[ Bayesian reasoning
and its application to physics measurements ]
(For most material in Italian, also concerning probability
and statistics,
see here)
BOOKS
"Bayesian reasoning in data analysis
 A critical introduction", World Scientific Publishing 2003
(soft cover 2013)
[details]
"Così è... probabilmente. Il saggio, l'ingenuo e la signorina Bayes",
with Dino Esposito (in Italian), December 2013 (PoD).
[indice]
Recensione sul sito Scienza per Tutti
[ Per le omonime conferenze divulgative a Frascati e a Roma 3
vedi qui ]
"L'improbabile mondo del Mago di Odds",
with Gianluca Testa (in Italian), June 2016 (PoD)
[copertina,
quarta di copertina
e indice]
Recensione sul sito Scienza per Tutti
PAPERS etc.
(included selected slides)

"What is the probability that a vaccinated person is
shielded from Covid19?
A Bayesian MCMC based reanalysis of published data
with emphasis on what should be reported as 'efficacy' "
(with Alfredo Esposito)
 arXiv:2102.11022 [stat.AP]
(Version v3 of 6 Sep 2021 contains the new subsection 6.1)
 Local copy: 2102.11022.pdf
 Browsable version
— conversion by latex2html 'not perfect'
(and that of version v3 is even worst,
missing completely the figures.)
 Good save us from the bloggers!
 The post (2021/03/10)
(screenshot, in the case it will not
be accessible in the future).
 It is a well known fact that ignorance, arrogance
and sloppy thinking
go often together.
 If you want to form your idea about a scientific paper,
please take your time to read it carefully, provided you have the
scientific background to understand it.

"Ratio of counts vs ratio of rates
in Poisson processes",
arXiv:2012.04455 [stat.ME]

"Inferring vaccine efficacies and their uncertainties.
A simple model implemented in JAGS/rjags" (with Alfredo Esposito)
(first release 19 November 2020)
The work has been finally posted on the arXiv as
arXiv:2102.11022 [stat.AP] (see above).

"Checking individuals and sampling populations with imperfect tests"
(with Alfredo Esposito),
arXiv:2009.04843 [qbio.PE].
(Besides the specific application, it contains general ideas about
inference and prediction, statistics and systematic contribution
to the uncertainty, role of priors, and so on)
 front page and contents
(pdf, 3 pages)
 local copy (pdf, 113 pages): 2009.04843.pdf
 browsable version (conversion
by latex2html 'not perfect')
 R and Jags code (Appendix B):
B1.R,
B2.R,
B3.R,
B4.R,
B5.R,
B6.R,
B7.R,
B8.R,
B9.bug,
B10.R,
B11.R,
B12.R,
B13.R,
B14.R.

"The Gauss' Bayes Factor",
arXiv:2003.10878 [math.HO]
About the Gauss derivation of the Gaussian:
[Note also the beginning of article 178, in which
Gauss comments about the "defectus" of the
"formula" just found ("formula nostra").
But, as he remarks at the end of the
paragraph, besides the fact that the "formula" goes
rapidly to zero, when the errors become very large,
"... ipsos errorum limites obsoluto rigore assignare,
rei natura numquam permittet".
]

"On a curious bias arising when the sqrt(χ^{2}/ν) scaling prescription is first applied to a subsample of the individual results",
arXiv:2001.07562 [physics.dataan]
(Browsable version  conversion
by latex2html 'not perfect')

"Sceptical combination of experimental results using JAGS/rjags
with application to the K^{±} mass determination",
arXiv:2001.03466 [physics.dataan]
(Browsable version  conversion
by latex2html 'not perfect')

"Pvalues: Meaning, misconceptions and
dangers, but also their practical utility if used cum grano salis",
invited talk at itENBIS
and INRIM Joint Workshop
"MSMM 2019 —
Mathematical and Statistical Methods for Metrology", Torino, 3031 May2019.
A related, well done video by Veritasium

"Talking about Probability, Inference and Decisions. Part 1: The Witches of Bayes",
with Noemi Cifani and Alba Gilardi (in Italian),
arXiv:1802.10432 [math.HO]
Per ulteriore informazione sul casus belli
vedi qui
(→ Le streghe di Bayes)).
Recensione di Scienza per Tutti (LNF)

"More lessons from the six box toy experiment",
arXiv:1701.01143
 Files of the sequences of Appendix A:
 To load the files in a
R session:
seq.B0 = as.vector( read.csv("seq_B0.txt")$x ) ; etc.

