text
stringlengths 2
6.14k
| file_name
stringclasses 10
values | node_id
stringlengths 36
36
|
|---|---|---|
and a family of smooth maps φα: (R∗
+)2→R∗
+×Rindexed by α∈ {1,0,−1}
defined by
φα(a,b) ={
(a,αlogb)ifα̸= 0
(1,logb)ifα= 0
Letǫ(i,j)be the sign of i−jfor(i,j)∈N2
n.
Definition 2.15. For any A∈Mn(R+),such that πPC(A) = (ci,j)(i,j)∈N2n
andπNR(A) = (di,j)(i,j)∈N2n,we define the matrix Φ(A) = (bi,j)(i,j)∈N2nwith
coefficients in Gby
∀(i,j)∈N2
n,bi,j=φǫ(i,j)(ci,j,di,j). | 2312.00001v1.pdf | 6c5333b7-4d49-4380-8dce-7c6ef8a491c3 |
12 JEAN-PIERRE MAGNOT
For the sake of the clarity of next theorem, we note by PCn(R∗
+)the
group of multiplicative pairwise comparisons matrices wit h coefficients in
R∗
+and byPCn(R)the group of additive paiwise comparisons matrices with
coefficients in R.
Theorem 2.16. Φ(Mn(R+))is identified as a Lie group with PCn(R∗
+)×
PCn(R)×Rn.With the notations of the last definition, the identification is
given by
bi,j↦→(ci,j,ǫ(i,j)di,j)∈PCn(R∗
+)×PCn(R)
and
A↦→(di,i)i∈Nn∈Rn.
The proof is straightforward.
Remark 2.17.The proof is straightforward.
Remark 2.17. In[29], for any Lie group G,we described pairwise comparisons
matrices PCn(G).We have here
PCn(R∗
+)×PCn(R) =PCn(G)
for the struture group already announced
G=TR∗
+=R∗
+×R.
As a consequence of this remark, applying one main construct ion of [29],
we get:
Theorem 2.18. Mn(R∗
+)projects to connections on ∆n×(R∗
+×R)discretized
by their holonomies, and the kernel of this projection is the space of pairwise
comparisons matrices (ai,j)(i,j)∈N2n∈Mn(R∗
+)such that
∀(i,j)∈N2,i̸=j⇒ai,j= 1.
We call pure pairwise comparisons matrix a (maybe non reciprocal) matrix
(ai,j)(i,j)∈N2n∈Mn(R∗
+)such that ∀i∈Nn,ai,i= 1.Remark 2.19. We may have made the choice to consider in our study only
pure pairwise comparisons matrices. This would have given a more elegant
formulation of last theorem, fully identifying discretize d connections with the
space of pairwise comparisons that we would have considered . However, for
the sake of completeness of the framework that we propose her e in view of
modelizations, we feel the need to consider the assumption ai,i̸= 1as non
void for applications. An example is briefly developped in th e appendix, before
follow-up investigations.
2.6.Inconsistency on Mn(R∗
+).Applying Theorem 2.18, one can define
consistent (maybe) non-reciprocal pairwise comparisons m atrices by the discretized
flat connections, the following way:
Definition 2.20. | 2312.00001v1.pdf | 655336c0-11c5-4759-8f37-20ebbcf73783 |
LetA∈Mn(R∗
+).Then the (maybe non reciprocal) pairwise
comparisons is consistent if it is a pure pairwise compariso ns matrix that
projects to a flat discretized connection on ∆n×(R∗
+×R).This space is
noted by CMn(R∗
+). | 2312.00001v1.pdf | bedb064c-3d86-4bd1-b66b-90711e3f8c1f |
ON RANDOM PAIRWISE COMPARISONS MATRICES AND THEIR GEOMETRY . 13
This definition clearly extends the notion of consistent (an d reciprocal)
pairwise comarisons matrix. In the sequel, we shall make no d ifference in
the notations between CMn(R∗
+)as a set of pairwise comparions matrices
with coefficients in Gand as a set of non-reciprocal pairwise comparisons in
Mn(R∗
+)because of the bijection mentioned before between the two su bsets,
and we can define
Definition 2.21. Ainconsistency index is a map ii:Mn(R∗
+)→R+
such that
ii−1(0) =CMn(R∗
+).
Ainconsistency indicator is an reciprocity index with values in [0,1].
This definition clearly defines inconsistency indexes (resp . indicators) ii
onMn(R∗
+)that restrict to inconsistency indexes (resp.indicators) onPCn
along the lines of Definition 2.4.
Example 2.22. The discretized Yang-Mills functional along the lines of [28]
and forG=TR∗
+determines an inconsistency index that extends the known
one inPCn.
Example 2.23. Letdbe a distance on Mn(R∗
+).Then, for any γ∈R∗
+,the
map
A∈Mn(R∗
+)↦→(
d(A,CM n(R∗
+)))γ
is an inconsistency index.
One can define then a total index from an inconsistency index a nd a non
reciprocity index.Let us finish by the following (again, dir ect) proposition:
Proposition 2.24 (Geometric mean method in non-reciprocal setting) .Let
us note by π′:PCn(G)→PCn(G)the map defined by the following formulas:
ifA= (ai,j)(i,j)∈N2n∈PCn(G),with
∀(i,j)∈N2n∈PCn(G),with
∀(i,j)∈N2
n,ai,j= (ci,j,di,j)∈R∗
+×R,
then the image
B= (bi,j)(i,j)∈N2n=π′(A),∀(i,j)∈N2
n,bi,j= (c′
i,j,d′
i,j)∈R∗
+×R
is such that:
(c′
i,j)(i,j)∈N2n=πCPC(
(ci,j)(i,j)∈N2n)
and
∀(i,j)∈N2
n, d′
i,j=1
n(n∑
k=1di,k−n∑
l=1dl,j)
. | 2312.00001v1.pdf | 7f55205e-b5d4-4113-b951-7fc7ae00ca04 |
k−n∑
l=1dl,j)
.
Then
•the restriction of π′toPCn(R∗
+)is equal to πCPC, wherePCn(R∗
+)
is understood as the set of matrices in PCn(G)which coefficients are
such that R−component equal to 0,
•and moreover, π′is a smooth projection from PCn(G)toCMn(R∗
+). | 2312.00001v1.pdf | 5306e55c-0cf0-4ebc-a8d6-5398c3a548bb |
14 JEAN-PIERRE MAGNOT
Proof. The first point follows from the definition of π′in the coefficients c′
i,j.
Analysing now the formula for the coefficient d′
i,j,we recgnize a geometric
mean method for an additive pairwise comparisons matrix, se e [43] for a
review. Therefore, the obtained matrix Bis a projection to consistent
pairwise comparisons matrices with coefficient in G. □
Therefore, we can propose the following projection:
Definition 2.25. πMC=π′◦Φ.
Theorem 2.26. The following diagram is commutative:
Mn(R∗
+) PCn
MCn CPCnπPC
πCM πCPC
πPC
Proof. By direct calculations of the coefficients of πCPC◦πPC(A)on one side
and ofπPC◦πCM(A)in the other side. □
3.Random pairwise comparisons
3.1.The settings.□
3.Random pairwise comparisons
3.1.The settings. Let us now consider coefficients of matrices in Mn(R∗
+)
as random variables in a sample probability space Ωwith values in Mn(R∗
+)
and we call it random pairwise comparisons matrix. The margi nal laws on
each coefficient ai,jof a random pairwise comparisons matrix (ai,j)(i,j)∈N2ndetermine the probability law of each coefficient. | 2312.00001v1.pdf | 7ed4d673-c617-41eb-9b80-ef57827b5da1 |
For the sake of motivation and in order to precise accessible situations
that are relevant to this picture, before the more developpe d example in the
appendix, we precise that typically the coefficients ai,jfollow the following
laws:
•a Gaussian law on the multiplicative Lie group R∗
+if the evaluations
are honnestly made by experts
•an uniform law withe respect to an interval on R∗
+, with respect
to the additive structure, if the errors are made by human mind
corrections or cut-off approximations made with respect to t he limits
of the measurement procedures and without real link to the pr oblem.
These two extreme cases are simple consequences of the celeb rated Benford’s
law [3], but the law of the random pairwise comparisons matri xAcan be
anything as well as the laws of the coefficients ai,j.
Remark 3.1. We have here to remark that, by randomness, the use of non-
reciprocal pairwise comparisons become natural, even in “h onest” evaluations
due to the natural approximation of the scores.Let us note by RMn(R∗
+)the space of random matrices and, extending
the notations, the first letter Rwill mean “random”. For example, RPCnis
the set of random pairwise comparisons matrices with suppor t inPCn. | 2312.00001v1.pdf | f8b51522-60a1-49b4-bc0b-5e2d78fdb0aa |
ON RANDOM PAIRWISE COMPARISONS MATRICES AND THEIR GEOMETRY . 15
Proposition 3.2. There is an inclusion map Mn(R∗
+)→RMn(R∗
+)yay
identifies (ai,j)(i,j)∈N2nwith⨂
(i,j)∈N2nδai,j(Dirac measures).
3.2.Stochastic analogs to deterministic constructions. In this section
and till the end of the paper, all the (finite dimensional, geo metric) mappings
are assumed measurable with respect to the considered measu res or probabilities.
This is authomatically fulfilled for continuous (and hence f or smooth) mappings
with sourve and target in finite dimensional manifolds.
3.2.1. Stochastic extension of projections. Let us consider the projections
already defined: πCPC, πPCandπCMand our remark is the following:
since these mappings are measurable, they extend to mapping s between the
corresponding probability spaces.That is, we have three pu sh-forward maps
(πCPC)∗:RPCn→RCPC n,
(πPC)∗:RMn(R∗
+)→RPCn
and
(πCM)∗:RMn(R∗
+)→RCMn
and their right inverse (that is, roughky speaking, π∗◦π∗=Idin the
adequate domain) given by the pull-back mappings:
(πCPC)∗:RCPC n→RPCn,
(πPC)∗:RPCn→RMn(R∗
+)
and
(πCM)∗:RCMn→RMn(R∗
+)
3.2.2. Stochastic indexes. We can now given the corresponding notions of
the the various indexes of indicators in a stochastic contex t.
Definition 3.3. Astochastic reciprocity index (orsr-index for short)is
a mapsri:RMn(R∗
+)→R+such that
sri−1(0) =RPCn. | 2312.00001v1.pdf | 20e48fa6-175b-4f6d-8afa-4ce67675065b |
Astochastic reciprocity indicator is a sr-reciprocity index with values in
[0,1].
The same approach by the support of the probability is propos ed for each
of the two following notions:
Definition 3.4. Astochastic inconsistency index (orsi-index for short)
is a map sii:RMn(R∗
+)→R+such that
sii−1(0) =CMn(R∗
+).
Ainconsistency indicator is an sr-index with values in [0,1]. | 2312.00001v1.pdf | 6909b2b3-1a3d-4f31-b235-dce1d2112699 |
16 JEAN-PIERRE MAGNOT
Definition 3.5. Anstochastic total index (orst-index for short) is a
mapsti:Mn(R∗
+)→R+such that
sti−1(0) =CPCn.
Astochastic total indicator is a st-index index with values in [0,1].
Theorem 3.6. Letibe a reciprocity index (resp. inconsistency indicator,
resp. total indicator). Then it defines sia sr-index (resp. si-index, resp.
st-index) by the formula:
si:X∈RMn(R∗
+)↦→∫
i(A)dX(A).
Moreover, sirestricts to ion Dirac random pairwise comparisons matrices.
The proof is obvious while considering the support of the map si.
Remark 3.7.The proof is obvious while considering the support of the map si.
Remark 3.7. We need an inconsistency indicator in this formula because
the map ineeds to be, in fact, in the class L1(dX)for anyX.This theorem
extends to any bounded inconsistency index, but we prefer here to use the
“normalized” approach proposed in [18]even if the resulting stochastic index
siwill be very exceptionaly a stochastic indicator. Since siis bounded, it will
be possible to normalize it by scaling in applications if nec essary.
3.3.Minimizing inconsitency: a sketch of two methods. We choose
here to expose the main steps of methods that one cn use for find ing a
consistent, maybe non reciprocal, or a reciprocal, or a cons istent and reciprocal
PC matrix. For the reasons explained in the presentation of t he organisation
of this work, the aspectts related to problems of convergenc e, uniqueness,
regularity are postponed to future works.
3.3.1. Minimization of a stochastic index.3.3.1. Minimization of a stochastic index. The main idea consists in reducing
the random matrix Xto a random matrix with support in CMn, PCnor
CPCndepending on the nature of the bounded index ichosen, by minimizing
the associated stochastic index
si:X∈P(Mn(R∗
+))↦→∫
RMn(R∗
+)i(A)dX(A).
