax{S}}}}\nu)\raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{\overline{ \relax{S}}}} \mu,\nu then let us here again denote \chi=(\mu^\star\sqcup_{{}_{{ \relax{S}}}}\gamma). If we have \chi\in { \relax{S}}\smallsetminus \overline{ \relax{S}} then

Let us consider \nu and \phi two distinct elements of {{{ \overline{\relax{S}}}}}{}^{{}^{pure}} and \mu another element of {{{ \overline{\relax{S}}}}}{}^{{}^{pure}}. Let us assume that \mu \not\!\!\!\underline{\perp}\nu,\phi. We have

In particular, \exists \chi\in {{{ \widetilde{\relax{S}}}}}\;\vert\; (\,(\nu\sqcap_{{}_{{{{ \overline{\relax{S}}}}}}}\phi) \raisebox{1.3ex}{\,\rotatebox{180}{}}\,_{{}_{{{{ \widetilde{\relax{S}}}}}}}\chi \;\;\;\;\textit{\rm and}\;\;\;\; \chi\,\underline{\perp}\,\mu\,)\,) implies \forall i\in I,\;\; \zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(i)}(\nu) = \zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(i)}(\phi).

Let us denote j,k\in \{1,\cdots,N\} with j<k such that for any i\in \{1,\cdots,N\}\smallsetminus \{j,k\} we have \zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(i)}(\nu) = \zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(i)}(\phi). Endly, let us denote \alpha:=\zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(j)}(\mu), \beta:=\zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(k)}(\mu), \alpha':=\zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(j)}(\nu), \beta':=\zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(k)}(\nu), \alpha'':=\zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(j)}(\phi), \beta'':=\zeta^{{}^{{\relax{S}}_{A_1}\cdots{\relax{S}}_{A_N}}}_{(k)}(\phi).
As long as \lambda:=\mu\sqcup_{{}_{\relax{S}}}(\nu\sqcap_{{}_{\overline{\relax{S}}}}\phi) exists in {{\relax{S}}}\smallsetminus {\overline{\relax{S}}}, we have (see (r@computationexamplecomputationexample))

We have the following basic property

We note the following sheaf condition

The ternary relation {\relax{r}}^U defined on the consistent subset U is symmetric under any permutation of its arguments.

[Veblen-Young’s first geometric property]

Let us consider \chi\in ({\check{\relax{S}}}\smallsetminus \overline{\relax{S}}). We have

Let U\subseteq {\relax{G}}^{\check{\relax{S}}} be a consistent subset of states, such that there exists \sigma_1,\sigma_2,\sigma_3,\sigma_4,\lambda\in U satisfying

then we have necessarily

Let us consider U\in { \relax{E}}^{\check{\relax{S}}} such that there exists \chi\in ({\check{\relax{S}}}\smallsetminus \overline{ \relax{S}}) with \chi\in U. The subset U is necessarily of one of the two following types

2em (Consistent Type 1) \exists \textit{\bf !} \alpha\in \Theta^{\overline{ \relax{S}}}(\chi) such that \forall \sigma\in U we have (\sigma\sqcap_{{}_{\overline{ \relax{S}}}}\chi)=\alpha. 2em (Consistent Type 2) \exists \alpha,\beta\in \Theta^{\overline{ \relax{S}}}(\chi),\alpha\not=\beta and \exists \sigma,\sigma'\in U\cap \overline{ \relax{S}}{}^{{}^{pure}} such that (\sigma\sqcap_{{}_{\overline{ \relax{S}}}}\chi)=\alpha and (\sigma'\sqcap_{{}_{\overline{ \relax{S}}}}\chi)=\beta.