"Basic probabilistic issues
in the Sciences and in Forensics
(hopefully) clarified by a Toy Experiment
modelled by a Bayesian Network"
Workshop on Bayesian Networks and Argumentation in Evidence Analysis,
2629 September 2016, Cambridge, UK.
Slides of the presentation (1.1MB).
Video available at the
INI website.

"Probability, Propensity and Probability of Propensities
(and of Probabilities)"
Invited talk at
MaxEnt 2016
arXiv:1612.05292 and
INI preprint ni16052
(FOS Program) of the
Isaac Newton Institute for Mathematical Sciences,
AIP Conference Proceedings 1853, 030001 (2017)
(Browsable version  conversion
by latex2html 'not perfect')

"The Waves and the Sigmas (To Say Nothing of the 750 GeV Mirage)"
Paper based on
the invited talk at
MaxEnt 2016,
arXiv:1609.01668,
pdf (511k),
INI preprint ni16046
(FOS Program) of the
Isaac Newton Institute for Mathematical Sciences.
 browsable version
 R script: p_value_Vs_BF.R
 For the Hugin Bayesian network mentioned in footnote 32
see here.
 Betting against
the 750 Mev mirage
 Related links/resources:
 Is Most Published Research Wrong?:
YouTube
by Veritasium (very well done!);
 I Fooled Millions Into Thinking Chocolate Helps Weight Loss.
Here's How, John Bohannon, GIZMODO, 27 May 2015.
 SYSY bettings:
 BBC Horizon Inside CERN (August 10, 2016): a very interesting documentary
on CERN and on the expextations related to the 750 GeV 'peak',
ending with the 5 August anouncement
that it was indeed a fluke.
(Unfortunatly the video is no longer available,
however the link is
left, so that interested people might find ways to access it.)
→ Available at Vimeo.com (February 2018)
 On misunderstanding (and cheating) based on pvalues,
plus phacking etc.,
see also here, and in particular
in the
additional related links,
in which you will learn that it is not just a question of journalists
and laypeople misunderstanding pvalues
(see in particular
what a famous book selling theorist writes about
"5sigma reliability",
showing total confusion)
 But someone still insists that
“Discoveries
have 5 sigmas”
because
“Quando una scoperta viene annunciata,
ogni dubbio deve essere stato fugato”
[By the way, you might be interested in the bet raised in
footnote 31 of
arXiv:1609.01668]

School
on Bayesian Analysis in Physics and Astronomy in Stellenbosch,
South Africa, November 2013:
→ see dedicated web page

"Probabilistic Reasoning in Frontier Physics", lecture at
HASCO Summer School 2013
(slides available at indico).

"Introducing Bayesian Reasoning in Measurements with a Toy Experiment"
Tutorial
at the 2015 IEEE I2MTC
Conference (Pisa, Italy, 11 May 2015):
Slides (pdf, 2.0M).

"Learning about probabilistic inference and forecasting
by playing with multivariate normal distributions
(with examples in R)"
arXiv:1504.02065:
pdf (3.3M).

"Bertrand 'paradox' reloaded
(with details on transformations of variables,
an introduction to Monte Carlo simulation
and an inferential variation of the problem)",
arXiv:1504.01361:
local pdf (4.9M)
"without rotations"
(the original arXiv pdf has two
rotated pages, a curious effect that disappears printing the file...
even on file − ??)
 Android app to generate the chords with the six methods
described in the paper: Bertrand.apk
(screenshot − ten chords each method)

