>>84 ようやく混乱の原因がわかりましたよ.結局, ナガサワによる隠れた変数理論の定義 ≠ (僕の認識での)局所的決定論 ということです. ナガサワ論文 On the Locality of Hidden-Variable Theories in Quantum Physics の2章の最初のほうにこうあります. We notice, however, that there is no standard definition of hidden-variable theories, although there are various definitions depending on the author. その後,割と普通っぽい定義2.1を書いて,これじゃだめっぽいので, 2.1を拡張した定義2.2を書いて,彼流のhidden-variable theoryを定義しています. やはり言葉の行き違いだったということです.
Abstract Claims on the non-existence of hidden-variable theories (resp. local hidden-variable theories) of quantum mechanics are examined. It is shown that Bell's locality contains an unnecessarily restrictive condition influenced by a narrow definition of hidden-variable theories, and hence one cannot find any local hidden-variable theory in his framework. A local hidden-variable model is constructed for spin-correlations of two particles against Bell's non-existence claim. In this context Mermin's non-existence claim is analyzed, and a local hidden-variable model for spin- correlations of three particles is constructed against Mermin's (as well as Greenberger-Horn- Zellinger's) non-existence claim.
>>97,105 というか,単に「隠れた変数理論」の意味が(少なくとも僕の)常識とは違ったってことだろ. どちらも間違っていない. そっち(?)の業界では there is no standard definition of hidden-variable theories は常識なのかい? もしそうならある程度しょうがないのかも知れんが..
Definition 2.2. For each prediction of a physical system let X denote a set of involved self-adjoint operators and a state ψ∈ H. A theory is a hidden-variable theory of quantum mechanics if it provides a probability measure P^X and a random variable h^X_B on a probability space such that for each observable B ∈ X its expectation <ψ, B ψ> in a state ψ∈ X is given through <ψ, B ψ> = ∫ P^X (dω) h^X_B (ω) where P^X ( resp. h^X_B ) indicates that it may depend on parameters in X (but not necessarily all of the elements in X).
Definition 2.1. Let P_ψ be a probability measure, and h_B(ω) a random variable on a measurable space {Ω, F}. If <ψ, B ψ> = ∫_Ω P_ψ(dω) h_B(ω) , then the random variable h_B is called dispersion-free representation of an observable B, and h_B(ω) is called the value of B in a dispersion-free state {ψ, ω} ∈ H×Ω. The parameters ω ∈ Ω are called hidden-variables. If a theory gives {P_ψ, h_B} for any observable B, then it is called a hidden-variable theory of quantum mechanics.
トリックは,125も指摘した点ですが, 隠れた変数の確率分布 P が ψだけでなく involved self-adjoint operators によっても 変わることを許すことにあるようです. involved という言い方が微妙ですが, 結局,実験で何を測定するかに依存して P が変わることを許してしまいます. つまり,測定器がどの方向のスピンを測るかによって,確率分布のほうが変化してくれるというモデルです.
で,ベルの場合は Single-measure hypothesis: the probability measure P in eqn (11) is fixed, independent of experiments (namely, independent of the orientations {a, b} of Stern-Gerlach magnets). (ここのmesureは「測定」ではなく「測度」) を暗黙に仮定しているが,我々はその仮定を外すというわけです. (もちろん,局所的な決定論を前提とする場合はこれは仮定ではないでしょう.) 啓蒙書の「普遍的な確率法則」はおそらくこれのことと思います.
It is shown that Bell's locality contains an unnecessarily restrictive condition influenced by a narrow definition of hidden-variable theories, and hence one cannot find any local hidden-variable theory in his framework.
3. LOCALITY OF HIDDEN-VARIABLE THEORIES One can consider the `locality of hidden-variable theories of quantum mechanics', but cannot speak of the `locality of quantum mechanics', since quantum mechanics itself has no mathematical structure to discuss the `locality'. This is actually one of the motivations why `hidden-variables' are considered. The discussion of Einstein-Podolsky-Rosen [7] involves the so-called `elements of physical reality'. We will not discuss, in this article, what they meant by `elements of physical reality'.
However, in the context of hidden variable theories, there will be no confusion if we identify `elements of physical reality' with `random variables (or dispersion-free representations)', and we will do so. Then the `locality of hidden-variable theories' can be a well-posed mathematical problem. We must, nonetheless, approach the problem of `locality' carefully, since the so-called `locality' is not uniquely defined, and it depends on the definitions of hidden-variable theories that one adopts, as will be seen. We begin with Bell's locality of hidden-variable theories.
We therefore give a definition of the locality by a pair of conditions:
(L.i) The random variable h_A(ω) (resp. h_B(ω)) in eqn (15) does not depend on b (resp. a), namely, the random variables must be locally determined.
(L.ii) The marginal distribution of the random variable h_A(ω) (resp. h_B(ω)) according to the probability measure P^X must be independent of b (resp. a), namely, the marginal distributions are locally determined.