This minimization problem is posed on the convex set P(Mn(R∗
+))where the
extremels are Dirac type measures, that is, elements of Mn(R∗
+).Beside the
necessary developments of adequate analysis on the space Mn(R∗
+), let us
point out some issues that cannot be treated in a “black box” t hat technicaly
solves the problem, but that envolves conceptual constrain ts for decision
makers:
•This method can be adapted to the constraints imposed by the
decision problem. | 2312.00001v1.pdf | fd558387-3f77-430a-9559-6cfff6db60c4 |
Indeed, not only the choice of the indicat ori,
but also a restricted space of random PC matrices, can be left to the
choice of the decision maker which makes this method quite fle xible. | 2312.00001v1.pdf | bd79fa80-ac5e-4b0b-8bef-db57a0ed6ac1 |
ON RANDOM PAIRWISE COMPARISONS MATRICES AND THEIR GEOMETRY . 17
•Besides this flexibility, the regularity of the functional is itself a
technical issue and we hope that we will be able to succeeed in
our future works that will precise the necessary regularity of the
stochastic index in order to have a numericaly robust method .
•The matrix Yobtained as a minimizer of siis still random, unless a
way to impose to find it on the border “∂RMn(R∗
+) =Mn(R∗
+)′′is
imposed.
3.3.2. Shortest distance and optimal transport. Therefore, one has to address
the problem of finding a (non random) pairwise comparisons ma trix from a
random one.A direct approach, will be discussed next in a geo metric way,
may consist in taking the expectation value of the matrix A.But it seems
to us a bit drastic, because the first goal of a consistencizat ion remains in
changing as few as possible (in a way that has to be controled numericaly)
the coefficients of the PC matrix. If this condition of minimal change is not
guaranteed, an (extreme) dictatorial behaviour would cons ist in imposing
a pre-determined consistent and reciprocal PC matrix to any problem. We
leave to the reader the production of a concrete example that can be modelled
with such approach.
Anyway, we have to propose an approach that controls the “de- randomization”
of the random PC matrix. Let us recall that any space of probab ilities is
a compact metric space, and for probabilities on a Riemannia n manifold,
the Wasserstein distance W2related to optimal transport problems is an
improved metric for which one has already a panel of properti es at hand, see
e.g. [44].[44]. Therefore, we can propose the following procedur e given random
pairwise comparisons matrix A:
(1) choose a compact subsetBinMn(R∗
+)such that∫
1BdA >1−ǫ
and normalize 1BdAas well as the Riemannian metric to make the
volume of Bwith respect to the new Riemannian metric equal to 1.
The values that are outside the set Bare typicaly non-representative,
or meaningless, values for the modelized problem.
(2) Consider the target space Tof Dirac measures in the desired space
(CM, PC orCPC) inB.
(3) Find C∈Tsuch that
W2(A,C) =minA′∈TW2(A,A′)
Let us now analyze, again from the viewpoint of the decision m aker, the
proposed method:
•This method furnishes a way to define a deterministic matrix w ith
respect to the minimization of a distance. This approach is t hen
conceptualy affordable. | 2312.00001v1.pdf | 6c6c76ba-9113-4842-97f3-3864ae0c3a8f |
This approach is t hen
conceptualy affordable.
•Moreover, the decision process can be acheived by the produc tion of
weights(wi)i∈Nnwhich explicitely proceed to a ranking, | 2312.00001v1.pdf | 54e3065c-4dbe-4795-b46d-1dc074044347 |
18 JEAN-PIERRE MAGNOT
•but this method does not depend on any index. This means that t he
minimization of the stochastic index
si:X∈P(Mn(R∗
+))↦→∫
RMn(R∗
+)i(A)dX(A)
has to be anyway performed before this second method (on the s pace
RMn) or after this methode (under the constrat of a minimization
on the space of Dirac-type measures) if the decision maker wa nts to
get a ranking really appropriate for the problem under consi deration.
4.A formal geometric picture for random pairwise
comparisons and geometric concerns
Let us start with a simplified picture.This is a bundle-like e xtension
construction, but due to tthe fact that we did not precise whi ch topology
we prefer to define on the space of random PC matrices, and also because
the probability space in general is not a manifold, in the bes t cases a metric
space [33] or a diffeological space [30, 11], the full geometr ic or topological
structures will remain non precised in this communication. We shall only
mention the following diagram:
RMn(R∗
+)Mn(R∗
+)
Mn(R∗
+) CPCnπCPC,∗
E E
πCPCwhereEis the (multiplicative) expectation
value.
Proposition 4.1. Last diagram is well defined and it commutes.
Proof. The proof is standard for additive pairwise comparisons, wh ere the
expectation value is taken in a standard way, and where the pr ojection
πCPCreads as a mean of the reciprocal coefficients followed by an ar ithmetic
mean on off-diagonal coefficients. | 2312.00001v1.pdf | cd2936bc-cf67-42f3-aaa7-c065e2781b77 |
the multiplicative expect ation value of the
multiplicative PC matrix is the exponential of the expectat ion value of this
additive PC matrix, while the geometric mean projection fol lows the same
rules. □
Therefore, we have exhibited a split of the space of random pa irwise
comparisons matrix into a fiber bundle where
(1) The projection associates at each random matrix Athe matrix E(A)
of the (multiplicative) expectation values of its coefficite nt, and the
geometric mean method is, in the sense that we have explained ,
compatible with this fibration
(2) Therefore the fibers can be identified with multiplicativ e random
pairwise comparisons matrices with expectation values equ al to1
where1is the neutral element of Mn(R∗
+).It has a “global section”
defined through Dirac measures. Each fiber can be heuristical y | 2312.00001v1.pdf | 21d74c74-10cc-4e00-867a-7d4acf254eb0 |
ON RANDOM PAIRWISE COMPARISONS MATRICES AND THEIR GEOMETRY . 19
identified as the stochastic variations with trivial expect ation values
of the given deterministic PC matrix on the base.
(3) Even if the base is a finite dimensioanal Lie group, the fibe rs are
truely infinite dimensional and do not carry a structure of e. g. Banach
manifold as in many known cases.
5.Outlook
The aim of this paper was to describe safely the basic necessa ry setting
for random pairwise comparisons, in order to adapt it and int erpret it in
view of applications. After this description, we see four di rections for future
investigations.
(1) The case when the non reciprocal pairwise comparisons ma trix is not
pure, that is, when the diagonal coefficients are not neutral, needs a
more deep treatise, maybe inspired by concepts in physics.
(2) The two procedures for reduction of a stochastic index th at we proposed
have to be implemented, both in terms of optimal control theo ry.This requires to gather the known results that are related, a nd produce
a proper algorithm which convergence is fully studied.
(3) For this, the geometric structures (metric geometry, di ffeology... ) of
the ambient space of random matrices as well as the useful geo metric
objects for the implimentation of the algorithms already me ntioned
have to be precised. Moreover, we get another motivation fro m the
fact that the two proposed types of structures, metric space s and
diffeologies, have not been deeply compared yet and such a stu dy in
a given example would be of theoretical interest.
(4) Finally, we did not address in this paper the question of m aximal
inconsistency of a submatrix of a random PC matrix.
Appendix: One motivating example on exchange rates
Let us consider three currencies A, B and C and let us explain w hat
happens in the corresponding exchange rates. | 2312.00001v1.pdf | 5f4e5e81-fef1-4de4-9498-5226e0937902 |
A B
CeA,CeA,B
eB,C
For almost any existing currency, one can see that
•The exchange rates eK,L,for(K,L)∈ {A,B,C}2,truely gives a
reciprocal ratio between the currencies, i.e. ∀(K,L)∈ {A,B,C}2,eL,K=
e−1
K,L.
•a loop that changes AtoB, B toCandCtoAmay give an
inconsistency
eA,BeB,CeC,A̸=eAA= 1. | 2312.00001v1.pdf | a8d3ad8a-ddc4-4246-8a9b-3e0e314f7fea |
20 JEAN-PIERRE MAGNOT
Moreover, any individual who wants to exchange e.g. AforBneeds to
contact a currency exchange agency which applies charges an d taxes to
exchanges, which implies
∀(K,L)∈ {A,B,C}2,eL,K̸=e−1
K,L,
and storing currencies on a bank account generates costs, wh ich reads as
∀L∈ {A,B,C},eL,L̸= 1.
These two items are non reciprocal pairwise comparisons.
References
[1] Albeverio, S.; Zegarlinski, B.; Construction of conver gent simplicial approximations
of quantum fields on Riemannian manifolds Comm. Mat. Phys. 13239-71 (1990)
[2] Bartl, D.; Ramík, J.; A new algorithm for computing prior ity vector of pairwise
comparisons matrix with fuzzy elements. Information Sciences ,615103?117 (2022).
[3] Benford, F.; The law of anomalous numbers. Proc. Am. Philos. Soc.Proc. Am. Philos. Soc. 78no 4, 551?572
(1938)
[4] Bozóki, S.; Rapcsák, T.; On Saaty’s and Koczkodaj’s inco nsistencies of pairwise
comparison matrices; J. Glob. Optim. 42no.2 (2008) 157-175
[5] Colomer, J-M.; Ramon Llull: from ?Ars electionis? to soc ial choice theory. Social
Choice and Welfare 40no2), 317?328 (2011).
[6] Crawford, R.; Williams, C.; A note on the analysis of subj ective judgement matrices.
Journal of Mathematical Psychology ,29387–405 (1985)
[7] A. Darko, A.; Chan, A. P. C.; Ameyaw, E. E.; Owusu, E. K.; Pä rn, E.; D.
J. Edwards, D. J.; Review of application of analytic hierarc hy process (AHP) in
construction. International Journal of Construction Management 19no5, 436? | 2312.00001v1.pdf | 626696fe-dbc3-4a88-85bb-ecb64deb37cb |
International Journal of Construction Management 19no5, 436? 452
(2019)
[8] Farinelli S., Vasquez S. Gauge invariance, geometry and arbitrage. J. Invest.
Strategies .1no 2, 2366 (2012).
[9] Fülöp, J.; A method for approximating pairwise comparis ons matrices by consistent
matrices J. Global Optimization 42, 423-442 (2008)
[10] Gacula Jr., M.C.; Singh, J.; Statistical Methods in Food and Consumer Research .
Academic Press (1984).
[11] Goldammer, N.; Magnot, J-P.; Welker, K.; On diffeologie s from infinite
dimensional geometry to PDE constrained optimization. To a ppear in Contemporary
Mathematics , volume dedicated to the proceedings of the special session “Recent
advances in diffeologies and their applications”, AMS-EMS- SMF congress, Grenoble,
France, July 2022.[12] Gromov, M.; Metric structures in Riemannian and non-Riemannian spaces , 2nd
ed. Birkauser, (1997).
[13] Hyde, R.A.; Davis, K.; Military applications of the ana lytic hierarchy process.
International Journal of Multicriteria Decision Makin g,2no3), 267 (2012)
[14] Illinski, K.; Gauge geometry of financial markets. J. Phys. A, Math. Gen. 335?14
(2000).
[15] Kobayashi, S.; Nomizu, K. (1963-1969) ‘ Foundations of Differential Geometry I and
IIWiley classics library ISBN:978-0471157335 and 978-04711 57328
[16] Koczkodaj, W.W. (1993) A new definition of consistency o f pairwise
comparisons, Math. Comput. Modelling 879-84 (1993).Comput. Modelling 879-84 (1993).
[17] Koczkodaj, W.; Orlowski, M.; An orthogonal basis for co mputing a consistent
approximation to a pairwise comparisons matrix. Computers and Mathematics with
Applications 34no10, 41?47 (1997) | 2312.00001v1.pdf | 3f585e62-1f49-4cb9-b145-70f1379544e5 |
ON RANDOM PAIRWISE COMPARISONS MATRICES AND THEIR GEOMETRY . 21
[18] Koczkodaj, W.W.; Magnot, J-P.; Mazurek, J.; Peters, J. F.; Rakhshani, H.;
Soltys, M.; Strza lka, D.; Szybowski, J.; Tozzi, A. (2017) On normalization of
inconsistency indicators in pairwise comparisons Int. J. Approx. Reasoning 8673-79.
https://doi.org/10.1016/j.ijar.2017.04.005
[19] Koczkodaj, W.W.; Smarzewski, R.; Szybowski, J. (2020) On Orthogonal
Projections on the Space of Consistent Pairwise Comparison s Matrices Fundamenta
Informaticae ,172, no. 4 379-397 https://doi.org/10.1016/j.ijar.2017.04. 005
[20] Koczkodaj, W.W.005
[20] Koczkodaj, W.W.; Szybowski, J.; Wajch, E., Inconsiste ncy indicator maps on groups
for pairwise comparisons, International Journal of Approx imate Reasoning, 69(2)
81-90, 2016.
[21] Koczkodaj, W.W.; Szwarc, R. (2014) Axiomatization of I nconsistency Indicators for
Pairwise Comparisons, Fundam. Inform. 132no 4, 485-500 (2014)
[22] Ku?akowski, K. Grobler-D?bska, K; W?s, J.; Heuristic r ating estima- tion: geometric
approach. Journal of Global Optimization ,62no3, 529?543 (2015).
[23] K. Kulakowski, J. Mazurek, and M. Strada. On the similar ity between ranking
vectors in the pairwise comparison method. Journal of the Opera- tional Research
Society ,01?10 (2021).Journal of the Opera- tional Research
Society ,01?10 (2021).