"Probability, propensity and probability of propensities",
talk given at the
Garching's Bayes
Forum, February 2013 (slides of the presentation in the
Forum web page).
 This work has been presented a few other times alsewhere,
and finally turned into a paper in occasion
of MaxEnt16 and of the visit to
the Isaac Newton Institute for Mathematical Sciences
(see above).
 Is Most Published Research Wrong?:
YouTube
by Veritasium (very well done!);
 I Fooled Millions Into Thinking Chocolate Helps Weight Loss.
Here's How, John Bohannon, GIZMODO, 27 May 2015.
 Higgs boson bettings:
 BBC Horizon Inside CERN (August 10, 2016): a very interesting documentary
on CERN and on the expextations related to the 750 GeV 'peak',
ending with the 5 August anouncement
that it was indeed a fluke.
(Unfortunatly the video is no longer available,
however the link is
left, so that interested people might find ways to access it.)
→ Available at Vimeo.com (February 2018)
 On misunderstanding (and cheating) based on pvalues,
plus phacking etc.,
see also here
 But someone still insists that
“Discoveries
have 5 sigmas”
because “Quando una scoperta viene annunciata,
ogni dubbio deve essere stato fugato”
[By the way, you might be interested in the bet raised in
footnote 31 of
arXiv:1609.01668]

"Introducing Bayesian Reasoning in Measurements with a Toy Experiment"
Tutorial
at the 2015 IEEE I2MTC
Conference (Pisa, Italy, 11 May 2015):
Slides (pdf, 2.0M).

"Learning about probabilistic inference and forecasting
by playing with multivariate normal distributions
(with examples in R)"
arXiv:1504.02065:
pdf (3.3M).

"Bertrand 'paradox' reloaded
(with details on transformations of variables,
an introduction to Monte Carlo simulation
and an inferential variation of the problem)",
arXiv:1504.01361:
local pdf (4.9M)
"without rotations"
(the original arXiv pdf has two
rotated pages, a curious effect that disappears printing the file...
even on file − ??)
 Android app to generate the chords with the six methods
described in the paper: Bertrand.apk
(screenshot − ten chords each method)

School
on Bayesian Analysis in Physics and Astronomy in Stellenbosch,
South Africa, November 2013:
→ see dedicated web page

"Probabilistic Reasoning in Frontier Physics", lecture at
HASCO Summer School 2013
(slides available at indico).

"Judgement Leanings, a graphical way to hopefully improve
the perception and the communication of the `weights of evidence'
and of the 'intensities of beliefs",
poster at PESFoG2012:
Slides (110k).

"Probably a discovery: Bad mathematics means rough scientific communication", arXiv:1112.3620v2:
pdf (383k)
[html version, containing
also additional related links].
More links on pvalues (added from June 2015):

"Reti Bayesiane: da modelli di conoscenza a strumenti inferenziali e decisionali" (con Serena Cenatiempo e Aldo Vannelli), Notiziario Tecnico di
Telecom Italia,
2010, Numero 3, pp 1625
(copia locale, pdf, 836k).

