Motto: "I have little patience with scientists who take a board
of wood, look at its thinnest part and drill a great number of
holes where drilling is easy." (Albert Einstein)
Warning[1]
In 2003, Petr Harmanec, a former colleague, visited the Catholic
University of Leuven as an "author" of Petr Hadrava's method of spectra
disentangling [2]. This method was developed
by Hadrava (published 1995,
A&AS 114,
393) between 1990 and 1993 as
an improvement of the method of cross-corellation (that is why the code
is named KOREL), first using simulated spectra and then demonstrated to
the colleagues at Ondrejov stellar dept. on the triple system 55 UMa.
They were distrustful of it until a similar method was published by
Simon and Sturm (1994), yet in the common publication on 55 UMa (Horn
et al. 1996,
A&A
309, 521) they agreed only to mention the applicability
of the method. The first publication based on the method was thus a study
on beta Cephei (Hadrava & Harmanec 1996,
A&A 315,
L401), followed by
several others next year. The method enabled to rise a common observational
project in the dept. on search for tidally induced oscillations in binaries (cf.
Holmgren
1996 and
Holmgren
et al. 1996), which was actually one of
aims already for the development of FOTEL code (cf. Hadrava 1986,
Hvar
Obs. Bull. 10, 1). The method of Fourier disentangling was further improved
by Hadrava for this purpose and entrusted to Harmanec, who was supposed
to take care for preparation of data for common publications, but he
presented it as his invention to others instead.
In 2005 in his letter to the dean of faculty, Harmanec explained the wrong
attribution of the method to him on the Leuven web-pages as a fault of his
host, which he was not aware of (despite of e-mails exchanged with him on
this matter in 2003).[3]
However, in their common publications on other stars of beta-Cephei type
they wrote similar incorrect claims.
For instance, in the paper Harmanec et al. (2004, A&A 422, 1013) they wrote:
"... we used ... KOREL ... to which we implemented calculation of weighted
rms errors of RVs derived via cross-correlation of disentangled spectrum
with the individual stellar spectra." In reality, the calculation of RVs
in the individual exposures and their deviations from the orbital solution
was included by Hadrava already in the early versions of KOREL to provide
an input into FOTEL for solution simultaneously with other published RVs
and light-curves and for treatment of errors of th/422
e solution. They are wrong
that the RVs are calculated using the cross-correlation. Harmanec tried to
include into the KOREL output calculation of rms from the RV differences,
what is unnecessary, and he did it with a mistake corrected by Hadrava.
In the paper A&A 455, 259 (2006) they even claim: "Although KOREL was not
developed to treat line-profile variations due to oscillations, Harmanec
et al. (2005) showed that the code is able to treat such a complex
combination of variability."
Harmanec started his stay at Leuven by creating "his manual" on Hadrava's
older code FOTEL for solution of radial-velocity and light curves. By the
method cut and paste he reproduced great part of the work "FOTEL - User's
Guide" written by Hadrava and examples of input files without giving proper
credit to the sources provided to the users by the author on internet
from the beginning of 90th. Harmanec copied great part of the LaTeX
source-file, reworded texts to the equations which he obviously did not
understand properly and sent it for comments to colleagues at Stellar dept.
of Ondrejov observatory as a
Release 1
of manual "written" by him (with an explanation that this is what he
intended to do for a long time). It was recommended to him by the author
that he should not publish such a confusing text and if he assumes to
have some notes he should write his comments separately with a proper
citation of the commented text. However, Harmanec argued that if he
has found the text on web, he may use it as he likes, but he cannot to
cite it, if it is not printed (despite from 1995, A&A 294, 135 till
1999 A&A 341, 867 he knew how to cite this electronic publication).
Harmanec then changed the title page to mimic a co-authorship and
published it as a
Release 2
on web anyway.[4]
To explain briefly the mistakes made by Harmanec, there follows comparison
of equations from "his manual" (denoted as Hec_#) and the final version
of the official
FOTEL 4 - User's Guide
by Hadrava (Had_#):
(Hec_1) = (Had_4) - form copied before author's final simplification using substitution (Had_5)
(Hec_2) = (Had_6)
(Hec_3) = (Had_7)
(Hec_4) = (Had_8) - Harmanec copied it with a mistype in the preliminary version
(Hec_5) = (Had_9) - the same mistype
(Hec_6) = (Had_10)
(Hec_7) = (Had_11)
(Hec_8)...(Hec_11) = equivalent of previous eqs. for simplified case k'=0
(and M=2 pi E). Because period changes in binaries are irregular, formulae
with expansion either in time or epoch are approximations only. (Cf. e.g. R.E.Wilson, Astrophys. and Sp. Sc. 296, 197.)