[24] Lahby, M.; Leghris, C.; Adib, A.; Network Selection Dec ision based on handover
history in Heterogeneous Wireless Networks; International Journal of Computer
Science and Telecommunications 3no 2, (2012) 21-25
[25] Liberatore, M.J.; Nydick, R. L.; Group decision making in higher education using
the analytic hierarchy process. Research in Higher Education ,38no5, 593?614
(1997).
[26] Mack, G.; Gauge theory of things alive Nucl.Phys. B, Proc.Suppl. 42923-925 (1995)
[27] Magnot, J-P.; A mathematical bridge between discretiz ed gauge theories in quantum
physics and approximate reasoning in pairwise comparisons Adv. Math. Phys. 2018
(2018), Article ID 7496762, 5 pages https://doi.org/10.11 55/2018/7496762
[28] Magnot, J-P.; Remarks on a new possible discretization scheme for gauge theories.
Int. J. Theor. Phys. 57no. 7, 2093-2102 (2018)
[29] Magnot, J-P. ; On mathematical structures on pairwise c omparisons matrices with
coefficients in a group arising from quantum gravity Helyion 5(2019) e01821
https://doi.org/10.1016/j.heliyon.2019.e01821
[30] Magnot, J-P.; On the differential geometry of numerical schemes and weak solutions
of functional equations. Nonlinearity 33, no. 12, 6835-6867 (2020)
[31] Magnot, J-P.; Mazurek, J.; Cernanova, V.; A gradient me thod for inconsistency
reduction of pairwise comparisons matrices Int. J. Approx. Reasonning 15246-58
(2023)
[32] Peterson, G.L.; Brown, T.C.; Economic valuation by the method of paired
comparison, with emphasis on evaluation of the transitivit y axiom.Land Economics
240-261 (1998)
[33] Prokhorov, Yu.V.; Convergence of random processes and limit theorems in
probability theory. Theor. Prob. Appl. 1157-214 (1956)
[34] Sen, S.; Sen, S.; Sexton, J.C.; Adams, D.H.; A geometric discretisation scheme
applied to the Abelian Chern-Simons theory Phys.Rev. E 613174-3185 (2000)
[35] Rovelli, C.; Vidotto, F. (2014) Covariant loop quantum gravity Cambridge university
press https://doi.org/10.1017/CBO9781107706910
[36] Saaty, T.; A scaling methods for priorities in hierarch ical structures; J. Math.
Psychol. 15(1977) 234-281
[37] Sasaki, Y.; Strategic manipulation in group decisions with pairwise comparisons: A
game theoretical perspective. European Journal of Operational Research 304no3,
1133?1139 (2023).
[38] Taylor, A.D. | 2312.00001v1.pdf | 47949e48-d23e-49d7-960b-3eceff98c363 |
[38] Taylor, A.D.; Social Choice and the Mathematics of Manipulation . Outlooks.
Cambridge University Press (2005). | 2312.00001v1.pdf | 1c003296-0216-4e45-8a07-b76e8b2cd8a3 |
22 JEAN-PIERRE MAGNOT
[39] Thurstone, L.L.; The Method of Paired Comparisons for S ocial Values. Journal of
Abnormal and Social Psychology , pages 384?400 (1927).
[40] Urbaniec, M.; So?tysik, M.; Prusak, A.; Ku?akowski, K. ; Wojnarowska, M.;
Fostering sustainable entrepreneurship by busi- ness stra tegies: An explorative
approach in the bioeconomy. Business Strategy and the Envir onment 31no1,
251?267 (2022).
[41] Sengupta, A.N. (2011) Yang-Mills in Two Dimensions and Chern-Simons in Three
, inChern- Simons Theory: 20 years after , Editors Jorgen Ellegaard Anderson,
Hans U. Boden, Atle Hahn, and Benjamin Himpel. AMS/IP Studie s in Advanced
Mathematics (pp. 311-320).
[42] Sengupta, B.; Tozzi, A.; Cooray, G.K.; Douglas, P.K.; Douglas, P.K.; F riston, K.J.; Towards a
Neuronal Gauge Theory. PLoS Biol. 14no 3, e1002400 (2016).
[43] Szybowski, J; Kulakowski, K.; Ernst, S.; Almost optima l manipulation of a pair of
alternatives ArXiv:2304.03060
[44] Villani, C.; Optimal transport, old and new Springer (2006)
[45] Wajch, E., From pairwise comparisons to consistency wi th respect to a group
operation and Koczkodaj’s metric. Int. J. Approx. Reason. 10651?62 (2019).
[46] Whitney, H. (1957) Geometric Integration Theory , Princeton University Press,
Princeton, , J.NJ.
[47] Young K. Foreign exchange markets as a lattice gauge the ory.Am. J. Phys. 67no
10, 862?868 (1999). | 2312.00001v1.pdf | ab87da38-aff9-46e2-81ef-6e5548e419b7 |
J. Phys. 67no
10, 862?868 (1999).
[48] Yuan, R.; Wu, Z.; Tu, J.; Large-scale group decision-ma king with incomplete fuzzy
preference relations: The perspective of ordinal consiste ncy.Fuzzy Sets and Systems
454100?124 (2023).
CNRS, LAREMA, SFR MATHSTIC, F-49000 Angers, France, and, Ly cée
Jeanne d’Arc, Avenue de Grande Bretagne, F-63000 Clermont- Ferrand
Email address :[email protected] | 2312.00001v1.pdf | 85240b36-5cea-4fd8-bfe4-cea2086dcf6c |
arXiv:2312.00002v1 [math.RT] 22 Jul 2023The Discontinuity Group
of a Locally Bounded Homomorphism
of a Lie Group into a Lie Group
Is Commutative
A. I. Shtern∗,∗∗,∗∗∗,1
∗Moscow Center for Fundamental and Applied Mathematics, Mos cow, 119991 Russia
∗∗Department of Mechanics and Mathematics, Moscow State Univ ersity, Moscow, 119991 Russia
∗∗∗Scientific Research Institute for System Analysis of the Rus sian Academy of Sciences
(FGU FNTs NIISI RAN), Moscow, 117312 Russia
E-mail:[email protected]
July 20, 2023
Abstract
We prove that the discontinuity group of every locally bound ed ho-
momorphism of a Lie group into a Lie group is not only compact a nd
connected, which is known, but is also commutative.1 Introduction
Recall that a (not necessarily continuous) homomorphism πof a topological
groupGinto a topological group His said to be relatively compact if there is a
neighborhood U=UeGof the identity element eGinGwhose image π(U)has
compact closure in H.
Obviously, a homomorphism into a locally compact group is re latively com-
pact if and only if it is locally bounded , i.e., there is a neighborhood UeGwhose
image is contained in some element of the filter Hof neighborhoods of eHinH
having compact closure.
Let us also recall the notion of discontinuity group of a homo morphismπof
a topological group Ginto a topological group H, which was introduced in [1]
and used in [2, 3].
Definition 1 LetGandHbe topological groups, and let U=UGbe the filter
of neighborhoods of eGinG. | 2312.00002v1.pdf | 101eb26d-415e-4e10-bdec-8e615f44e970 |
For every (not necessarily continuous) locally
relatively compact homomorphism πofGintoH, the set DG(π) =⋂
U∈Uπ(U)
is called the discontinuity group of π.
1 | 2312.00002v1.pdf | ab58153a-95da-427f-8430-15c567fcc6e3 |
Here and below, the bar stands for the closure in the correspo nding topology
(therefore, here the closure is taken with respect to the top ology ofH). (See [2,
Definition 1.1.1].)
The discontinuity group of a homomorphism has some importan t properties.
Let us list some of them.
Letπbe a (not necessarily continuous) locally relatively compa ct homomor-
phism of a connected Lie group Ginto a topological group H. Then the set
DG(π)is a compact subgroup of the topological group Hand a compact normal
subgroup of the closed subgroup π(G)ofH.
Moreover, the filter basis {π(U)|U∈U}converges to DG(π), and the
homomorphism πis continuous if and only if DG(π) ={eH}. (See [2, Theo-
rem 1.1.2].)
Under the same conditions, DG(π)is a compact connected subgroup of H.
(See [2, Lemma 1.1.6].)(See [2, Lemma 1.1.6].)
LetGbe a connected locally compact group, let Nbe a closed normal
subgroup of G, and letπbe a locally relatively compact homomorphism of G
into a topological group H(for instance, a locally bounded homomorphism
into a locally compact group). Let Mbe the discontinuity group of the re-
striction DG(π|N). ThenMis a closed normal subgroup of the compact dis-
continuity group DG(π), and the corresponding quotient group DG(π)/Mis
isomorphic to the discontinuity group DG(ψ)of the homomorphism ψofG
obtained as the composition of the homomorphism πand the canonical homo-
morphismπ(G)→π(G)/M. (See [2, Lemma 1.1.7].)
In this note we prove that the discontinuity group of every lo cally bounded
homomorphism of a Lie group into a Lie group is not only compac t and con-
nected but also commutative. | 2312.00002v1.pdf | 9235ea1d-e558-405b-88e1-f4e7c3bbbf49 |
2 Preliminaries
The following refinement of Lemma 1.1.9 of [2] obviously hold s.
Theorem 1 LetGbe a topological group, let G′be the commutator subgroup
ofG(in the algebraic sense and in the topology induced by that of G), and
letπbe a locally relatively compact homomorphism of Ginto some topological
groupH. Then the commutator subgroup of the discontinuity group of the ho-
momorphism πis a subgroup of the discontinuity group of the restriction π|G′
ofπto the commutator subgroup G′ofG:
DG(π)′⊂DG(π|G′);
moreover, DG(π)′is a normal subgroup of DG(π|G′).
Proof. The group DG(π)′is a normal subgroup of DG(π|G′)indeed since,
by the invariance of G′inG,DG(π)′is obviously invariant with respect to inner
automorphisms of G, and hence normal in DG(π|G′).The proof of Lemma 1.1.9 in [2] is a model for the proof of the fo llowing
main theorem.
2 | 2312.00002v1.pdf | dfac3783-addf-42c7-b2d7-7d439849363d |
3 Main result
Theorem 2 The discontinuity group of every locally bounded homomorphis m
of a Lie group into a Lie group is commutative.
Proof. LetGandHbe Lie groups, and let πbe a locally bounded ho-
momorphism of GintoH. LetG9be the connected component of G. The
discontinuity group of πcoincides obviously with the discontinuity group of the
restrictionπ|G0. Thus, we may assume that Gis connected. Let DG(π)⊂H
be the discontinuity group of π.
Consider the adjoint representation AdHofH.The composition AdH◦
πis a locally bounded linear representation of G. By Corollary 1.3.3 of [2],
the discontinuity group DG(AdH◦π)is a commutative compact connected
subgroup of H.
By the very definition DG(π) =⋂
U∈Uπ(U)of the discontinuity group of a
homomorphism and by the continuity of AdH, we see that, since the kernel of
AdHcoincides with the center ZHof the group H, it follows that DG(AdH◦
π) =⋂
U∈U(AdH◦π)(U)coincides with AdH(DG(π)). Therefore, DG(AdH◦
π)is the image of DG(π)under the factorization of Hby the center ZH.
However, both ZHandDG(AdH◦π)are commutative groups. There-
fore,DG(π)is solvable.There-
fore,DG(π)is solvable. Let Z0
Hbe the connected component of the identity
elementeofZHand letDbe the complementary discrete subgroup of ZH
toZ0
Hsuch thatD∩Z0
H={e}. Then the connected component of DG(π)is
a connected solvable Lie subgroup of the compact Lie group DG(π). Hence,
this component is commutative. The whole group DG(π)is generated by this
commutative connected component of the identity element an d the central sub-
groupD. Therefore, DG(π)is commutative, as was to be proved.
4 Discussion
As is known, Theorem 2 cannot be extended even to connected co mpact groups,
since they can have discontinuous non-one-dimensional fini te-dimensional rep-
resentations [1]–[3].
Funding
The research was supported by SRISA RAS according to the proj ect FNEF-
2022-0007 (Reg. no. 1021060909180-7-1.2.1). | 2312.00002v1.pdf | 30bf311c-009f-43d2-a80d-0bc563adaa07 |
no. 1021060909180-7-1.2.1).
References
[1] A. I. Shtern, A version of van der Waerden’s theorem and a p roof of
Mishchenko’s conjecture on homomorphisms of locally compa ct groups, Izv.
Math., 2008, 72 (1), 169–205.
3 | 2312.00002v1.pdf | 9db59e6f-3d4c-4f6c-b97c-035c393b94f1 |
[2] A. I. Shtern, Finite-Dimensional Quasirepresentation s of Connected Lie
Groups and Mishchenko’s Conjecture, J. Math. Sci. (N. Y.), 2 009, 159 (5),
653–751.
[3] A. I. Shtern, Locally Bounded Finally Precontinuous Fin ite-Dimensional
Quasirepresentations of Locally Compact Groups, Sb. Math. , 2017, 208
(10), 1557–1576.
