Degree project
Optimality of uncertainty principles for joint time- frequency representations
Author: Leonie Callies
Supervisor: Patrik Wahlberg
Examiner: Joachim Toft
Date: 2014-06-10
Course Code: 4MA11E
Subject: Mathematics
joint time-frequen y representations
Leonie Callies
June 17, 2014
Abstra t
Thestudyofjointtime-frequen y representationsis alargeeld ofmathemati s
andphysi s,espe iallysignalanalysis. BasedonHeisenberg's lassi alun ertainty
prin iplevariousinequalitiesforsu htime-frequen ydistributions havebeenstud-
ied. The obje tive of this thesis is to examine the role that Gaussian fun tions,
in luding those with a hirp ontribution, play in inequalities for the Short-Time
Fourier transformand the Wignerdistribution. We show that Grö henig'sun er-
taintyprin iplesfortheShort-TimeFouriertransformarenotoptimalwithregardto
thesefun tions. AsfortheWignerdistributionweshowhowanexistingun ertainty
prin iplebyJanssen anbemodiedtorea hoptimalityforChirpGaussians.
1 Introdu tion 3
2 Time-Frequen y representations 4
2.1 TheFourier Transform . . . 4
2.2 TheShort-Time Fourier Transform (STFT) . . . 6
2.3 Time-Frequen y representations ofCohen's lass . . . 7
3 Un ertainty prin iplesfor the pair
(f, ˆ f )
10 3.1 Heisenberg'sun ertainty prin iple . . . 103.2 Un ertaintyprin iple byCohen . . . 10
3.3 Un ertaintyprin iple byDonohoand Stark . . . 11
4 Un ertainty prin iplesfor the Short-Time Fourier Transform 12 4.1 Grö henig's inequalities . . . 12
4.1.1 Originalversion . . . 12
4.1.2 Adaptation to thealternative denitionof theSTFT . . . 13
4.2 Non-optimalityof thestrong un ertainty prin iple . . . 15
4.2.1 Gaussian fun tions . . . 15
4.2.2 ChirpGaussians . . . 19
5 Un ertainty prin iplesfor the Wignerdistribution 22 5.1 Janssen'sun ertaintyprin iple . . . 22
5.2 Equality inJanssen's prin iple. . . 23
5.3 Astronger lowerboundfor the ase
M = I
. . . . . . . . . . . . . . . . . 275.4 Astronger lowerboundfor a general matrix
M
. . . . . . . . . . . . . . . 286 Con lusion 31
Referen es 32
Appendix 33
In the early 19 th
entury Jean-Baptiste Joseph Fourier laid the foundations for what is
now alled Fourier Analysis. He dis overed that periodi fun tions an be expressedas
seriesofsineand osinefun tionsofdis retefrequen ies,theso- alledFourierseries. The
equivalentexpressionfornon-periodi fun tionsistheFouriertransformationwhi hhasa
ontinuousratherthandis retefrequen yspe trum. Theappli ationsofFourieranalysis
are ountless, both inphysi s andmathemati s. Inthis work we willbe on erned with
theeldof signalanalysis. Hen e,fun tionswilloftenbe alledsignalsand the ommon
variablesaretime andfrequen y.
When a fun tion des ribes the behavior ofa signal in time, thenthe Fourier transform
des ribesits behaviorinthe frequen y domain. Both fun tionsdes ribe thesignal om-
pletely be ause ea h of the representations an be omputed from the other, but when
dealing with only one of them it la ks an important part of the information. It would
be desirable to simultaneouslydes ribe thesignal'sbehavior inthetime and inthefre-
quen y domain, even determine the frequen y at a spe i point in time, a so- alled
instantaneous frequen y. However, a ording to the un ertainty prin iple dis overed by
WernerKarlHeisenberginthemiddleofthe20 th
enturyitisimpossibletoexpresssu h
an instantaneous frequen y. His inequality was originally formulated for quantities in
quantum me hani s. Yet, it an be arriedoverto signalanalysis.