"Improved iterative Bayesian unfolding",
arXiv:1010.0632v1:
ps.gz (384k),
pdf (321k).
 "On the Peirce's balancing reasons rule
failure in his "large bag of beans" example",
arXiv:1003.3659v1:
ps.gz (113k),
pdf (148k).
 "A defense of Columbo
(and of the use of Bayesian inference in forensics):
A multilevel introduction to probabilistic reasoning",
arXiv:1003.2086v2:
ps.gz (3.4M),
pdf (771k) and html
(the automatic translation from latex is quite poor).
 "L'inferenza probabilistica:
ruolo nelle scienze sperimentali e suggerimenti
per il suo insegnamento"
(Probabilistic inference: role in the experimental sciences
and suggestion for its teaching)
invited contribution
to the
Dossier Statistica,
TRECCANI Scuola, February 2010.
Copia locale [nel 2017 mi sono accorto che i link
del sito Treccani non erano più validi, e anche
quello che
trovavano ancora i
motori di ricerca puntava nel 'vuoto'.
 "On the socalled Boy or Girl Paradox",
arXiv:1001.0708v1:
ps.gz (93K) and
pdf (150K).
 Sleeping Beauty problem:
 "About the proof of the so called
exact classical confidence intervals.
Where is the trick?",
physics/0605140v2:
ps.gz (73K) and
pdf (125K).
 "Dalle osservazioni alle ipotesi scientifiche"
(From the observations to the scientific hypotheses), in Italian:
invited contribution
to
TRECCANI Scuola, Aprile 2006.
Versione locale.
 "Fits, and especially linear fits, with errors on both axes,
extra variance of the data points and other complications",
physics/0511182:
ps.gz (191K),
pdf (259K) and
HTML.
 The Fermi's Bayes Theorem
 “C'è statistica e statistica”,
risposta ad una domanda su “statistica gaussiana e statistica
bayesiana”
rivolta a Scienza per tutti dei Laboratori Nazionali di Frascati, pubblicata su Scaffali, 30 marzo 2005 (pdf, 45kB;
copia locale).
 "Telling the truth with statistics".
Lectures for the CERN Academic Training, 2125 February 2005.
(The title of lectures, commented
in the first lecture, was proposed by the organisers.)
 "From Observations to Hypotheses: Probabilistic Reasoning
Versus Falsificationism and its Statistical Variations".
Invited talk at the 2004 Vulcano Workshop on Frontier Objects
in Astrophysics and Particle Physics,
Vulcano (Italy) May 2429, 2004,
physics/0412148v2:
ps.gz (111k) or
pdf (259k) file;
HTML version.
 "Inferring the success parameter p of a binomial model
from small samples affected by background",
physics/0412069:
ps.gz (1.1M) or
pdf (403k) file;
HTML version.
 "Asymmetric uncertainties: sources, treatment and possible
dangers",
physics/0403086:
ps.gz (357k) or
pdf (236k) file;
HTML version.
 "Bayesian inference in processing experimental data: principles and basic applications", invited paper for Reports on Progress in Physics
[66 (2003)1383]
Preprint:
physics/0304102:
ps.gz (281k),
pdf (473k),
dvju (304k);
HTML version.
The same issue of RPP contains also a review by
Volker Dose on
"Bayesian inference in physics: case studies":
66 (2003)1421
(many related publications of Dose and collaborators on
data analysis can be found in the
IPP
Bayesian Data Analysis Group web page).
 "Bayesian model comparison applied to the
ExplorerNautilus 2001 coincidence data",
with Pia Astone and Sabrina D'Antonio,
grqc/0304096:
ps.gz (164k),
pdf (160k) and
html (160k) versions.
Published in Class. Quant. Grav. 20 (2003) 769.
Note added (December 2004): as stated in
my contribution to Vulcano04
(footnote 9),
this paper should be taken more for its methodological contents
than for the physical outcome.
 "Search for correlation between GRB's detected
by BeppoSAX and gravitational wave detectors
EXPLORER and NAUTILUS", astroph/0206431
(with Explorer/Nautilus and BeppoSAX Collaborations). Published
in Phys Rev D66 (2002) 102002.
 "Minimum bias legacy of search results",
Seventh Topical Seminar on "The legacy of LEP and SLC",
Siena, Italy, 811 October 2001.
hepph/0201070:
ps.gz (35k) or
PDF (143k) file.
Nucl Phys Proc Suppl 109 (2002) 148152.
 "Inference of rho and eta of the CKM matrix 
A simplified, intuitive approach" (based on a
series of lectures
given at the VI LNF Spring School "Bruno Touscheck",
Frascati, Italy, May 2001,
hepph/0107067):
ps.gz (1.1MB) or
PDF (1.6MB) file.
html version
 "2000 CKMTriangle Analysis: A critical review
with updated experimental inputs and theoretical
parameters" (with M. Ciuchini, E. Franco, V. Lubicz,
G. Martinelli, F. Parodi, P. Roudeau A. Stocchi)
hepph/0012308:
ps.gz (486k)
or PDF (1.2MB) file.
Published in JHEP 0107 (2001) 013
and highly cited.
I had some fun with referees also with this paper:
For more info and plots about the Unitary Triangle analysis, see
www.UTfit.org.
 ``Role and meaning of subjective probability: some
comments on common misconceptions'', talk at the
XXth International Workshop on
Bayesian Inference and Maximum Entropy Methods in Science and
Engineering, July 813,2000, Gif sur Yvette (Paris), France:
postscript file or
PDF file
(also at arXiv: physics/0010064). AIP Conference Proceedings (Melville) Vol. 568 (2001) 2330.
 ``Teaching Bayesian statistics in the scientific curricula'',
The ISBA Newletter (later renamed 'bulletin'),
Vol. 7,
No. 1 (March 2000) pp. 1819,
(local txt file).
 ``Uncertainties due to imperfect knowledge of systematic
effects: general considerations and approximate formulae''
(with Mirko Raso),
CERNEP/2000026, February 2000,
hepex/0002056:
postscript file or
PDF file.
A reduced version of this paper has become a part of
Chapter 12 of
Bayesian reasoning in data analysis.
 Workshop on
`Confidence Limits' at
CERN, Jan. 1718, 2000:
CERN document server.