4 | 2312.00002v1.pdf | 1fd8c5c2-4167-463f-8ccb-0821cf3af269 |
arXiv:2311.00001v1 [quant-ph] 1 Aug 2023A Fisher Information Perspective of Relativistic
Quantum Mechanics
Asher Yahalom
Ariel University, Ariel 40700, Israel
(E-mail: [email protected] )
Abstract. In previous papers we have shown how Schr¨ odinger’s equatio n which in-
cludes an electromagnetic field interaction can be deduced f rom a fluid dynamical
Lagrangian of a charged potential flow that interacts with an electromagnetic field.
The quantum behaviour was derived from Fisher information t erms which were added
to the classical Lagrangian. It was thus shown that a quantum mechanical system is
drived by information and not only electromagnetic fields.
This program was applied also to Pauli’s equations by removi ng the restriction
of potential flow and using the Clebsch formalism. Although t he analysis was quite
successful there were still terms that did not admit interpr etation, some of them
can be easily traced to the relativistic Dirac theory.Here w e repeat the analysis
for a relativistic flow, pointing to a new approach for derivi ng relativistic quantum
mechanics.
Keywords: Spin, Fluid dynamics, Electromagnetic interaction.
1 Introduction
Quantum mechanics, is usually interpreted by the Copenhagen scho ol ap-
proach. The Copenhagen approach defies the ontology of the qua ntum wave
function and declares it to be completely epistemological (a tool for estimat-
ing probability of certain measurements) in accordance with the Kan tian [1]
conception of reality, and its denial of the human ability to grasp any thing
”as it is” (ontology). However, historically we also see the developme nt of
another school of prominent scholars that interpret quantum me chanics quite
differently. This school believed in the reality of the wave function. I n their
view the wave function is part of reality much like an electromagnetic fi eld is. | 2311.00001v1.pdf | 4545bcca-022f-4e2c-a416-ee857a848683 |
This approach that was supported by Einstein and Bohm [2,3,4] has re sulted in
other interpretations of quantum mechanics among them the fluid r ealization
championed by Madelung [5,6] which stated that the modulus square o f the
wave function is a fluid density and the phase is a potential of the velo city field
of the fluid. However, this approach was constrained to wave func tions of spin
less electrons and could not take into account a complete set of att ributes even
for slow moving (with respect to the speed of light) electrons.
16thCHAOS Conference Proceedings, 13 - 16 June 2023, Heraklion, Crete,
Greece
C. H. Skiadas (Ed)
©2023 ISAST | 2311.00001v1.pdf | 2f20dc38-2808-473d-9021-3377d984d3b8 |
A non relativistic quantum equation for a spinor was first introduced by
Wolfgang Pauli in 1927 [7]. This equation is based on a two dimensional op -
erator matrix Hamiltonian. Two dimensional operator matrix Hamilton ians
are currently abundant in the literature ([8] - [21]) and describe man y types of
quantum systems. It is natural to inquire wether such a theory ca n be given
a fluid dynamical interpretation. This question is of great importanc e as sup-
porters of the non-realistic Copenhagen school of quantum mech anics usually
use the spin concept as a proof that nature is inherently quantum a nd thus
have elements without classical analogue or interpretation. A Bohm ian anal-
ysis of the Pauli equation was given by Holland and others [3], however , the
analogy of the Pauli theory to fluid dynamics and the notion of spin vo rticity
were not considered. This state of affairs was corrected in [22] int roducing spin
fluid dynamics.The interpretation of Pauli’s spinor in terms of fluid density and velocit y
variables leads us directly to the nineteenth century seminal work o f Clebsch
[23,24] which is stronglyrelated tothe variationalanalysisoffluids. V ariational
principles for barotropic fluid dynamics are described in the literatur e. A four
function variational principle for an Eulerian barotropic fluid was dep icted by
Clebsch [23,24] and much later by Davidov [25] who’s main purpose was t o
quantize fluid dynamics. The work was written in Russian, and was larg ely
unknown in the west. Lagrangian fluid dynamics (which takes a differe nt ap-
proach than Eulerian fluid dynamics) was given a variational descript ion by
Eckart [26]. | 2311.00001v1.pdf | 674c5805-249e-4906-bfb2-c2b614504c61 |
Ignoring both the work of Clebsch (written in German) a nd the
work of Davidov (written in Russian) initial attempts in the English writ ten
literature to formulate Eulerian fluid dynamics using a variational prin ciple,
were given by Herivel [28], Serrin [29] and Lin [30]. However, the variat ional
principles developed by the above authors were cumbersome relying on quite a
few ”Lagrange multipliers” and auxiliary ”potentials”. The total num ber of in-
dependent functions in the above formulations are from eleven to s even, which
are much more than the four functions required for the Eulerian an d continuity
equations of a barotropic flow. Thus those methods did not have pr actical use.
Seliger & Whitham [31] have reintroduced the variational formalism of Cleb-
sch depending on only four variables for barotropic flow.Lynden-B ell & Katz
[32] have described a variational principle in terms of two functions t he load
λand density ρ. However, their formalism contains an implicit definition for
the velocity ⃗ vsuch that one is required to solve a partial differential equation
in order to obtain both ⃗ vin terms of ρandλas well as its variations. Much
the same criticism holds for their general variational for non-baro tropic flows
[33]. Yahalom & Lynden-Bell [34] overcame the implicity definition limitatio n
by paying the price of adding an additional single variational variable. This
formalism allows arbitrary variations (not constrained) and the defi nition of
⃗ vis explicit.The original work of Clebsch and all the following publications
assume a non-relativistic fluids in which the velocity of the flow is much s lower
than the speed of light in vacuum c. This is of course to be expected as the
work of Clebsch preceded Einstein’s work on special relativity by for ty eight
years. This can also be based on practical basis as relativistic flows a re hardly
encountered on earth. | 2311.00001v1.pdf | cd0a516e-b94e-43b0-b6ca-c718259490e7 |
The standard approach to relativistic flows is based on the energy- moment-
um tensor [37,35,36], however, this approach is not rigorous becaus e the defini-
tion of an energy-momentum tensor can only be done if a Lagrangian density is
provided [38]. However, no Lagrangiandensity was known for relativ istic flows.
In this work we intend to expand Clebsch work to relativistic flow and t hus
amend this lacuna with aderived Lagrangiandensity for a relativisticfl ow from
which one can obtain rigorously the energy-momentum tensor of hig h velocity
flows.
A fundamental issue in the fluid interpretation of quantum mechanic s still
remains. This refers to the meaning of thermodynamic quantities. T hermody-
namics concepts like specific enthalpy, pressure and temperature are related to
the specific internal energy defined by the equation of state as a u nique func-
tion of entropy and density.The internal energy is a part of any La grangian
density related to fluid dynamics. The internal energy functional c an in prin-
ciple be explained on the basis of the microscopic composition of the flu id
using statistical physics. That is the atoms and molecules from which the fluid
is composed and their interactions impose an equation of state. How ever, a
quantum fluid has no structure and yet the equations of both the s pin less [5,6]
and spin [22] quantum fluid dynamics shows that terms analogue to int ernal
energies appear. One thus is forces to inquire where do those inter nal energies
originate? Of course one cannot suggest that the quantum fluid ha s a mi-
croscopic sub structure as this will defy current empirical evidenc e suggesting
that the electron is a point particle. The answerto this question com es from an
entirely different scientific discipline known as measurement theory [4 3,44,46]. | 2311.00001v1.pdf | 1593af8d-9b98-4ec2-9d38-aec803412893 |
Fisher information is a basic notion of measurement theory, and is a m easure
of measurement quality of any quantity. It was demonstrated [46 ] that this
notion is the internal energy of a spin less electron (up to a proport ionality
constant) and can interpret sum terms of the internal energy of an electron
with spin. Here we should mention an attempt to derive most physical theories
from Fisher information as described by Frieden [47]. It was suggest ed [48]
that there exist a velocity field such that the Fisher information will g iven a
complete explanation for the spin fluid internal energy. It was also s uggested
that one may define comoving scalar fields as in ideal fluid mechanics, h owever,
this was only demonstrated implicitly but not explicitly. A common featu re of
previous work on the fluid & Fisher information interpretation of qua ntum me-
chanics, is the negligence of electromagnetic interaction thus sett ing the vector
potential to zero. This makes sense as the classical ideal fluids disc ussed in
the literature are not charged.Hence, in order to make the compa rison easier
to comprehend the vector potential should be neglected. Howeve r, one cannot
claim a complete description of quantum mechanics lacking a vector po tential
thus ignoring important quantum phenomena such as the Zeeman eff ect which
depends on a vector potential through the magnetic field, this was taken care
of in [49,50]. However, this previous work assumed a non-relativistic fl ow. In
the current paper we study a relativistic flow and thus suggest a ne w route
leading to relativistic quantum mechanics which is based on a relativistic fluid
dynamics with a Lorentz invariant Fisher information term. | 2311.00001v1.pdf | 61614dcc-68ad-4823-b1ec-14ddec91ca6b |
Wewillbeginthispaperbyintroducingavariationalprincipleforarelat ivis-
tic charged classical particle with a vector potential interaction an d a system
of the same. This will be followed by the Eckart [26] Lagrangian variat ional
principles generalized for a relativistic charged fluid. We then introdu ce an
Eulerian-Clebsch variational principle for a relativistic charged fluid. Finally
the concept of Fisher information will allow us to suggest a new appro ach to
relativistic quantum fluids.
2 Trajectories Through Variational Analysis
We consider a particle travelling in spacetime of a constant metric.Th e action
Aof such a particle is:
A=−mc∫
dτ−e∫
Aαdxα (1)
In the above τis the trajectory interval:
dτ2=⏐⏐ηαβdxαdxβ⏐⏐=|dxαdxα| (2)
xαaretheparticlecoordinates(themetricraisesandlowersindicesac cordingto
the prevailingcustom), misthe particlemass, eisthe chargeand Aαisthe four
vector potential which depend on the particle coordinates. Aαtransforms as a
four dimensional vector. Variational analysis results in the following equations
of motion:
mduα
dτ=−e
cuβ(∂βAα−∂αAβ), uα≡dxα
dτ, ∂α≡∂
∂xα, ∂β≡ηβα∂α
(3)
in which the metric ηαβis the Lorentz metric:
ηαβ= diag (1 ,−1,−1,−1). | 2311.00001v1.pdf | 21cca3a7-de38-4346-977d-5939c2e747ae |
(4)
2.1 Partition to Space & Time
Given a space-time with a Lorentz metric the partition into spatial an d tempo-
ral coordinates is trivial. The spatial coordinates are ⃗ x= (x1,x2,x3) and the
temporal coordinate is x0. As we measure time in the units of seconds which
differ from the space units of meters, we introduce x0=ct, in which cconnects
the different units. The velocity is defined as:
⃗ v≡d⃗ x
dt, v=|⃗ v|, v α≡dxα
dt= (⃗ v,c). (5)
In a similar way we dissect Aαinto temporal and spatial pieces:
Aα= (A0,A1,A2,A3)≡(A0,⃗A)≡(φ
c,⃗A) (6) | 2311.00001v1.pdf | 0351c40d-4ef5-4e68-b70a-892e60bf4365 |
the factor1
cin the last term allows us to obtain the equations in MKS units,
it is not needed in other types of unit systems. Through equation (6 ), we can
define a magnetic field:
⃗B=⃗∇×⃗A (7)
(⃗∇has the standard meaning) and the electric field:
⃗E=−∂⃗A
∂t−⃗∇φ (8)
For the subluminal case v < cwe may write dτ2as:
dτ2=c2dt2(1−v2
c2), dτ =cdt√
1−v2
c2=cdt
γ,γ≡1√
1−v2
c2(9)
And using the above equations the spatial piece of equation (3) is de duced:
d
dt(mγ⃗ v) =d
dt
m⃗ v√
1−v2
c2
=e(
⃗E+⃗ v×⃗B)
(10)
2.2 The Lagrangian
We may write the action (1) as a temporal integral and thus define a La-
grangian:
A=∫t2
t1Ldt, L =L0+Li
L0≡ −mcdτ
dt=−mc2
γ=−mc2√
1−v2
c2≃1
2mv2−mc2,
Li≡ −eAαdxα
dt=e(⃗A·⃗ v−φ). (11)
in the above the ≃symbol signifies a classical (low speed) approximation. | 2311.00001v1.pdf | 193baf8b-0ee5-454b-9ddb-216a7a46502e |
We
notice that the interaction part of the Lagrangian is the same for h igh and low
speeds while the kinetic part takes a different and simpler form for th e low
speed cases.
2.3 The Action & Lagrangian for a System of Particles
Considerasystemof Nparticleseachwithanindex n∈[1−N],acorresponding
massmn, charge en. Each particle will have a trajectory xα
n(τn) in which τn
measures the interval already propagated along the trajectory . Thus:
uα
n≡dxα
n
dτn. (12)
We will assume as usual that the particle trajectories pierce throu gh time
”planes”, and the ”plane” tis pierced at position vector ⃗ xn(t), see figure 1 | 2311.00001v1.pdf | 43968024-c954-4f81-a6b0-6ddf19d39a46 |
Fig.1.Schematic drawing of two trajectories piercing a time ”plan e” which is illus-
trated as a straight line.