The issue of joint time-frequen y representation is a question in ongoing resear h and
various ways have been found. Among them are the Short-Time Fourier Transform
(STFT),distributionsofCohen's lassliketheWignerdistribution,Gaboranalysisorthe
Wavelettransformtonameonlyafew. InequalitiesanalogoustoHeisenberg'sun ertainty
prin iple arevalidfor su h joint representations.
Chapter two and three serve as an introdu tion to time-frequen y representations and
un ertainty prin iples. Subsequently, we fo us on inequalities for theSTFT derived by
Karlheinz Grö henig and show that they are not optimal for Gaussian fun tions. The
last hapterisdevotedtoanun ertaintyprin iple fortheWignerdistributionthatisdue
to A.J.E.M. Janssenand theroleof Gaussian fun tionsinhisinequality.
This hapter serves as an introdu tion to the basi s of Fourier and Time-Frequen y
Analysis. In general analysis is done on
L p
, the Bana h spa e of measurable fun tionson themeasurespa e
S
where the normis hosen to bekf k p := Z
S |f (x)| p dx 1 p
with
1 ≤ p < ∞
. Mostoftenwewill on ernourselveswithL 2
whi hturnsintoaHilbertspa eafter introdu ing the innerprodu tof
f, g ∈ L 2
ashf, gi =
Z
S
f (x)g(x) dx.
Two fundamental inequalities ne essary for the study of un ertainty prin iples are due
to Hölder and Cau hy. They aretaken for granted and used without proof throughout
this work.
Theorem 2.1 (Hölder's inequality). Let
S
be a measure spa e and1 ≤ p, q ≤ ∞
with1
p + 1 q = 1
. Letfurtherf ∈ L p (S)
andg ∈ L q (S)
. Thenf g ∈ L 1 (S)
andkf gk 1 ≤ kf k p kgk q .
(1)Cau hy'sinequality isa spe ial aseof Hölder'sstatement:
Theorem 2.2 (Cau hy-S hwarz inequality).
|hf, gi| ≤ kf k 2 kgk 2 .
(2)Thisisobtained bysetting
p = q = 2
inHölder's inequality:|hf, gi| = | Z
S f (x)g(x)dx| ≤ Z
S |f (x)g(x)| dx = kf gk 1 ≤ kf k 2 kgk 2 .
Themeasurespa e
S
inthisdis ussionisusuallythespa eR
,equippedwiththeLebesgue measure, andthe fun tionsf
andg
arereal- or omplex-valued. Mostresults an easily be transferred to higher dimensionsbut this is not done hereto not ompli ate mattersunne essarily. The quantity
|f (x)| 2
is often alled the energy density of the fun tionf
. Asexplained by Cohenin [Coh95 ℄ the namehas its origin in physi s. Iff
des ribesthe ele tri eld, then the quantity
|f (x)| 2
orresponds to the ele tri energy density. Analogies an be foundinother areasof physi s, aswell.2.1 The Fourier Transform
All time-frequen y representations onsidered inthe following are basedon theFourier
transform. A little in onvenien e of time-frequen y analysisis theexisten e of multiple
versions of the denition of the Fourier transform. Consequently this arries over to
the denitions of the time-frequen y representations and makes omparisons of works
dieronlyinamultipli ativefa torandafa torintheexponential. CohenandDanguse
equation(3)fromdenition2.3intheirwork([Coh95 ℄and[PGT13℄),whereasGrö henig,
JanssenandCarypisuseequation(4)intheirwork([Grö01 ℄,[FS03 ℄,[Jan89 ℄and[Car14 ℄).
Denition 2.3. TheFourier transform of afun tion
f ∈ L 1 (R)
isdened asf (ω) = ˆ 1
√ 2π Z
R
f (x)e −ixω dx,
(3)alternatively:
f (ω) = ˆ Z
R
f (x)e −2πixω dx.
(4)Asnotation for theFourier transform of
f
bothf ˆ
andF[f ]
willbe used.The rst denitionis theone to be usedin this report. The results found by thelatter
three authors need therefore be adaptedto thedierent denitionof the Fourier trans-
form.