Let us consider a maximal consistent subset of states U\in { \relax{E}}^{\check{\relax{S}}} and let us assume that there exists \lambda\in ({\check{\relax{S}}}\smallsetminus \overline{ \relax{S}}) and \sigma_1,\sigma_2,\sigma_3,\sigma_4\in \overline{ \relax{S}}{}^{{}^{pure}} such that \lambda,\sigma_1,\sigma_2,\sigma_3,\sigma_4\in U and such that {\relax{r}}^U(\lambda,\sigma_1,\sigma_2) and {\relax{r}}^U(\lambda,\sigma_3,\sigma_4) and \forall \mu,\nu,\rho\in \{\sigma_1,\sigma_2,\sigma_3,\sigma_4\}, \;\neg\; {\relax{r}}^U(\mu,\nu,\rho). Then, the consistent subset U is necessarily of consistent type 2 .

[Veblen-Young’s second geometric property]

[Veblen-Young’s third geometric property]
We will consider \sigma_1,\sigma_2,\sigma_3,\sigma_4\in { \relax{E}}^{\check{\relax{S}}}. We have the following result

{ \relax{E}}^{\check{\relax{S}}} is equipped with a structure of projective sheaf. In other words, apart from the sheaf condition (r@VeblenYoung0VeblenYoung0), we have proved the Veblen-Young’s geometric axioms of projective spaces (r@VeblenYoung3VeblenYoung3)(r@VeblenYoung2VeblenYoung2)(r@VeblenYoung1VeblenYoung1).

Let us consider a consistent subset of states U\in { \relax{E}}^{\check{\relax{S}}} and let us assume that there exists \chi\in ({\check{\relax{S}}}\smallsetminus \overline{ \relax{S}}) and that the consistent subset U is of consistent type 2.
Let us now assume that U is orthogonally complete. Then, we have necessarily \chi\in ({\widecheck{\relax{S}}}\smallsetminus \overline{ \relax{S}}).
Conversely, let us assume that \chi\in ({\widecheck{\relax{S}}}\smallsetminus \overline{ \relax{S}}) and that U is maximal, then we have

and then necessarily U is orthogonally complete.

Let us consider an orthogonally complete consistent subset such that there exists \chi\in ({\widecheck{\relax{S}}}\smallsetminus \overline{ \relax{S}}) with \chi\in U. We adopt the same notation as in (r@thirdcoveringpropertySbar0thirdcoveringpropertySbar0) (r@thirdcoveringpropertySbar0bisthirdcoveringpropertySbar0bis) (r@thirdcoveringpropertySbarviicthirdcoveringpropertySbarviic) (r@thirdcoveringpropertySbarviibthirdcoveringpropertySbarviib) for \gamma and for \varphi_{\alpha} and \psi_{\alpha} with \alpha any element of \Theta^{\overline{ \relax{S}}}(\chi).
Let us now assume that U is of consistent type 1, i.e. \exists \textit{\bf !} \delta\in \Theta^{\overline{ \relax{S}}}(\chi) such that \forall \sigma\in U we have (\sigma\sqcap_{{}_{\overline{ \relax{S}}}}\chi)=\delta. Let us fix this element \delta.
Then, we are in one of the two following cases
(1) \delta\in \Theta^{\overline{ \relax{S}}}(\chi)\smallsetminus \{\gamma\}. Then, there exists \chi'\in {\widecheck{\relax{S}}}\smallsetminus {\overline{\relax{S}}} such that

(2) \delta=\gamma. Then, there exists \chi'\in {\widecheck{\relax{S}}}\smallsetminus {\overline{\relax{S}}} such that

