際際滷

際際滷Share a Scribd company logo
Richard D. Gill, Bell mini-conference February 11 2023
Comment on
Contextuality or Nonlocality: What
Would John Bell Choose Today?
by Marian Kupczynski
Entropy 2023, 25, 280 (13pp.). https://doi.org/10.3390/e25020280
Entropy 2023, 25, 280 (13pp.). https://doi.org/10.3390/e25020280
 Random variables are represented by nodes in a DAG
 Their joint probability distribution is built up by composing
conditional distributions of the variable in each node given its
node parents
 (Variables in root nodes are statistically independent of one
another)
 Why this works: the nodes of a DAG can be ordered so that all
arrows are pointing the same way
Express in a graphical model
i, j 1, 2
i, j
Ai(1, i) Bj(2, j)
~
~
Settings
(may be correlated)
Source hidden variables
(may be correlated)
Context dependent
instrument hidden variables
(may be correlated)
Alices outcome
+/1
Bobs outcome
+/1
Notice:
 If the instrument hidden variables i, j are allowed to be
statistically dependent, then any four joint probability distributions
of pairs of setting outcomes (Ai, Bj) are allowed
 If the instrument hidden variables are assumed to be
uncorrelated, then CHSH holds (Gill and Lambare)
 MKs escape route: the detection loophole. Outcome can also be
zero, and MK goes on to study correlations conditional on two
non-zero outcomes
 The cause of the muddle: bad notation
The dilemma
Kupczynksis quandrary

More Related Content

Bell mini conference RDG.pptx

  • 1. Richard D. Gill, Bell mini-conference February 11 2023 Comment on Contextuality or Nonlocality: What Would John Bell Choose Today? by Marian Kupczynski Entropy 2023, 25, 280 (13pp.). https://doi.org/10.3390/e25020280
  • 2. Entropy 2023, 25, 280 (13pp.). https://doi.org/10.3390/e25020280
  • 3. Random variables are represented by nodes in a DAG Their joint probability distribution is built up by composing conditional distributions of the variable in each node given its node parents (Variables in root nodes are statistically independent of one another) Why this works: the nodes of a DAG can be ordered so that all arrows are pointing the same way Express in a graphical model
  • 4. i, j 1, 2 i, j Ai(1, i) Bj(2, j) ~ ~ Settings (may be correlated) Source hidden variables (may be correlated) Context dependent instrument hidden variables (may be correlated) Alices outcome +/1 Bobs outcome +/1 Notice:
  • 5. If the instrument hidden variables i, j are allowed to be statistically dependent, then any four joint probability distributions of pairs of setting outcomes (Ai, Bj) are allowed If the instrument hidden variables are assumed to be uncorrelated, then CHSH holds (Gill and Lambare) MKs escape route: the detection loophole. Outcome can also be zero, and MK goes on to study correlations conditional on two non-zero outcomes The cause of the muddle: bad notation The dilemma Kupczynksis quandrary