(actually each ”plane” is three dimensional). Thus one can define a ve locity
⃗ vn≡d⃗ xn
dt. The action and Lagrangian for each point particle are as before:
An=−mnc∫
dτn−en∫
Aα(xν
n)dxαn=∫t2
t1Lndt, L n≡L0n+Lin
L0n≡ −mnc2
γn≃1
2mnv2
n−mnc2, Lin≡en(
⃗A(⃗ xn,t)·⃗ vn−φ(⃗ xn,t))
.(13)
The action and Lagrangian of the system of particles is:
As=∫t2
t1Lsdt, L s=N∑
n=1Ln.(14)
The variational analysis follows the same lines as for a single particle an d we
obtain a set of equations of the four dimensional form:
mnduα
n
dτn=−en
cuβ
n(∂βAα
n−∂αAβn), n∈[1−N]. (15)
Or the three dimensional form:
d
dt(γn⃗ vn) =en
mn[
⃗ vn×⃗B(⃗ xn,t)+⃗E(⃗ xn,t)]
, n∈[1−N].(16)
in which we do not sum over repeated Latin indices
3 A Relativistic Charged Fluid - the Lagrangian
Approach
3.1 The Action and Lagrangian
The dynamics of the fluid is determined by its composition and the forc es
acting on it. | 2311.00001v1.pdf | a87bbd48-fbc3-42c6-ac22-e75e45089060 |
The fluid is made of ”fluid elements” [26,27], practically a ”flu id
element” is a point particle which has an infinitesimal mass dM⃗ α, infinitesimal
chargedQ⃗ α, position four vector x⃗ αν(τ⃗ α) andu⃗ αν(τ⃗ α)≡dx⃗ αν(τ⃗ α)
dτ⃗ α. Here the
continuous vector label ⃗ αreplaces the discrete index nof the previous section. | 2311.00001v1.pdf | 6c9dac78-035d-4f6a-9597-9e6961ef0386 |
As the ”fluid element” is not truly a point particle it has also an infinitesim al
volumedV⃗ α, infinitesimal entropy dS⃗ α, and an infinitesimal internal energy
dEin ⃗ α. The action for each ”fluid element” are according to equation (11) as
follows:
dA⃗ α=−dM⃗ αc∫
dτ⃗ α−dQ⃗ α∫
Aµ(xν
⃗ α)dxµ⃗ α+dAin ⃗ α,
dAin ⃗ α≡ −∫
dEin ⃗ αdt.(17)
The Lagrangian for each ”fluid element” can be derived from the abo ve expres-
sion as follows:
dA⃗ α=∫t2
t1dL⃗ αdt, dL ⃗ α≡dLk⃗ α+dLi⃗ α−dEin ⃗ α
dLk⃗ α≡ −dM⃗ αc2
γ⃗ α≃1
2dM⃗ αv⃗ α(t)2−dM⃗ αc2
dLi⃗ α≡dQ⃗ α(
⃗A(⃗ x⃗ α(t),t)·⃗ v⃗ α(t)−φ(⃗ x⃗ α(t),t))
.(18)
all the above quantities are calculated for a specific value of the labe l⃗ α, while
the action and Lagrangianof the entire fluid, should be summed (orin tegrated)
over all possible ⃗ α’s. That is:
L=∫
⃗ αdL⃗ α
A=∫
⃗ αdA⃗ α=∫
⃗ α∫t2
t1dL⃗ αdt=∫t2
t1∫
⃗ αdL⃗ αdt=∫t2
t1Ldt. (19)
It is customary to define densities for the Lagrangian, mass and ch arge of every
fluid element as follows:
L⃗ α≡dL⃗ α
dV⃗ α, ρ⃗ α≡dM⃗ α
dV⃗ α, ρc⃗ α≡dQ⃗ α
dV⃗ α,ein ⃗ α≡dEin ⃗ α
dV⃗ α(20)
Each of the above quantities may be thought of as a function of the location
⃗ x, where the ”fluid element” labelled ⃗ αhappens to be in time t, for example:
ρ(⃗ x,t)≡ρ(⃗ x⃗ α(t),t)≡ρ⃗ α(t) (21)
It is also customary to define the specific internal energy ε⃗ αas follows:
ε⃗ α≡dEin ⃗ α
dM⃗ α⇒ρ⃗ αε⃗ α=dM⃗ α
dV⃗ αdEin ⃗ α
dM⃗ α=dEin ⃗ α
dV⃗ α=ein ⃗ α (22)
Thus we can write the following equations for the Lagrangian density :
L⃗ α=dL⃗ α
dV⃗ α=dLk⃗ | 2311.00001v1.pdf | 0af804b3-173c-456c-ae7e-b1c1e97e9633 |
α=dL⃗ α
dV⃗ α=dLk⃗ α
dV⃗ α+dLi⃗ α
dV⃗ α−dEin ⃗ α
dV⃗ α=Lk⃗ α+Li⃗ α−ein ⃗ α
Lk⃗ α≡ −ρ⃗ αc2
γ⃗ α≃1
2ρ⃗ αv⃗ α(t)2−ρ⃗ αc2,
Li⃗ α≡ρc⃗ α(
⃗A(⃗ x⃗ α(t),t)·⃗ v⃗ α(t)−ϕ(⃗ x⃗ α(t),t))
. (23) | 2311.00001v1.pdf | 4d665450-fb53-4875-990c-b7325da1e977 |
The above expression allows us to write the Lagrangian as a spatial in tegral:
L=∫
⃗ αdL⃗ α=∫
⃗ αL⃗ αdV⃗ α=∫
L(⃗ x,t)d3x (24)
which will be important for later sections of the current paper.
3.2 Variational Analysis
Returning now to the variational analysis we introduce the symbols ∆⃗ x⃗ α≡⃗ξ⃗ α
to indicate a variation of the trajectory ⃗ x⃗ α(t) (we reserve the symbol δin
the fluid context, to a different kind of variation, the Eulerian variat ion to be
described in the next section). Notice that:
∆⃗ v⃗ α(t) =∆d⃗ x⃗ α(t)
dt=d∆⃗ x⃗ α(t)
dt=d⃗ξ⃗ α(t)
dt.(25)
And thus according to equation (9):
∆(1
γ⃗ α)
=−γ⃗ α⃗ v⃗ α(t)
c2d⃗ξ⃗ α(t)
dt, ∆γ ⃗ α=γ3
⃗ α⃗ v⃗ α(t)
c2d⃗ξ⃗ α(t)
dt.(26)
In an ideal fluid the ”fluid element” does exchange mass, nor electric charge,
nor heat with other fluid elements, so it follows that:
∆dM⃗ α=∆dQ⃗ α=∆dS⃗ α= 0. | 2311.00001v1.pdf | 6b1d91b1-b0f2-4fb2-8cde-056b80ff7c9a |
(27)
Moreover, according to thermodynamics a change in the internal e nergy of a
”fluid element” satisfies the equation in the particle’s rest frame:
∆dEin ⃗ α0=T⃗ α0∆dS⃗ α0−P⃗ α0∆dV⃗ α0, (28)
the first term describes the heating energy gained by the ”fluid elem ent” while
the second terms describes the work done by the ”fluid element” on neighbour-
ing elements. T⃗ α0is the temperature of the ”fluid element” and P⃗ α0is the
pressure of the same.As the rest mass of the fluid element does no t change
and does not depend on any specific frame we may divide the above ex pression
bydM⃗ αto obtain the variation of the specific energy as follows:
∆ε⃗ α0=∆dEin ⃗ α0
dM⃗ α=T⃗ α0∆dS⃗ α0
dM⃗ α−P⃗ α0∆dV⃗ α0
dM⃗ α
=T⃗ α0∆s⃗ α0−P⃗ α0∆1
ρ⃗ α0=T⃗ α0∆s⃗ α0+P⃗ α0
ρ2
⃗ α0∆ρ⃗ α0. s⃗ α0≡dS⃗ α0
dM⃗ α(29)
in which s⃗ α0is the specific entropy of the fluid element in its rest frame.It
follows that:∂ε0
∂s0=T0,∂ε0
∂ρ0=P0
ρ2
0. (30)
Another important thermodynamic quantity that we will use later is t he En-
thalpy defined for a fluid element in its rest frame as:
dW⃗ α0=dEin ⃗ α0+P⃗ α0dV⃗ α0. (31) | 2311.00001v1.pdf | b2595bdb-5977-4ba8-b8f5-563fc0095141 |
and the specific enthalpy:
w⃗ α0=dW⃗ α0
dM⃗ α=dEin ⃗ α0
dM⃗ α+P⃗ α0dV⃗ α0
dM⃗ α=ε⃗ α0+P⃗ α0
ρ⃗ α0. (32)
Combining the above result with equation (30) it follows that:
w0=ε0+P0
ρ0=ε0+ρ0∂ε0
∂ρ0=∂(ρ0ε0)
∂ρ0. | 2311.00001v1.pdf | 307de169-3e9e-4c5b-bb51-1cf19fbfa033 |
(33)
Moreover:
∂w0
∂ρ0=∂(ε0+P0
ρ0)
∂ρ0=−P0
ρ2
0+1
ρ0∂P0
∂ρ0+∂ε0
∂ρ0=−P0
ρ2
0+1
ρ0∂P0
∂ρ0+P0
ρ2
0=1
ρ0∂P0
∂ρ0.
(34)
As we assume an ideal fluid, there is no heat conduction or heat radia tion,
and thus heat can only be moved around along the trajectory of th e ”fluid
elements”, that is only convection is taken into account. Thus ∆dS⃗ α0= 0 and
we have:
∆dEin ⃗ α0=−P0∆dV⃗ α0. (35)
Our next step would to be to evaluate the variation of the volume elem ent.(35)
Our next step would to be to evaluate the variation of the volume elem ent.
However, before we do this we establish some relations between the rest frame
and any other frame in which the fluid element is in motion (this frame is
sometimes denoted the ”laboratory” frame). First we notice that at the rest
frame there is no velocity (by definition), hence according to equat ion (9):
dτ=cdt0=cdt√
1−v2
c2=cdt
γ⇒dt0=dt
γ. (36)
It is well known that the four volume is Lorentz invariant, hence:
dV0dt0=dVdt=dVdt0γ,⇒dV0=γdV. (37)
Thus:
ρ0=dM
dV0=1
γdM
dV=ρ
γ,⇒ρ=γρ0.(38)
Moreover, the action given in equation (17) is Lorentz invariant, th us:
dEin ⃗ α0dt0=dEin ⃗ αdt=dEin ⃗ αdt0γ⇒dEin ⃗ α0=γdEin ⃗ α,dEin ⃗ α=dEin ⃗ α0
γ
(39)
We are now at a position to calculate the variation of the internal ene rgy of a
fluid element:
∆dEin ⃗ α=∆(1
γ)
dEin ⃗ α0+1
γ∆dEin ⃗ α0. (40)
Taking into account equation (35) and equation (37) we obtain:
∆dEin ⃗ α=∆(1
γ)
dEin ⃗ α0−1
γP0∆dV⃗ α0=∆(1
γ)
dEin ⃗ α0−1
γP0∆(γdV⃗ α).
(41) | 2311.00001v1.pdf | 3df8d19a-7dee-4913-9104-b2cc39cb3496 |
Thus using the definition of enthalpy given in equation (31) we may writ e:
∆dEin ⃗ α=∆(1
γ)
(dEin ⃗ α0+P0dV⃗ α0)−P0∆dV⃗ α=∆(1
γ)
dW⃗ α0−P0∆dV⃗ α.