Let us consider U\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} such that U\subseteq \overline{ \relax{S}}{}^{{}^{pure}}. Let us moreover consider \lambda,\sigma_1,\sigma_2,\sigma_3,\sigma_4\in U, all distinct, such that {\relax{r}}^U(\lambda,\sigma_1,\sigma_2) and {\relax{r}}^U(\lambda,\sigma_3,\sigma_4) and \forall\mu,\nu,\rho\in \{\sigma_1,\sigma_2,\sigma_3,\sigma_4\}, \;\neg\; {\relax{r}}^U(\mu,\nu,\rho). Endly, we will suppose that the plane to which \lambda,\sigma_1,\sigma_2,\sigma_3,\sigma_4 belong is not a starred plane.
Then, there exists \mu\in { \relax{G}}^{\widecheck{\relax{S}}}\smallsetminus \overline{ \relax{S}} and there exists U'\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \mu, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U' such that {\relax{r}}^{U'}(\mu,\sigma_1,\sigma_3) and {\relax{r}}^{U'}(\mu,\sigma_2,\sigma_4), and there exists \nu\in { \relax{G}}^{\widecheck{\relax{S}}}\smallsetminus \overline{ \relax{S}} and there exists U'\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \nu, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U' such that {\relax{r}}^{U'}(\nu,\sigma_2,\sigma_3) and {\relax{r}}^{U'}(\nu,\sigma_1,\sigma_4).

We will consider \sigma_1,\sigma_2,\sigma_3,\sigma_4\in {\overline{\relax{S}}}{}^{{}^{pure}} and \lambda \in { \relax{G}}^{\widecheck{\relax{S}}}\smallsetminus {\overline{\relax{S}}} and we will assume the existence of U\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \lambda, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U and such that {\relax{r}}^U(\lambda,\sigma_1,\sigma_2) and {\relax{r}}^U(\lambda,\sigma_3,\sigma_4) and \forall\omega,\nu,\rho\in \{\sigma_1,\sigma_2,\sigma_3,\sigma_4\}, \;\neg\; {\relax{r}}^U(\omega,\nu,\rho). Endly, we will assume that the plane to which \sigma_1,\sigma_2,\sigma_3,\sigma_4 belong is not a starred plane.
Then, there exists \mu\in {\overline{\relax{S}}}{}^{{}^{pure}} and there exists U'\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \mu, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U' such that {\relax{r}}^{U'}(\mu,\sigma_1,\sigma_3) and {\relax{r}}^{U'}(\mu,\sigma_2,\sigma_4) and there exists \chi\in { \relax{G}}^{\widecheck{\relax{S}}}\smallsetminus {\overline{\relax{S}}} and there exists U''\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \chi, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U'' such that {\relax{r}}^{U''}(\chi,\sigma_2,\sigma_3) and {\relax{r}}^{U''}(\chi,\sigma_1,\sigma_4), or there exists \mu\in {\overline{\relax{S}}}{}^{{}^{pure}} and there exists U'\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \mu, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U' such that {\relax{r}}^{U'}(\mu,\sigma_2,\sigma_3) and {\relax{r}}^{U'}(\mu,\sigma_1,\sigma_4) and there exists \chi\in { \relax{G}}^{\widecheck{\relax{S}}}\smallsetminus {\overline{\relax{S}}} and there exists U''\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \chi, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U'' such that {\relax{r}}^{U''}(\chi,\sigma_1,\sigma_3) and {\relax{r}}^{U''}(\chi,\sigma_2,\sigma_4).