Gaussian fun tions play an important role in this report. Hen e, it is worth to note
Hörmander's theorem on the Fourier transform of a omplex-valued Gaussian fun tion
(see [Hör85 ℄)
Theorem 2.4 (Fourier transform of a omplex-valued Gaussian). Let
f : R → C, x 7→
e −dx 2 /2
withd ∈ C
andRe(d) ≥ 0
. ThentheFouriertransformoff
isanotherGaussianfun tion
f : R ∋ ω 7→ d ˆ − 1 2 e − ω2 2d .
(5)If
Re(d) = 0
thenf : R ∋ ω 7→ |d| ˆ − 1 2 e πi 4 +i 2Im(d) ω2 .
(6)Remark 2.5. For
Im(d) = 0
andRe(d) > 0
denition (5 )redu es tof (ω) = ˆ √ 1
2π
R
R f (x)e −iωx dx
asinequation (3) .Afewpropertiesareessentialfor thestudy ofFourier analysis. TheFourier transformis
invertible andtheinversionisgiven by
Theorem 2.6 (InversionFormula). Let
f ∈ L 1 (R)
andf ∈ L ˆ 1 (R)
, thenf (x) = 1
√ 2π Z
R
f (x)e ˆ ixω dω.
Remark2.7. Ingeneralthe Fourier transformis amapfrom
L 1
to thespa e of ontin-uousfun tionsthatvanishat innity. Inthat aseadditional requirementsfor
f
havetobe setsothatthe inversionformulaholds. However,thisisnot the aseforthefun tions
thatareof relevan efor thisreport.
Ifin addition
f ∈ L 2 (R)
, thenthe Fourier transform hasthepropertyof preservingtheenergy ofa signal. Thisstatement is dueto Plan herel.
Theorem 2.8 (Plan herel). Let
f ∈ L 1 ∩ L 2 (R)
. Thenkf k 2 = k ˆ f k 2
.An important operator for thefollowing studyis the onvolutionof fun tions.
Denition 2.9 (Convolution). Let
f, g ∈ L 1 (R)
. Then(f ∗ g)(x) =
Z
R f (t)g(x − t) dt
is alledthe onvolution of
f
andg
.Theorem 2.10 (Convolutiontheorem). Let
f, g ∈ L 1 (R)
. The Fourier transformof theonvolved fun tions is
F[f ∗ g] = √
2πF[f ]F[g].
Thestatement isproven bysimple al ulation usingFubini's theorem.
2.2 The Short-Time Fourier Transform (STFT)
In signal analysis the fun tion
f
represents for instan e an a ousti signal. It ontainsinformationaboutthetemporalbehavior,inthe aseofthea ousti signaltheamplitude
overtime, whereas the Fourier transform ontains information about the frequen y be-
havior, thatis, whi hfrequen ieswere present. Unfortunately,neitherfun tionprovides
any information whatsoever about the hange of the spe trum over time. However, in
many ases this is very relevant. An analogy that Grö henig introdu es motivates the
sear hforsimultaneoustime-frequen y representations. Composershavefound awayto
express timeand frequen y behavior at the same time sothat musi ians areable to re-
onstru tthe omposer'sideasfromasheetofmusi . Foramusi iantheverti alposition
of the note in the staves provides information about the frequen y and the horizontal
arrangement ofthenotesinformationaboutthedurationofea hnoteandthetimewhen
thefrequen y isto beplayed.
Arstandratherintuitiveattempttoa hievethesameintime-frequen yrepresentation
is the Short-Time Fourier transform (STFT). The idea behind it is to emphasize the
fun tion of interest
f
on smaller subsets of the domain before determining the Fourier transform. TheSTFT anberegardedasalo alversionofthe lassi alFouriertransform.Toemphasize thefun tion,
f
ismultipliedbyaso- alled window fun tiong
. Thismightbe a fun tion
g(x) ≡ 1
on a ompa t set that goes to zero smoothly outside this set.Other frequent window fun tions are Gaussian fun tions. The hoi e of the window is
ru ialand highlydependsonthe respe tive obje tive. It isimportant thatthewindow
fun tion emphasizes
f
but at thesame time doesn't alter it signi antly on thesubset.An illustrationofthis method an beseen ingure1.