(42)
We shall now calculate the variation of the volume element.(42)
We shall now calculate the variation of the volume element. Suppose a t a time
tthe volume of the fluid element labelled by ⃗ αis described as:
dV⃗ α,t=d3x(⃗ α,t) (43)
Using the Jacobian determinant we may relate this to the same elemen t at
t= 0:
d3x(⃗ α,t) =Jd3x(⃗ α,0), J≡⃗∇0x1·(⃗∇0x2×⃗∇0x3) (44)
In which ⃗∇0is taken with respect to the coordinates of the fluid elements at
t= 0:⃗∇0≡(∂
∂x(⃗ α,0)1,∂
∂x(⃗ α,0)2,∂
∂x(⃗ α,0)3).As both the actual and varied ”fluid
element” trajectories start at the same point it follows that:
∆dV⃗ α,t=∆d3x(⃗ α,t) =∆J d3x(⃗ α,0) =∆J
Jd3x(⃗ α,t) =∆J
JdV⃗ α,t,
(∆d3x(⃗ α,0) = 0).(45)
The variation of Jcan be easily calculated as:
∆J=⃗∇0∆x1·(⃗∇0x2×⃗∇0x3)+⃗∇0x1·(⃗∇0∆x2×⃗∇0x3)+⃗∇0x1·(⃗∇0x2×⃗∇0∆x3),
(46)
Now:
⃗∇0∆x1·(⃗∇0x2×⃗∇0x3) =⃗∇0ξ1·(⃗∇0x2×⃗∇0x3)
=∂kξ1⃗∇0xk·(⃗∇0x2×⃗∇0x3) =∂1ξ1⃗∇0x1·(⃗∇0x2×⃗∇0x3) =∂1ξ1J.⃗∇0x1·(⃗∇0∆x2×⃗∇0x3) =⃗∇0x1·(⃗∇0ξ2×⃗∇0x3)
=∂kξ2⃗∇0x1·(⃗∇0xk×⃗∇0x3) =∂2ξ2⃗∇0x1·(⃗∇0x2×⃗∇0x3) =∂2ξ2J. | 2311.00001v1.pdf | b32fdf2a-3509-44f2-a3f9-d642014b121c |
⃗∇0x1·(⃗∇0x2×⃗∇0∆x3) =⃗∇0x1·(⃗∇0x2×⃗∇0ξ3)
=∂kξ3⃗∇0x1·(⃗∇0x2×⃗∇0xk) =∂3ξ3⃗∇0x1·(⃗∇0x2×⃗∇0x3) =∂3ξ3J.(47)
Combining the above results, it follows that:
∆J=∂1ξ1J+∂2ξ2J+∂3ξ3J=⃗∇·⃗ξ J. (48)
Which allows us to calculate the variation of the volume of the ”fluid elem ent”:
∆dV⃗ α,t=⃗∇·⃗ξ dV⃗ α,t.(49)
And thus the variation of the internal energy given in equation (42) is:
∆dEin ⃗ α=∆(1
γ)
dW⃗ α0−P0⃗∇·⃗ξ dV⃗ α,t. (50) | 2311.00001v1.pdf | 91cff25d-910d-459a-9399-f2dc140e9138 |
Taking into account equation (26) this takes the form:
∆dEin ⃗ α=−P⃗ α0⃗∇·⃗ξ⃗ αdV⃗ α,t−γ⃗ α⃗ v⃗ α(t)
c2dW⃗ α0·d⃗ξ⃗ α(t)
dt.(51)
The variation of internal energy is the only novel element with respe ct to
the system of particles scenario described in the previous section, thus the rest
of the variation analysis is straight forward.Varying equation (17) we obtain:
∆dA⃗ α=∫t2
t1∆dL⃗ αdt, ∆dL ⃗ α=∆dLk⃗ α+∆dLi⃗ α−∆dEin ⃗ α
∆dLk⃗ α=−dM⃗ αc2∆(1
γ⃗ α)
=dM⃗ αγ⃗ α⃗ v⃗ α(t)·d⃗ξ⃗ α(t)
dt,
∆dLi⃗ α=dQ⃗ α(
∆⃗A(⃗ x⃗ α(t),t)·⃗ v⃗ α(t)+⃗A(⃗ x⃗ α(t),t)·∆⃗ v⃗ α(t)
−∆φ(⃗ x⃗ α(t),t)). | 2311.00001v1.pdf | 00682685-1bf7-48f7-b219-99e8db0763ac |
(52)
We can now combine the internal and kinetic parts of the varied Lagr angian
taking into account the specific enthalpy definition given in equation ( 32):
∆dLk⃗ α−∆dEin ⃗ α=dM⃗ αγ⃗ α((
1+w0
c2)
⃗ v⃗ α(t)·d⃗ξ⃗ α(t)
dt+P⃗ α0⃗∇·⃗ξ⃗ αdV⃗ α,t.(53)
The electromagnetic interaction variation terms are not different t han in the
low speed (non-relativistic) case, see for example equations A47 an d A48 of
[49], and their derivation will not be repeated here:
d⃗FL⃗ α≡dQ⃗ α[
⃗ v⃗ α×⃗B(⃗ x⃗ α(t),t)+⃗E(⃗ x⃗ α(t),t)]
(54)
and:
∆dLi⃗ α=d(dQ⃗ α⃗A(⃗ x⃗ α(t),t)·⃗ξ⃗ α)
dt+d⃗FL⃗ α·⃗ξ⃗ α. (55)
Introducing the shorthand notation:
¯λ≡1+w0
c2, λ≡γ¯λ=γ(
1+w0
c2)
.(56)
The variation of the action of a relativistic single fluid element is thus:
∆dA⃗ α=∫t2
t1∆dL⃗ αdt= (dM⃗ αλ⃗ α⃗ v⃗ α(t)+dQ⃗ α⃗A(⃗ x⃗ α(t),t))·⃗ξ⃗ α⏐⏐⏐t2
t1
−∫t2
t1(dM⃗ αd(λ⃗ α⃗ v⃗ α(t))
dt·⃗ξ⃗ α−d⃗FL⃗ α·⃗ξ⃗ α−P⃗ α0⃗∇·⃗ξ⃗ αdV⃗ α,t)dt.(57)
The variation of the total action of the fluid is thus:
∆A=∫
⃗ αdA⃗ α=∫
⃗ α(dM⃗ αλ⃗ α⃗ v⃗ α(t)+dQ⃗ α⃗A(⃗ x(⃗ α,t),t))·⃗ξ⃗ α⏐⏐⏐⏐t2
t1
−∫t2
t1∫
⃗ α(dM⃗ αd(λ⃗ α⃗ v⃗ α(t))
dt·⃗ξ⃗ α−d⃗FL⃗ α·⃗ξ⃗ α−P⃗ α0⃗∇·⃗ξ⃗ αdV⃗ α)dt.(58) | 2311.00001v1.pdf | 2ea307a3-1fa7-4e41-b89b-c42501bd0675 |
Now according to equation (20) we may write:
dM⃗ α=ρ⃗ αdV⃗ α, dQ ⃗ α=ρc⃗ αdV⃗ α (59)
using the above relations we may turn the ⃗ αintegral into a volume integral and
thus write the variation of the fluid action in which we suppress the ⃗ αlabels:
∆A=∫
(ρλ⃗ v+ρc⃗A)·⃗ξdV⏐⏐⏐⏐t2
t1−∫t2
t1∫
(ρd(λ⃗ v)
dt·⃗ξ−⃗fL·⃗ξ−P0⃗∇·⃗ξ)dVdt.(60)
in the above we introduced the Lorentz force density:
⃗fL⃗ α≡d⃗FL⃗ α
dV⃗ α=ρc⃗ α[
⃗ v⃗ α×⃗B(⃗ x⃗ α(t),t)+⃗E(⃗ x⃗ α(t),t)]
. | 2311.00001v1.pdf | 425e25da-4b6f-47c9-806d-8ae3c31b9c41 |
(61)
Now, since:
P0⃗∇·⃗ξ=⃗∇·(P0⃗ξ)−⃗ξ·⃗∇P0, (62)
and using Gauss theorem the variation of the action can be written a s:
∆A=∫
(ρλ⃗ v+ρc⃗A)·⃗ξdV⏐⏐⏐⏐t2
t1
−∫t2
t1[∫
(ρd(λ⃗ v)
dt−⃗fL+⃗∇P0)·⃗ξdV−∮
P0⃗ξ·d⃗Σ]
dt.(63)
It follows that the variation of the action will vanish for a ⃗ξsuch that ⃗ξ(t1) =
⃗ξ(t2) = 0andvanishingonasurfaceencapsulatingthefluid, butothert hanthat
arbitrary only if the Euler equation for a relativistic charged fluid is sa tisfied,
that is:
d(λ⃗ v)
dt=−⃗∇P0
ρ+⃗fL
ρ(64)
for the particular case that the fluid element is made of identical micr oscopic
particleseachwith amass mand a charge e, it followsthat the massand charge
densities are proportional to the number density n:
ρ=m n, ρ c=e n⇒⃗fL
ρ=k[
⃗ v×⃗B+⃗E]
,k≡e
m(65)
thus except from the terms relatedto the internalenergythe eq uationis similar
to that of a point particle.For a neutral fluid one obtains the form:
d(λ⃗ v)
dt=−⃗∇P0
ρ. (66)
Some authors prefer to write the above equation in terms of the en ergy per
element of the fluid per unit volume in the rest frame which is the sum of the
internal energy contribution and the rest mass contribution:
e0≡ρ0c2+ρ0ε0. (67) | 2311.00001v1.pdf | c4a1657f-b8ca-4c97-96c4-1410308f7ede |
It is easy to show that:
¯λ= 1+w0
c2=e0+P0
ρ0c2. (68)
And using the above equality and some manipulations we may write equa tion
(66) in a form which is preferable by some authors:
(e0+P0)γ
c2d(γ⃗ v)
dt=−⃗∇P0−γ2
c2dP0
dt⃗ v. (69)
In experimentalfluid dynamicsit is moreconvenientto describea fluid in terms
ofquantitiesataspecificlocation,ratherthanquantitiesassociat edwithunseen
infinitesimal ”fluid elements”. This road leads to the Eulerian descript ion of
fluid dynamics and thinking in terms of flow fields rather than in terms o f a
velocity of ”fluid elements” as will be discussed in the next section.4 An Eulerian Charged Fluid - the Clebsch Approach
In this section we follows closely the analysis of [22,49,50] with the mod ifica-
tion of taking into account the relativistic corrections, this implies ta king into
account an action which is invariant under Lorentz transformation s. Let us
consider the action:
A ≡∫
Ld3xdt, L ≡ L 0+L2+Li
L0≡ −ρ(c2
γ+ε) =−ρ0(c2+ε0) =−e0,L2≡ν∂ν(ρ0uν)−ρ0αuν∂νβ,
Li≡ −ρcAνvν, v ν≡dxν
dt. | 2311.00001v1.pdf | 6e0b5891-d9cd-4bee-b05c-682c397e21ad |
(70)
In the non relativistic limit we may write:
L0≃ρ(1
2v2−ε−c2) (71)
Taking into account that:
uµ=γ(c,⃗ v) (72)
and also that ρ=γρ0according to equation (38), it is easy to write the above
Lagrangian densities in a space-time formalism:
L2=ν[∂ρ
∂t+⃗∇·(ρ⃗ v)]−ραdβ
dtLi=ρc(
⃗A·⃗ v−φ)
(73)
In the Eulerian approach we consider the variational variables to be fields,
that functions of space and time. We have two such variational var iables the
vector velocity field ⃗ v(⃗ x,t) and density scalar field ρ(⃗ x,t). The conservation of
quantities such as the label of the fluid element, mass, charge and e ntropy are | 2311.00001v1.pdf | 8310885a-1668-44aa-81b1-8d79e44a9d03 |
dealtbyintroducingLagrangemultipliers ν,αinsuchawaythatthevariational
principle will yield the following equations:
∂ρ
∂t+⃗∇·(ρ⃗ v) = 0
dβ
dt= 0 (74)
Provided ρis not null those arejust the continuity equation which ensuresmas s
conservation and the conditions that βis comoving and is thus a label. | 2311.00001v1.pdf | e669d429-5b07-472f-851f-de22ddc7fe20 |
Let us
now calculate the variation with respect to β, this will lead us to the following
results:
δβA=∫
d3xdtδβ[∂(ρα)
∂t+⃗∇·(ρα⃗ v)]
−∮
d⃗S·⃗ vραδβ−∫
d⃗Σ·⃗ vρα[δβ]−∫
d3xραδβ|t1
t0(75)
Hence choosing δβin such a waythat the temporal and spatial boundary terms
vanish (this includes choosing δβto be continuous on the cut if one needs to
introduce such a cut) in the above integral will lead to the equation:
∂(ρα)
∂t+⃗∇·(ρα⃗ v) = 0 (76)
Using the continuity equation (74) this will lead to the equation:
dα
dt= 0 (77)
Hence for ρ̸= 0 both αandβare comovingcoordinates.This is why in the Eu-
lerian approach we are obliged to add the Lagrangian density L2. The specific
internal energy ε0defined in equation (22) is dependent on the thermodynamic
properties of the specific fluid. That is it generally depends through a given
”equation of state” on the density and specific entropy. In our ca se we shall
assume a barotropic fluid, that is a fluid in which ε0(ρ0) is a function of the
densityρ0only. Other functions connected to the electromagnetic interact ion
such as the potentials ⃗A,φare assumed given function of coordinates and are
not varied. Another simplification which we introduce is the assumptio n the
fluid element is made of microscopic particles having a given mass mand a
chargee, in this case it follows from equation (65) that:
ρc=kρ.(78)
Let us now take the variational derivative with respect to the dens ityρ, we
obtain:
δρA=∫
d3xdtδρ[−c2
γ−w0δρ0
δρ−∂ν
∂t−⃗ v·⃗∇ν+k(⃗A·⃗ v−φ)]
+∮
d⃗S·⃗ vδρν+∫
d⃗Σ·⃗ vδρ[ν]+∫
d3xνδρ|t1
t0(79) | 2311.00001v1.pdf | 735f77d8-0d82-4352-943d-69910e666fda |
Or as:
δρA=∫
d3xdtδρ[−c2+w0
γ−∂ν
∂t−⃗ v·⃗∇ν+k(⃗A·⃗ v−φ)]
+∮
d⃗S·⃗ vδρν+∫
d⃗Σ·⃗ vδρ[ν]+∫
d3xνδρ|t1
t0(80)
in which w0=∂(ρ0ε0)
∂ρ0is the specific enthalpy in the rest frame of the fluid
element (see equation (33)).Hence provided that δρvanishes on the boundary
of the domain, on the cut and in initial and final times the following equa tion
must be satisfied:
dν
dt=∂ν
∂t+⃗ v·⃗∇ν=−c2+w0
γ+k(⃗A·⃗ v−φ) (81)
In the above we notice that taking a time derivative for a fixed label ⃗ α(also
known as a material derivative) of any quantity gtakes the form:
dg(⃗ α,t)
dt=dg(⃗ x(⃗ α,t),t)
dt=∂g
∂t+d⃗ x
dt·⃗∇g=∂g
∂t+⃗ v·⃗∇g(82)
oncegis considered to be a field dependent on ⃗ x,t.Finally Let us take an arbitrary variational derivative of the above a ction
with respect to ⃗ v, taking into account that:
δ⃗ v1
γ=−γ⃗ v·δ⃗ v
c2(83)
This will result in:
δ⃗ vA=∫
d3xdtρδ⃗ v·[γ⃗ v−w0
ρδρ0
δ⃗ v−⃗∇ν−α⃗∇β+k⃗A]
+∮
d⃗S·δ⃗ vρν+∫
d⃗Σ·δ⃗ vρ[ν]. | 2311.00001v1.pdf | 532c4388-f704-4c9f-a3a8-35c8723a4187 |
(84)
However:
δρ0
δ⃗ v=ρδ1
γ
δ⃗ v=−ργ⃗ v
c2(85)
Taking in account the definition of λ(see equation (56)), we thus have:
δ⃗ vA=∫
d3xdtρδ⃗ v·[λ⃗ v−⃗∇ν−α⃗∇β+k⃗A]
+∮
d⃗S·δ⃗ vρν+∫
d⃗Σ·δ⃗ vρ[ν]. (86)
the above boundary terms contain integration over the external boundary∮
d⃗S
and an integral over the cut∫
d⃗Σthat must be introduced in case that νis
not single valued, more on this case in later sections.The external b oundary
term vanishes; in the case of astrophysical flows for which ρ= 0 on the free
flow boundary, or the case in which the fluid is contained in a vessel wh ich | 2311.00001v1.pdf | 79a95e74-067f-4675-8670-c8f4c0f24b74 |
induces a no flux boundary condition δ⃗ v·ˆn= 0 (ˆnis a unit vector normal
to the boundary). The cut ”boundary” term vanish when the veloc ity field
varies only parallel to the cut that is it satisfies a Kutta type conditio n. If the
boundary terms vanish ⃗ vmust have the following form:
λ⃗ v=α⃗∇β+⃗∇ν−k⃗A (87)
this is a generalization of Clebsch representation of the flow field (se e for ex-
ample [26], [40, page 248]) for a relativistic charged flow.