Paper based on the invited talk
Confidence limits: what is the problem?
Is there the solution?
(eprint hepex/0002055):
postscript file or
PDF file
(Mathematica file
containing the Newton problem and the box example, see
also the errata below!).

Contributed paper with G. Degrassi: Constraining the
Higgs boson mass through the combination of direct
search and precision measurement result (an update
of our 1999 EPJ paper):
postscript file or
PDF file (also
hepph/0001269).

Short contribution during the pannel and discussion session:
Quantum mechanics and probability (with comments on
confidence intervals):
slides
 "Conclusions" of the Workshop.
 Proceedings published as CERN Report 2000005.
 20/08/2001: errata about the
numbers of the "original" problem to Newton.
 ``Sceptical combination of experimental results: General
considerations and application to epsilonprime/epsilon",
CERNEP/99139, October 1999,
hepex/9910036:
postscript file or
PDF file. This paper will remain
unpublished. People interested in sociology of science might
want to give a look at
1st referee report,
my replay,
2nd referee report,
2nd replay and
final verdict.
This paper, with little changes, has become
Chapter 11 of
Bayesian reasoning in data analysis.
Please refer to that book for citations.
 ``Inferring the intensity of Poisson processes at
limit of detector sensitivity (with a case study on
gravitational wave burst search)"
(with Pia Astone,
special acknowledgements to Maddalena...),
CERNEP/99126, August 1999,
hepex/9909047:
postscript file or
PDF file.
The substance of this paper has become a part of
Chapter 13 of
Bayesian reasoning in data analysis.
 "Teaching statistics in the physics curriculum.
Unifying and clarifying role of subjective probability",
invited contribution to the special
theme issue
of the American Journal of Physics on
Thermal and Statistical Physics (H. Gould and J. Tobochnik eds.):
 "Bayesian reasoning in high energy physics.
Principles and applications",
CERN Yellow Report 9903, July 1999 (vi + 175 pages).
 html version (latex2html translation
not really perfect...)
 ps/pdf version (by parts):
 Introduction + contents (8 pages):
ps file, PDF file;
 Part I  Subjective probability in Physics?
Scientific reasoning in conditions of uncertainty
(36 pages):
ps file, PDF file;
 Part II  Updated version of "Bayesian primer"
(78 pages):
ps file, PDF file;
 Part III  Other comments, examples and applications
(44 pages):
ps file, PDF file;
 Concluding matter + bibliography (15 pages):
ps file, PDF file;
(Also available at
CERN Document Server).
For FAQ's and clarifications see
here (all comments are welcome!).
A reviewed and extended version of
the report has been
published in 2003 by
World Scientific Publishing.
 ``Overcoming prior anxiety'',
invited contribution to the
monographic
issue of the Revista de la Real Academia de Ciencias
on Bayesian Methods in the Sciences (J.M. Bernardo ed.),
Año 1999, Vol. 93, Número 3;
local PDF file
(also available as
physics/9906048).
 Tutorial at Los Alamos Laboratory (March 13, 1999): for writeup of matter concerning
lessons see report based on CERN lectures.
The "six boxes" example, of which the
viewgraphs
are available in pdf format at LNAL, has been included
in the AJP paper ("Teaching statistics ..."),
see above in this list.
 ``Constraints on the Higgs Boson Mass from Direct Searches
and Precision Measurements''
(with G. Degrassi),
DFPD99/TH/02, February 1999,
hepph/9902226,
published in Eur. Phys. J. C10 (1999) 663675:
local postscript file or
PDF file of preprint.
You might be interested in what a `first rate' frequentist referee
would think about the paper: see
referee report.
Updated analysis as contribution to the Workshop on `Confidence
limits' (see above):
hepph/0001269.
 "Observation of a tossed
coin remaining vertical"
(15 Nov 1998),
plus other oddities.