Denition 2.11. The Short-Time Fourier transform of
f ∈ L 2 (R)
with respe t tog ∈ L 2 (R)
is dened asV g f (x, ω) = 1
√ 2π Z
R f (t)g(t − x)e −itω dt,
(7)−1.5 0 −1 −0.5 0 0.5 1 1.5 2 2.5 3 0.2
0.4 0.6 0.8 1 1.2 1.4
t
g(t−x) f(t)
g(t)
f(t)g(t−x)
Figure1:Visualizationofanexemplary fun tion
f
,awindowfun tiong
andtheir prod-u twhi h isneededto determine theSTFT
or alternatively Grö henig'sversion:
V g f (x, ω) = Z
R f (t)g(t − x)e −2πitω dt
(8)with
x, ω ∈ R
.Remark 2.12. The STFT is not limited to fun tions in
L 2 (R)
. Depending on therespe tive ontext other spa es su h as
L 1 (R)
,the S hwartz spa eS(R)
or thespa e ofdistributions
S ′ (R)
an beused, aswell.To simplifyexpressions like (7)the two operators
T x
andM ω
areintrodu ed su h thatT x f (t) = f (t − x),
M ω f (t) = e iω·t f (t).
The operator
T x
is alleda time-shift andM ω
a frequen y-shift. Equation (7 ) an then be written asV g f (x, ω) = 1
√ 2π hf, M ω T x gi,
provided thatthe hosenfun tion spa eis aHilbertspa e withan innerprodu t.
2.3 Time-Frequen y representations of Cohen's lass
Starting from the 1940s onsiderable progress was made in the study of joint time-
frequen y distributions. In the1960sLeon Cohendis overed thatmany ofthedistribu-
tions that hadbeen found ould be derived byone and thesame method and unitedin
a lass that is now alled Cohen's lass. Cohen presents the method, the lass and an
overviewofthe most important distributions in[Coh95 ℄.
C f (x, ω) = 1 4π 2
Z Z Z
f (u + 1
2 τ )f (u − 1
2 τ )Φ(θ, τ )e −iθx−iτ ω+iθu du dτ dθ.
Thefun tion
Φ(θ, τ )
is alledthekernel ofthedistribution.Distributions of Cohen's lass are often alled quadrati or bilinear be ause the signal
is bilinearly involved. The properties of the distribution are determined by the kernel.
Dependingontheappli ationdierentpropertiesmightbedesirable. Alldistributionsof
Cohen's lasshave ertainshift-propertiesin ommon,thatis, translationsand modula-
tions of the signalimply orresponding shiftsof thedistribution [MH97℄. Janssenoers
a omprehensivedis ussionof furtherpropertiesin[MH97℄. A ompressed versionofhis
listisgiven inthe following. For details see, for instan e,Janssen'sor Cohen's work.
It isdesirable thatthefollowing onditions holdfor all
f
andfor allx, ω ∈ R
:•
Corre t marginal onditionsZ
R
C f (x, ω) dω = |f (x)| 2 Z
R
C f (x, ω) dx = | ˆ f (ω)| 2
•
Corre t total energyZ Z
R 2
C f (x, ω) dx dω = kf k 2
•
Weak supportpropertyf (x) = 0, |x| > a ⇔ C f (x, ω) = 0, |x| > a, ω ∈ R, a > 0 f (ω) = 0, ˆ |ω| > b ⇔ C f (x, ω) = 0, |ω| > b, x ∈ R, b > 0
•
Strongsupportpropertyf (x) = 0 ⇔ C f (x, ω) = 0, ω ∈ R f (ω) = 0 ˆ ⇔ C f (x, ω) = 0, x ∈ R
•
Real-valuednessC f (x, ω) ∈ R, x, ω ∈ R
•
Moyal's onditionZ Z
R 2
C f (x, ω)C g (x, ω) dx dω = |hf, gi| 2
•
Non-negativity onditionC f (x, ω) ≥ 0, x, ω ∈ R
dilation, onvolution and the Fourier transform might be required for a time-frequen y
distribution. The best-known time-frequen y distribution of this lass is the Wigner
distribution. It was introdu ed by Eugene Wigner in the 1930s and has been of great
importan e eversin e.