4.1 Euler’s equations
We shall now show that a velocity field given by equation (87), such th at the
functions α,β,νsatisfy the corresponding equations (74,81,77) must satisfy
Euler’s equations.Let us calculate the material derivative of λ⃗ v:
d(λ⃗ v)
dt=d⃗∇ν
dt+dα
dt⃗∇β+αd⃗∇β
dt−kd⃗A
dt(88)
It can be easily shown that:
d⃗∇ν
dt=⃗∇dν
dt−⃗∇vn∂ν
∂xn=⃗∇(
−c2+w0
γ+k⃗A·⃗ v−kφ)
−⃗∇vn∂ν
∂xn
d⃗∇β
dt=⃗∇dβ
dt−⃗∇vn∂β
∂xn=−⃗∇vn∂β
∂xn(89)
In which xnis a Cartesian coordinate and a summation convention is assumed. | 2311.00001v1.pdf | b2cc27ac-36ac-4fe1-a890-e881a847abc2 |
Inserting the result from equations (89) into equation (88) yields:
d(λ⃗ v)
dt=−⃗∇vn(∂ν
∂xn+α∂β
∂xn)+⃗∇(
−c2+w0
γ+k⃗A·⃗ v−kφ)
−kd⃗A
dt
=−⃗∇vn(λvn+kAn)+⃗∇(−c2+w0
γ+k⃗A·⃗ v−kφ)−k∂t⃗A−k(⃗ v·⃗∇)⃗A
=−1
γ⃗∇w0+k⃗E+k(vn⃗∇An−vn∂n⃗A), (90)
in the above we have used the electric field defined in equation (8).We notice
that according to equation (7):
(vn⃗∇An−vn∂n⃗A)l=vn(∂lAn−∂nAl) =ǫlnjvnBj= (⃗ v×⃗B)l,(91)
Hence we obtain the Euler equation of a charged relativistic fluid in the form:
d(λ⃗ v)
dt=−1
γ⃗∇w0+k[
⃗ v×⃗B+⃗E]
=−1
ρ⃗∇P0+k[
⃗ v×⃗B+⃗E]
,(92)
since (see equation (34)):
⃗∇w0=∂w0
∂ρ0⃗∇ρ0=1
ρ0∂P0
∂ρ0⃗∇ρ0=1
ρ0⃗∇P0.(93)
The above equation is identical to equation (64) and thus proves th at the Euler
equations can be derived from the action given in equation (73) and h ence all
the equations of charged fluid dynamics can be derived from the abo ve action
without restricting the variations in any way. | 2311.00001v1.pdf | d57aedaf-7f15-4197-886e-a3d9dcadee44 |
4.2 Simplified action
The reader of this paper might argue that the authors have introd uced un-
necessary complications to the theory of relativistic fluid dynamics b y adding
three more functions α,β,νto the standard set ⃗ v,ρ. In the following we will
show that this is not so and the action given in equation (70) in a form s uitable
for a pedagogic presentation can indeed be simplified.It is easy to sh ow that
defining a four dimensional Clebsch four vector:
vµ
C≡α∂µβ+∂µν= (1
c(α∂tβ+∂tν),α⃗∇β+⃗∇ν) = (1
c(α∂tβ+∂tν),⃗ vC) (94)
and a four dimensional electromagnetic Clebsch four vector:
vµ
E≡vµ
C+kAµ= (1
c(α∂tβ+∂tν+kφ),⃗ vC−k⃗A). (95)
It follows from equation (81) and equation (87) that:
vµ=−vEµ
λ⇒⃗ v=⃗ vE
λ. | 2311.00001v1.pdf | e5e5f464-40bc-42c0-900f-334b5af1fd7f |
(96)
Eliminating ⃗ vthe Lagrangiandensity appearingin equation(73) canbe written
(up to surface terms) in the compact form:
L[ρ0,α,β,ν] =ρ0[
c√
vEµvµ
E−ε0−c2]
(97)
This Lagrangian density will yield the four equations (74,77,81), afte r those
equations are solved we can insert the potentials α,β,νinto equation (87) to
obtainthephysicalvelocity ⃗ v. Hence,thegeneralchargedrelativisticbarotropic
fluid dynamics problem is changed such that instead of solving the Eule r and
continuity equations we need to solve an alternative set which can be derived
from the Lagrangian density ˆL.5 Conclusion
The current workwhich is a continuationofpreviousstudies [22,46,4 8] in which
we demonstrate how Pauli’s spinor can be interpreted in terms of spin fluid
using a generalized Clebsch form which is modified to include the electro mag-
netic vector potential affecting a charged fluid. The theory is desc ribed by an
action and a variational principle and the fluid equations are derived a s the
extrema of the action. The similarities as well as the pronounced diffe rences
with barotropic fluid dynamics were discussed.
A fundamental obstacle to the fluid interpretation of quantum mec hanics
still exist. This is related to the origin of thermodynamic quantities wh ich are
part of fluid mechanics in the quantum context. For classical fluid th e ther-
modynamic internal energy implies that a fluid element is not a point par ticle
but has internal structure. In standard thermodynamics notion s as specific | 2311.00001v1.pdf | c1adedeb-a1df-4fa6-aa39-1ebefda00556 |
enthalpy, pressure and temperature are derived from the specifi c internal en-
ergy equation of state. The internal energy is a required compone nt of any
Lagrangian density attempting to depict a fluid. The unique form of t he in-
ternal energy can be derived in principle relying on the basis of the at oms and
moleculesfrom whichthe fluid is composedandtheir interactionsusing statisti-
cal physics. However, the quantum fluid has no such microscopic st ructure and
yet analysis of both the spin less [5,6] and spin [22] quantum fluid shows that
terms analogue to internal energies appear. Thus one is forced to ask where
do those internal energies originate, surely the quantum fluid is dev oid of a mi-
croscopic sub structure as this will defy the empirically supported c onception
of the electron as a point particle. The answer to this inquiry originat ed from
measurement theory [43]. Fisher information a basic concept of mea surement
theory is a measure of the quality of the measurement.It was show n that this
concept is proportional to the internal energy of Schr¨ odinger’s spin-less elec-
tron which is essentially a theory of a potential flow which moves unde r the
influence of electromagnetic fields and fisher information forces. F isher infor-
mation can also explain most parts of the internal energy of an elect ron with
spin. This puts (Fisher) information as a fundamental force of nat ure, which
has the same status as electromagnetic forces in the quantum mec hanical level
of reality. Indeed, according to Anton Zeilinger’s recent remark to the press, it
is quantum mechanicsthat demonstratesthat information is more f undamental
than space-time.
We have highlighted the similarities between the variational principles o f
Eulerian fluid mechanics and both Schr¨ odinger’s and Pauli’s quantum m echan-
ics as opposed to classical mechanics. The former have only linear tim e deriva-
tives of degrees of freedom while that later have quadratic time der ivatives.The former contain terms quadratic in the vector potential ⃗Awhile the later
contain only linear terms.
While the analogies between spin fluid dynamics classic Clebsch fluid dy-
namics are quite convincing still there are terms in spin fluid dynamics t hat
lack classical interpretation. It was thus suggested that those t erm originate
from arelativistic Clebschtheory which wasthe main motivation to the current
paper. Indeed following the footsteps of the pervious papers [46 ,48] we may
replace the internal energy in equation (97) with a Lorentz invarian t Fisher
information term to obtain a new Lagrangian density of relativistic qu antum
mechanics of a particle with spin:
L[ρ0,α,β,ν] =ρ0[
c√
vEµvµ
E−c2]
−ℏ2
2m∂µa0∂µa0, a 0≡√ρ0
m. | 2311.00001v1.pdf | 92ec3686-0758-4bcd-b84c-1bc813a95cff |
(98)
in the above mis the particle’s mass and ℏis Planck’s constant divided by 2 π.
A side benefit of the above work is the ability to canonically derive the
stress energy tensor of a relativistic fluid.
As the current paper is of limited scope, we were not able to compare the
above lagrangian with its low speed limit and derive the relevant quantu m
equation, hopefully this will be done in a following more expanded paper .
Not less important is the comparison between the fluid route to relat ivistic
quantum mechanics and the more established route of the Dirac equ ation, this | 2311.00001v1.pdf | 609564b3-ffbb-4208-9535-5ec8af1aa7c8 |
certainly deserve and additional paper which I hope to compose in th e near
future.
References
1. Kant, I. (1781). Critik der reinen Vernunft.
2. D. Bohm, Quantum Theory (Prentice Hall, New York, 1966) section 12.6
3. P.R. Holland The Quantum Theory of Motion (Cambridge University Press, Cam-
bridge, 1993)
4. D. Durr & S. Teufel Bohmian Mechanics: The Physics and Mathematics of Quan-
tum Theory (Springer-Verlag, Berlin Heidelberg, 2009)
5. E. Madelung, Z. Phys., 40322 (1926)
6. R. Englman and A. Yahalom ”Complex States of Simple Molecu lar Systems” a
chapter of the volume ”The Role of Degenerate States in Chemi stry” edited by
M. Baer and G. Billing in Adv. Chem. Phys. Vol. 124 (John Wiley & Sons 2002).
[Los-Alamos Archives physics/0406149]
7.[Los-Alamos Archives physics/0406149]
7. W. Pauli (1927) Zur Quantenmechanik des magnetischen Ele ktrons Zeitschrift f¨ ur
Physik (43) 601-623
8. R. Englman, A.Yahalom and M. Baer, J. Chem. Phys.109 6550 ( 1998)
9. R. Englman, A. Yahalom and M. Baer, Phys. Lett. A 251223 (1999)
10. R. Englman and A. Yahalom, Phys. Rev. A 601802 (1999)
11. R. Englman, A.Yahalom and M. Baer, Eur. Phys. J. D 81 (2000)
12. R. Englman and A.Yahalom, Phys. Rev. B 612716 (2000)
13. R. Englman and A.Yahalom, Found. Phys. Lett. 13329 (2000)
14. | 2311.00001v1.pdf | af10f206-9712-4c1f-8896-f1031bf3f5bf |
Phys. Lett. 13329 (2000)
14. R. Englman and A.Yahalom, The Jahn Teller Effect: A Permanent Presence in
the Frontiers of Science in M.D. Kaplan and G. Zimmerman (editors), Proceed-
ings of the NATO Advanced Research Workshop, Boston, Sep. 20 00(Kluwer,
Dordrecht, 2001)
15. M. Baer and R. Englman, Chem. Phys. Lett. 33585 (2001)
16. A. Mebel, M. Baer, R. Englman and A. Yahalom, J.Chem. Phys .1153673 (2001)
17. R. Englman & A. Yahalom, ”Signed Phases and Fields Associ ated with De-
generacies” Acta Phys. et Chim., 34-35, 283 (2002). [Los-Al amos Archives -
quant-ph/0406194]
18.[Los-Al amos Archives -
quant-ph/0406194]
18. R. Englman, A. Yahalom and M. Baer,”Hierarchical Constr uction of Finite Di-
abatic Sets, By Mathieu Functions”, Int. J. Q. Chemistry, 90 , 266-272 (2002).