[restricted Veblen-Young’s third geometric property]
Let us consider U\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} and \lambda, \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U such that {\relax{r}}^U(\lambda,\sigma_1,\sigma_2) and {\relax{r}}^U(\lambda,\sigma_3,\sigma_4) and \forall\mu,\nu,\rho\in \{\sigma_1,\sigma_2,\sigma_3,\sigma_4\}, \;\neg\; {\relax{r}}^U(\mu,\nu,\rho). Endly, we will assume that the plane to which \sigma_1,\sigma_2,\sigma_3,\sigma_4 belong is not a starred plane.
From Lemma r@nondegenerateallsigmapurenondegenerateallsigmapure, we know that \sigma_1,\sigma_2,\sigma_3,\sigma_4\in \overline{\relax{S}}{}^{{}^{pure}}. From Lemma r@natureorthocompleteUnatureorthocompleteU, we know that, as long as U is orthogonally complete, we have \lambda\in {\relax{G}}^{\widecheck{\relax{S}}}. Endly, from Lemma r@firstlemmawidecheckfirstlemmawidecheck and Lemma r@caseanalysisUSpurecaseanalysisUSpure, we know that there exists U'\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}} with \sigma_1,\sigma_2,\sigma_3,\sigma_4\in U' and \lambda'\in U'\cap {\relax{G}}^{\widecheck{\relax{S}}} satisfying {\relax{r}}^{U'}(\lambda',\sigma_1,\sigma_3) and {\relax{r}}^{U'}(\lambda',\sigma_2,\sigma_4), and there exists U''\in { \relax{E}}_{\underline{\perp}}^{\check{\relax{S}}},\sigma_1,\sigma_2,\sigma_3,\sigma_4\in U'' and \lambda''\in U''\cap {\relax{G}}^{\widecheck{\relax{S}}} satisfying {\relax{r}}^{U''}(\lambda'',\sigma_1,\sigma_4) and {\relax{r}}^{U''}(\lambda'',\sigma_2,\sigma_3).

[Orthogonality properties]
The restriction to {\relax{G}}^{\widecheck{\relax{S}}} of the orthogonality relation \underline{\perp} satisfies the following properties.

[Irreducibility]
The space of states { \relax{S}} equipped with its real structure (\overline{ \relax{S}},\star) satisfies the following irreducibility property

Let { \relax{m}}_1,{ \relax{m}}_2\in \overline{\relax{M}}_{{}_{{ \relax{S}}}} be two real measurements on the space of states { \relax{S}} (they are real morphisms from { \relax{S}} to { \relax{B}}).
If there exists a real morphism \Psi \in \overline{\relax{C}}({ \relax{S}},{ \relax{S}}\widetilde{\otimes}{ \relax{S}}) such that \forall \sigma\in \overline{\relax{S}},\;(\zeta^{{{ \relax{S}}}{{ \relax{S}}}}_{(1)}\circ \Psi)(\sigma) = (\zeta^{{{ \relax{S}}}{{ \relax{S}}}}_{(2)}\circ \Psi)(\sigma) = \sigma, then there exists a real morphism \Psi_{({ \relax{m}}_1,{ \relax{m}}_2)} \in \overline{\relax{C}}({ \relax{S}},{ \relax{B}}\widetilde{\otimes}{ \relax{B}}) such that \forall \sigma\in \overline{\relax{S}}, (\zeta^{{{ \relax{B}}}{{ \relax{B}}}}_{(1)}\circ \Psi_{({ \relax{m}}_1,{ \relax{m}}_2)})(\sigma) = { \relax{m}}_1(\sigma) and (\zeta^{{{ \relax{B}}}{{ \relax{B}}}}_{(2)}\circ \Psi_{({ \relax{m}}_1,{ \relax{m}}_2)})(\sigma) = { \relax{m}}_2(\sigma).

[No-broadcasting theorem]
If the space of states { \relax{S}} (equipped with its real structure (\overline{ \relax{S}},\star)) is not deterministic, it does not allow the broadcasting.

If there exists two pairs of real measurements {\phi}_1\in \overline{ \relax{C}}({ \relax{S}}_A,{ \relax{B}}) and {\phi}_2\in \overline{ \relax{C}}({ \relax{S}}_A,{ \relax{B}}), {\rho}_1\in \overline{ \relax{C}}({ \relax{S}}_D,{ \relax{B}}) and {\rho}_2\in \overline{ \relax{C}}({ \relax{S}}_D, { \relax{B}}) such that there exists no state \Lambda \in { \relax{B}}^{{\widetilde{\otimes}}4} satisfying simultaneously the four relations (r@bell1bell1) (r@bell2bell2) (r@bell3bell3) (r@bell4bell4), then the state {\Sigma} is a Bell non-local state.

Bell non-local states do exist in {{{ \widetilde{S}}}}_{AB}.