Denition 2.14. The Wignerdistribution ofa fun tion
f ∈ L 2 (R)
is dened asW f (x, ω) = 1
2π Z
R
f (x + y
2 )f (x − y
2 )e −iωy dy.
(9)This an be obtained from the general time-frequen y distribution of theorem 2.13 by
settingthe kernel
Φ(θ, τ ) = 1
.An alternative,widely useddenitionof theWignerdistribution is
W f (x, ω) = Z
R
f (x + y
2 )f (x − y
2 )e −2πiωy dy.
(10)ThisversionisbasedontheFouriertransformasinequation(4 )andusedbyGrö henig,
Janssenand Carypis.
There is no perfe t time-frequen y representation be ause no distribution of Cohen's
lass an fulll all the requirements mentioned above asexplained indetail in [MH97℄.
Infa t,foreverydistributionofCohen's lassthatsatisesthemarginal onditionsthere
is at least one fun tion
f
su h that the distribution takes negative values. However, among Cohen's distributions the Wigner distribution is losest to being non-negative.Moreover, it an be shown that the Wigner distribution satises ten out of the twelve
onditions in Janssen's list. The fa t that it an take on negative values ould be seen
asitsmost signi ant aw be ause thisleads tointerpretation di ultiesinphysi s. As
explainedbyJanssenin[Jan89℄this an be over ome byweightingthedistributionwith
ertain square-integrable fun tions. The se ond property that the Wigner distribution
doesn't satisfy is the strong support ondition. A ommon inherent attribute of all
bilineardistributions areso- alled ross-terms. Theseareterms thatareex lusively due
to the bilinear involvement of the fun tion but la ka physi al meaning and hen e lead
to interpretationdi ulties.
The STFT isnot a dire tmember of Cohen's lassbut thesquare of its absolutevalue
is. Thisdistributionis alledtheSpe trogram.
3 Un ertainty prin iples for the pair
(f, ˆ f )
3.1 Heisenberg's un ertainty prin iple
In1927 thephysi istWernerHeisenbergformulated hisfamousun ertaintyprin iple for
quantumme hani s thatstatesthat ertainpairsof quantitiessu haspositionand mo-
mentum ofa parti le or energy andtime annot both bedetermined arbitrarilypre ise.
There are numerous un ertainty prin iples in the literature, not only in quantum me-
hani sbut alsoin otherelds ofphysi sand mathemati s,manyof whi h arebasedon
Heisenberg'sun ertainty prin iple. The versionof interest of Heisenberg's prin iple for
this study isaninequalityfor afun tion and itsFourier transform.
Theorem 3.1 (Heisenberg's un ertainty prin iple). Let
f ∈ L 2 (R)
anda, b ∈ R
arbi-trary. Then
Z
R (x − a) 2 |f (x)| 2 dx 1/2 Z
R (ω − b) 2 | ˆ f (ω)| 2 dω 1/2
≥ 1
2 kf k 2 2
(11)Equalityholdsifandonlyif
f (x) = De ib(x−a) · e − (x−a)2 2c
forsomeD ∈ C
andc ∈ R, c > 0
.The fun tions
|f | 2
and| ˆ f | 2
an be regarded asprobability densityfun tions (aslongasf ∈ L 2 (R)
andkf k 2 = 1
). The integrals in(11 ) an thenbeseen asthevarian e of|f | 2
and
| ˆ f | 2
. The prin iple states that if eitherf
orf ˆ
is strongly on entrated around a point,inotherwords,its varian e issmall,theotherone annotbe on entrated aswell,inotherwordsits varian ehasto be large.
Thisinterpretation suggeststhe widelyusedand insignalanalysispopularnotation
σ x σ ω ≥ 1
2 ,
(12)provided
kf k 2 = 1
. The quantitiesσ x 2
andσ 2 ω
are thevarian e of the fun tion and theFouriertransform,respe tively. They analsobeseenasthedurationandthebandwidth
ofthe signal.