 "Bayesian Reasoning versus Conventional Statistics in High
Energy Physics",
invited talk at the XVIII International Workshop
on Maximum Entropy and Bayesian Methods
(Maxent98):
postscript file or
PDF file
(also at Los Alamos xxx archive:
physics/9811046).
html
version at Caltech
 "Tutorial on measurement uncertainty" at
EC Summer School: Bayesian Signal Processing, Cambridge, July 1998.
(No proceedings; note in preparation with some
original material only presented there).
 "Jeffreys Priors versus Experienced Physicist Priors 
Arguments against Objective Bayesian Theory",
contributed paper to
"Valencia 6"
(1998): postscript file or
PDF file
(also at
physics/9811045).
 ``Bayesian Reasoning in High Energy Physics:
Principles and Applications'',
Academic Training at CERN (2529 May 1998)
Video and slides at the
CERN Document Server
(Like the staircases
at Hogwarts School, CERN web sites
like to move: if you find this link brocken, please help yourself
using your preferred search engine)
 ``Measurements errors and measurement uncertainty  critical
review and proposals for teaching'', Internal Report n. 1094,
Phys. Dept., May 1998:
postscript file or
PDF file,
in italian (86 pages):
this is my manifesto concerning the use of subjective
probability for the evaluation of uncertainty in measurement
as well as for a teaching method of physics laboratory
based on the Bayesian motto "Learning by Experience".
 ``A theory of measurement uncertainty based on conditional
probability''
ROME110791996, Nov 1996,
physics/9611016
(presented at JSM 1996 at Chicago, IL).
 ``Probability and Measurement Uncertainty in Physics 
a Bayesian Primer''
DESY95242, Roma1 N. 1070,
hepph/9512295
(original
postscript file also in DESY).
An updated version is contained in the CERN Yellow Report
9903 (see above).
A rivised version is now included in
Bayesian Reasoning
Several ideas of this paper have been applied
in the following notes by Mirko Raso
(including Bayesian unfolding and treatment of correlations due
to systematics):
 ZEUS96132,
"Measurement of
F_{2}ep
from the 1995 shifted vertex data":
ps and
pdf file.
 ZEUS98074,
"An update of the
F_{2}ep using the 1995
shifted vertex data":
ps
and pdf file;
html version.
 ZEUS98075,
"A measurement of the
sigma_{ep} for Q^{2}
< 0.3 GeV^{2}, as a by product of the
F_{2}ep measurement":
ps and
pdf file;
html version.
 Bayesian unfolding:
→Improved version
A curious introduction on Youtube.
Other papers
with some relevant issue on data analysis:
 ZEUS Note (1995) advertising ISO Guide:
postscript file; pdf file
 ``On the use of the covariance matrix
to fit correlated data'',
INSPIRE entry,
including the paper in pdf format
(local copy)
Preprint: DESY93175, December 1993,
Nucl. Instr. and Meth. in Phys. Res. A346 (1994) 306.
[Some authors call the effect described in the
paper "D'Agostini bias"]
 ``Quarkgluon jet separation in the photoproduction region
with a neural network algorithm''
(with G. Barbagli and D. Monaldi),
Roma1992, Feb. 1992;
Proc. of Workshop on Physics at HERA, Hamburg, Germany, October 2930, 1991;
 "Limits on the electron compositeness
from the Bhabha scattering
at PEP and PETRA".
25th Rencontre de Moriond on Electroweak Interactions
and Unified Theories, Les Arcs, France, Mar 411, 1990
 "Determination of alpha_s and sin^2theta(W) from the
R measurements at PEP and PETRA",
22th Rencontre de Moriond, Les Arcs, France, Mar 1521 1997,
Editions Frontières, pp. 325336.
 "Determination of α_{s} and the Z^{0}"
mass from measurements of the total hadronic crosssection
in e^{+}e^{} annihilation (with W. de Boer
and G. Grindhammer),
Phys.Lett.B 229 (1989) 160168.
 "Determination of alpha_s and sin^2theta(W) from measurements
of the total hadronic cross section in e+e annihilation"
(with CELLO Collaboration),
Phys. Lett. 183B (1987) 400.
Back to G.D'Agostini Home Page