[Los-Alamos Archives -physics/0406126]
19. R. Englman, A. Yahalom, M. Baer and A.M. Mebel ”Some Exper imental. and
Calculational Consequences of Phases in Molecules with Mul tiple Conical Inter-
sections” International Journal of Quantum Chemistry, 92, 135-151 (2003).
20. R. Englman & A. Yahalom, ”Phase Evolution in a Multi-Comp onent System”,
Physical Review A, 67, 5, 054103-054106 (2003). [Los-Alamo s Archives -quant-
ph/0406195]
21.[Los-Alamo s Archives -quant-
ph/0406195]
21. R. Englman & A. Yahalom, ”Generalized ”Quasi-classical ” Ground State of an
Interacting Doublet” Physical Review B, 69, 22, 224302 (200 4). [Los-Alamos
Archives - cond-mat/0406725]
22. A. Yahalom ”The Fluid Dynamics of Spin”. Molecular Physi cs, Published online:
13 Apr 2018.
http://dx.doi.org/10.1080/00268976.2018.1457808
(arXiv:1802.09331 [physics.flu-dyn]).
23. Clebsch, A., Uber eine allgemeine Transformation der hy drodynamischen Gle-
ichungen. J. reine angew. Math. 1857,54, 293–312. | 2311.00001v1.pdf | 711387a9-c1d4-4118-bff9-1160e12bd64a |
24. Clebsch, A., Uber die Integration der hydrodynamischen Gleichungen. J. reine
angew. Math. 1859,56, 1–10.
25. B. Davydov, ”Variational principle and canonical equat ions for an ideal fluid,”
Doklady Akad. Nauk , vol. 69, 165-168, 1949. (in Russian)
26. C. Eckart, ”Variation Principles of Hydrodynamics ,” Phys. Fluids , vol. 3, 421,
1960.
27. F.P. Bretherton ”A note on Hamilton’s principle for perf ect fluids,” Journal
of Fluid Mechanics / Volume 44 / Issue 01 / October 1970, pp 19 3 1 DOI:
10.1017/S0022112070001660, Published online: 29 March 20 06.
28. J. W. Herivel Proc. Camb. Phil. Soc., 51, 344 (1955)
29.Phil. Soc., 51, 344 (1955)
29. J. Serrin, ‘Mathematical Principles of Classical Fluid Mechanics’ inHandbuch der
Physik,8, 148 (1959)
30. C. C. Lin , ‘Liquid Helium’ inProc. Int. School Phys. XXI (Academic Press)
(1963)
31. R. L. Seliger & G. B. Whitham, Proc. Roy. Soc. London , A305, 1 (1968)
32. D. Lynden-Bell and J. Katz ”Isocirculational Flows and t heir Lagrangian and
Energy principles”, Proceedings of the Royal Society of Lon don. Series A, Math-
ematical and Physical Sciences, Vol. 378, No. 1773, 179-205 (Oct. 8, 1981).
33. J. Katz & D. Lynden-Bell 1982, Proc. R. Soc. Lond. A 381263-274.
34.R. Soc. Lond. A 381263-274.
34. Asher Yahalom and Donald Lynden-Bell ”Variational Prin ciples for Topological
Barotropic Fluid Dynamics” [”Simplified Variational Princ iples for Barotropic
Fluid Dynamics” Los-Alamos Archives - physics/ 0603162] Ge ophysical & Astro-
physical Fluid Dynamics. 11/2014; 108(6). DOI: 10.1080/03 091929.2014.952725.
35. Weinberg, S. Gravitation and Cosmology: Principles and Applications of the Gen-
eral Theory of Relativity ; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 1972.
36. Misner, C.W.; Thorne, K.S.; Wheeler, J.A. Gravitation ; W.H. Freeman & Com-
pany: New York, NY, USA, 1973.
37. Padmanabhan, T. Gravitation: Foundations and Frontiers ; Cambridge University
Press: Cambridge, UK, 2010.
38.38. H. Goldstein , C. P. Poole Jr. & J. L. Safko, Classical Mech anics, Pearson; 3
edition (2001).
39. Lynden-Bell, D., Consequences of one spring researchin g with Chandrasekhar.
Current Science 1996,70(9), 789–799.
40. Lamb, H., Hydrodynamics , 1945 (New York: Dover Publications).
41. H.K. Moffatt, ”The degree of knottedness of tangled vorte x lines,” J. Fluid Mech. ,
vol. 35, 117, 1969.
42. E. Schr¨ odinger, Ann. d. Phys. 81109 (1926). English translation appears in E.
Schr¨ odinger, Collected Papers in Wave Mechanics (Blackie and Sons, London,
1928) p. 102
43. R. A. Fisher Phil. Trans. R. Soc. London 222, 309.
44. | 2311.00001v1.pdf | 2d00de2d-0839-4444-832b-e06036c6a1d2 |
Trans. R. Soc. London 222, 309.
44. A. Yahalom ”Gravity and the Complexity of Coordinates in Fisher Information”
International Journal of Modern Physics D, Vol. 19, No. 14 (2 010) 2233-2237,
©World Scientific Publishing Company DOI: 10.1142/S0218271 810018347.
45. L. Mandel andE. Wolf, Optical Coherence and Quantum Optics (UniversityPress,
Cambridge, 1995) section 3.1
46. A. Yahalom ”The Fluid Dynamics of Spin - a Fisher Informat ion Perspective”
arXiv:1802.09331v2 [cond-mat.] 6 Jul 2018. Proceedings of the Seventeenth Is-
raeli - Russian Bi-National Workshop 2018 ”The optimizatio n of composition,
structure and properties of metals, oxides, composites, na no and amorphous ma-
terials”.
47. B. R. Frieden Science from Fisher Information: A Unification (Cambridge Uni-
versity Press, Cambridge, 2004) | 2311.00001v1.pdf | 3c14e4e5-738b-40e3-8f8f-31914dc46ee7 |
48. Asher Yahalom ”The Fluid Dynamics of Spin - a Fisher Infor mation Perspective
and Comoving Scalars” Chaotic Modeling and Simulation (CMS IM) 1: 17-30,
2020.
49. Yahalom, A. Fisher Information Perspective of Pauli’s E lectron. Entropy 2022,
24, 1721. https://doi.org/10.3390/e24121721.
50. Yahalom, A. (2023). Fisher Information Perspective of P auli’s Electron. In: Ski-
adas, C.H., Dimotikalis, Y. (eds) 15th Chaotic Modeling and Simulation Interna-
tional Conference. CHAOS 2022. Springer Proceedings in Com plexity. Springer,
Cham. https://doi.org/10.1007/978-3-031-27082-6 26 | 2311.00001v1.pdf | 63ffb586-729d-4a7a-b229-a27a13eab726 |
arXiv:2311.00002v1 [math.NT] 27 Oct 2023THE TRIANGULAR NUMBERS ARE FINITELY STABLE
Luan Alberto Ferreira
Abstract. This small note proves that the set of triangular numbers is a finitely stable
additive basis. This, together with a previous result by the author, shows that triangular
numbers and squares are, among all polygonal numbers, the on ly ones that are finitely stable
additive bases.
1.Introduction
IfA,B⊆N={0,1,2,3,4,5,...}, we define
A+B={a+b;a∈A, b∈B}.
Also, ifh∈N,h̸= 0, then we write
hA=A+···+A
htimes={a1+···+ah;a1,...,a h∈A}.
If there is h∈N,h̸= 0, such that hA=N, thenAis said to be an additive basis.The smallest h∈Nsatisfying this equality is called the order of A, and is denoted by
h=o(A). For example, Lagrange’s four-square theorem says that the set of the squares
N2={0,1,4,9,16,25,...}is an additive basis whose order is four, i.e., o(N2) = 4. Finally,
ifA⊆Nandn∈N, then we write
A(n) = #{a∈A; 1≤a≤n}.
In [1], it was defined that an additive basis Aisfinitely stable if the order of Ais equal
the order of A∪F, for all finite subset F⊆N, and it was proved the following theorem:
Theorem 1.1. LetAbe an additive basis such that o(A) =h≥3. If
lim
n→∞((h−2)A)(n)
n= 0
and
limsup
n→∞((h−1)A)(n)
n<1,
thenAis finitely stable. | 2311.00002v1.pdf | 4e665c6c-7549-46e7-b6b9-80e238acb5c0 |
Using this result, it was shown that the set of the squares N2is a finitely stable additive
basis. But theorem 1.1also implies that the set of triangular numbers T={0,1,3,6,10,...}
is finitely stable. The purpose of this humble note is to prove this and to show that among
all polygonal numbers, the triangular numbers and the squar es are the only ones that are
finitely stable additive bases.
1 | 2311.00002v1.pdf | bd47898e-b892-49b1-96c0-480d718510b1 |
2 THE TRIANGULAR NUMBERS ARE FINITELY STABLE
2.The result
For the main result we will need the following facts. For comp leteness, a proof of the
second will be given.
Theorem 2.1 (Gauss’ num = ∆+∆+∆ theorem) .The set of the triangular numbers Tis
an additive basis of order 3.
Proof. See [3], chapter 1, section 1.7. □
Proposition 2.2. Ifn≡5or8 (mod 9) , thennis not the sum of two triangular numbers.
Proof. Suppose by absurd that n≡5 (mod 9) is the sum of two triangular numbers. Then
there exists m,x,y∈Nsuch that
9m+5 =x(x+1)
2+y(y+1)
2.
This implies
18m+10 =x(x+1)+y(y+1).
Taking this equation mod 9,
1≡x(x+1)+y(y+1) (mod 9) .
But this equation does not have a solution mod 9, contradiction.But this equation does not have a solution mod 9, contradiction. Analogously for the case
n≡8 (mod 9) . □
Corollary 2.3. The triangular numbers Tis a finitely stable additive basis.
Proof. Sinceo(T) = 3, the previous proposition shows that the hypotheses of theo rem1.1
are fulfilled for T. ThenTis a finitely stable additive basis. □
This corollary, together with the next two results, show tha t the only polygonal numbers
that are finitely stable additive bases are the squares N2and triangular numbers T.
Theorem 2.4 (Fermat-Cauchy polygonal number theorem) .Letk∈N,k≥3. IfNk
denotes the set of the k-gonal numbers, then o(Nk) =k.
Proof. See [2]. □
Theorem 2.5 (Legendre polygonal number theorem) .Letm≥3andN≥28m3. | 2311.00002v1.pdf | 55469897-1737-4667-9c5e-8a21be9bc3e3 |
Ifmis
odd, then Nis the sum of four polygonal numbers of order m+2. Ifmis even, then Nis
the sum of five polygonal numbers of order m+2, at least one of which is 0or1.
Proof. See [3], chapter 1, section 1.7. □
Corollary 2.6. The only polygonal numbers that are finitely stable additive bases are the
squaresN2and triangular numbers T. | 2311.00002v1.pdf | 1db7925b-e2ad-405b-9c64-6520244663e4 |
THE TRIANGULAR NUMBERS ARE FINITELY STABLE 3
Proof. Fixm∈N,m≥3. Now consider the set
Nm+2∪{j∈N;j <28m3}
and compare the order of this set with the order of Nm+2. □
References
[1] Ferreira, L. A.: Finitely stable additive bases. Bulletin of the Australian Mathematical Society (3)97
(2018), 360–362.
[2] Nathanson, M. B.: A short proof of Cauchy’s polygonal num ber theorem. Proceedings of the American
Mathematical Society (1)99(1987), 22–24.
[3] Nathanson, M. B.: Additive Number Theory: The Classical Bases . Grad. Texts in Math. 164, Springer,
New York, 1996.
Instituto Federal de Educação, Ciência e Tecnologia de São Pa ulo, campus Itaquaquece-
tuba. IFSP - ITQ: Rua Primeiro de Maio, 500, Itaquaquecetuba - SP, Brazil.CEP: 08571-050
Email address :[email protected] | 2311.00002v1.pdf | 1c32e2b1-8c3b-49c0-a21d-886460974af7 |