3.2 Un ertainty prin iple by Cohen
StartingfromHeisenberg's lassi alun ertaintyprin ipleCohenderivesastrongerbound
in[Coh95 ℄. Inhis onsiderations hexes thearbitraryparameters
a
andb
asa = hxi =
Z
R x|f (x)| 2 dx b = hωi =
Z
R ω| ˆ f (ω)| 2 dω.
Thevarian es intime andfrequen y (orduration and bandwidth)thenbe ome
σ x 2 = Z
R (x − hxi) 2 |f (x)| 2 dx σ ω 2 =
Z
R (ω − hωi) 2 | ˆ f (ω)| 2 dω.
Inhisworkhe onsiderssignals
f (x) = ρ(x)e iϕ(x)
whereρ(x)
istheamplitudeofthesignalovertime and
ϕ(x)
a real-valuedfun tion des ribing thefrequen y. The ovarian e of a signalis dened asCov = Z
R
xϕ ′ (x)|f (x)| 2 dx − hxihωi.
Using thesedenitions Cohen statesand provesthefollowing un ertainty prin iple:
Theorem 3.2 (Cohen's un ertainty prin iple). Let
f ∈ L 2 (R)
,kf k 2 = 1
. Thenσ x σ ω ≥ 1
2
p 1 + 4Cov 2 .
(13)Equalityisattainedonlyforfun tionsoftheform
f (x) = Ce −a(x−hxi) 2 e −i(b(x−hxi) 2 +hωix)
,
where
a, b ∈ R, a > 0
andC ∈ C
su h thatkf k 2 = 1
.3.3 Un ertainty prin iple by Donoho and Stark
Another instrument to express theun ertainty dis overed by Heisenberg is thesupport
of afun tion. A qualitative versionofthe un ertainty prin iple is given bythetheorem
ofBenedi ks([FS03 ℄):
Theorem 3.3. Assume
f ∈ L 1 (R)
(orf ∈ L p (R)
). If|suppf | < ∞
and|supp ˆ f | < ∞
,then
f ≡ 0
.Sin eafun tionanditsFourierTransform annotbothhave ompa t support,asstated
intheorem3.3, itis onvenient to dene theessential support of afun tion asin[FS03℄
whi histhesubsetofthedomainthat oversmostofthefun tion. Thishelpstoformulate
quantitative un ertainty prin iples.
Denition 3.4 (essential support). A fun tion
f ∈ L 2 (R)
isε
- on entrated on a mea- surablesetT ⊂ R
,ifZ
T c |f (x)| 2 dx 1/2
≤ εkf k 2
with
0 ≤ ε ≤ 1
.T
is alledthe essential support off
.DonohoandStark introdu ed anun ertaintyprin iple basedontheideaoftheessential
support[Grö01℄.
Theorem 3.5. Suppose that
f ∈ L 2 (R)
,f 6= 0
isε T
- on entrated onT ⊂ R
andf ˆ
isε Ω
- on entrated onΩ ⊂ R
. Then|T ||Ω| ≥ (1 − ε T − ε Ω ) 2 .
Thisrelation istherefore alledan essential support ondition.
Transform
Asall time-frequen y representations are basedon theFourier transform itis only nat-
ural that un ertainty prin iples for the pair
(f, ˆ f )
arry over to joint time-frequen y distributions. This hapterisdevotedto inequalitiesfor theSTFT thatwerederived byKarlheinz Grö henig.
4.1 Grö henig's inequalities
4.1.1 Original version
Theweak un ertainty prin ipleasproposedbyGrö henigin[Grö01 ℄isanother example
oftheuseoftheessentialsupportintrodu edindenition3.4 . Inthefollowingtheresults
byGrö heningwillbepresentedusinghisalternativedenitionsoftheFouriertransform
and theSTFT(equation (4)and (8 )).
Theorem 4.1 (Weak un ertainty prin iple). Suppose that
kf k 2 = kgk 2 = 1
and thatU ⊂ R 2
and0 ≤ ε ≤ 1
are su h thatZ Z
U |V g f (x, ω)| 2 dx dω ≥ 1 − ε.
Then
|U| ≥ 1 − ε
where|U|
isthe measure ofU
.A stronger estimate for the essential support of
|V g f | 2
an be derived by using Lieb'sinequalities as they are stated and proven in [Grö01 ℄. The proof is in luded in the
appendix.