(Hec_12) = (Had_12)
(Hec_13) = (Had_13)
(Hec_14) = (Had_43)
(Hec_15), (Hec_16) = Harmanec's awkward excercise from calculus and simple
theory of atmospheres without direct relevance to the code Fotel
(Hec_17) = approximation mentioned in text
(Hec_18) = (Had_38), however, Harmanec gives completely wrong interpretation
to this formula - it is not the same quantity as in (Hec_17)
(Hec_19) = (Had_37)
(Hec_20) = given in the text of original
(Hec_21) - this generally obvious relation between the magnitudes of system
and its components is misleading in context of model explanation, where l.h.s.
contains constant parameters and r.h.s. variables
(Hec_22) = (Had_39)
(Hec_23) is a corrupted form of the final eq. for photometric model
(Had_49). Without the omitted terms the previous explanation is useless.
Equally, the eqs. (Had_14) and (Had_16), which are basic for fitting
of radial-velocities are omitted by Hec, as well as the whole passage
on errors.
After his return from Belgium, Harmanec promised and declared his
interest in a fair collaboration in future. He asked the author
for a series of lectures for his students and him to explain
the method of disentangling in details. But when Harmanec with his
student[5] failed to find a
spectrum of second component in one studied system, they submitted
a paper presenting the method of disentangling of telluric lines as
"their novel
approach". After protests by colleagues who knew that this
application was introduced by Hadrava many years before, they
canceled explicit claims about their authorship, but they anyway
published the confusing paper without citation of author's
description of the method.
A similar "novel approach" has been applied by P. Harmanec also
with respect to other colleagues. For instance, J. Kubat contributed
to our common paper A&A 319, 867 (1997) with calculation of
synthetic spectra, which he explained in a paragraph: "Synthetic
spectra for our analysis ... code SPEFO". P. Harmanec then used
other spectra calculated by J. Kubat in papers A&A 405, 1087
(2003) and A&A 416, 669 (2004) explaining them by practically
identical words without a co-authorship of the true author.
Due to persisting problems of this kind, it is no more possible
to provide freely the codes FOTEL and KOREL.
[1] The purpose of this web-page was
originally to explain the technical mistakes spread between users
of FOTEL and KOREL codes, which were, moreover, wrongly attributed
to the author. However, P. Harmanec and C. Aerts requested that
the criticism of these mistakes must be forbidden by the director
of the Astronomical Institute. Ethic commission established at the
Astronomical Institute to investigate the case then asked to explain
the ethic aspects more explicitly and let the judgement to the readers.
Cf. with the
Ethic codex
of Academy of Sciences or Ethical issues of
A&A.
The Ethic commission concluded, that the "Harmanec's" manual is
not a plagiate from the point of view of Law. The Czech
Author Law
(121/2000) does not define plagiarism at all.
[2] Motivation of this visit is
obvious from the annotation published at
http://cwisdb.cc.kuleuven.ac.be/research/P/team/team221218.htm
in the list of "All research projects of
research team Institute of Astronomy":
"Disentangling of high-resolution spectral lines profiles of massive
close binary stars with a pulsating component. - The purpose of the short
visit is to collaborate intensively on the disentangling of spectra of
5 targets in the seismic database at the Institute of Astronomy. These
targets turn out to be members of a close binary so that the orbital
and pulsational motion need to be deconvolved. The candidate
developed methodology for such disentangling. He will help applying
it to the 5 stars. The deconvolution has to be performed before any
seismic study can be started for these stars. - Project number 3E030041,
Responsible: Conny Aerts, Duration of the project 2003 09 - 2003 11."
[3]
The corresponding part of Harmanec's letter in translation reads:
"... In the materials enclosed by Associate Professor Hadrava to his letter
is also an undated copy of web-page of the Catholic University at Leuven,
where between the research projects of the Leuven Astronomical Institute
is mentioned my stay with explanation that I am the author of
the method of spectra decomposition. I got known about it only this September
and I fully understand that when Dr. Hadrava found this text on web
he must hold the suspicion that I do not behave fair. ...
I treat this point as quite principal and it is very unpleasant, that
there is only my word of honour that I was not aware of this message for
the whole period of my stay. There could be given an independent evidence
only by the Belgian participants of my seminar from the fall 2003, who
should confirm that I clearly from the very beginning emphasized the
Hadrava's authorship of both codes. ...
For myself I explain this text only by a sight from distance, probably
resulting from the fact that on most of then published applications of
both codes to particular objects we shared authorship with Assoc. Prof.
Hadrava."
[4]
Cf. with the rule III.1 of
Ethic codex
on co-authorship. It was found to be a "common habit" by Scientific
Council of the Faculty of Mathematics and Physics (cf. point IV.2. of its
minutes).
Really, similar cases occured in other Czech universities recently
(cf., eg.,
1,
2,
3).
It is explained and causal relation between such habits and the use
of quantitative criteria in evaluation of scientists
(advocated by
some deans) is discussed in
this paper.
[5] Cf. note 4 to
Ethic
framework declared by the Ministery of Education.