Theorem 4.2 (Lieb). Let
f, g ∈ L 2 (R)
, thenkV g f k p p =
Z Z
R 2 |V g f (x, ω)| p dx dω
( ≤ 2 p
(kf k 2 kgk 2 ) p
if2 ≤ p < ∞,
≥ 2 p
(kf k 2 kgk 2 ) p
if1 ≤ p ≤ 2.
The rstinequalityallowed Grö henigto establishthefollowing stronger versionof the
weak un ertainty prin iple 4.1 :
Theorem 4.3(Strongun ertaintyprin iple). Suppose that
kf k 2 = kgk 2 = 1
. IfU ⊂ R 2
and
0 ≤ ε ≤ 1
are su hthatZ Z
U |V g f (x, ω)| 2 dx dω ≥ 1 − ε,
then
|U| ≥ (1 − ε) p−2 p p 2
p−2 2
for all p > 2.
(14)In parti ular,
|U| ≥ sup
p>2 (1 − ε) p−2 p p 2
p−2 2
≥ 2(1 − ε) 2 .
(15)Let
f ˆ C
bethe Fourier transforma ordingto equation(3)andf ˆ G
theFourier transformasdenedinequation(4)where
C
abbreviatesCohenandG
Grö henig. Lettherespe -tiveShort-timeFourier transformsbe
V g C f
andV g G f
. Therelationbetween thedierentdenitions aregiven by:
f ˆ C (ω) = 1
√ 2π f ˆ G ( ω
2π )
(16)V g C f (x, ω) = 1
√ 2π V g G f (x, ω
2π ).
(17)Inthefollowingtheindex
”C”
willbeomittedbutitisunderstoodthatCohen'sdenitionisused. Theweak un ertainty an berestated as
Theorem4.4(Theweakun ertaintyprin iple-adapted). Supposethat
kf k 2 = kgk 2 = 1
andthat
U ⊂ R 2
and0 ≤ ε ≤ 1
are su h thatZ Z
U |V g f (x, ω)| 2 dx dω ≥ 1 − ε.
Then
|U| ≥ 2π(1 − ε)
where|U|
is the measure ofU
.Proof. The proof is essentially the same as given by Grö henig in [Grö01 ℄. Using the
denitionoftheSTFT(2.11 ),Cau hy'sinequality2.2andtheisometryoftime-frequen y
shiftsthefollowing holds:
|V g f (x, ω)| = | 1
√ 2π Z
R f (t)g(t − x)e −itω dt|
= 1
√ 2π |hf, M ω T x gi|
≤ 1
√ 2π kf k 2 kM ω T x gk 2
≤ 1
√ 2π kf k 2 kgk 2 = 1
√ 2π .
Thelastequalityresults fromtheassumption thatbothfun tionsarenormalized. Thus,
itfollows immediatelythat
1 − ε ≤ Z Z
U |V g f (x, ω)| 2 dx dω ≤ Z Z
U
1
2π dx dω = 1 2π |U|
whi h anbe rewrittenas
|U| ≥ 2π(1 − ε)
.Lieb'sinequalities intheir adaptedversionaregiven by
Theorem 4.5 (Lieb - adapted). Let
f, g ∈ L 2 (R)
, thenkV g f k p p =
Z Z
R 2 |V g f (x, ω)| p dx dω
≤ 2π 1 p−2 p 2
p
(kf k 2 kgk 2 ) p
if2 ≤ p < ∞,
≥ 2π 1 p−2 p 2
p
(kf k 2 kgk 2 ) p
if1 ≤ p ≤ 2.
(18)
kV g G f (x, ω)k p p = Z Z
R 2 |V g G f (x, ω)| p dx dω
= Z Z
R 2
1
√ 2π V g G f (x, ω 2π )
p (2π) p 2 1
2π dx dω
= (2π) p−2 2 Z Z
R 2 |V g C f (x, ω)| p dx dω
= (2π) p−2 2 kV g C f (x, ω)k p p
and theestimation
V g G f (x, ω) p
p ≤ 2 p
kf k 2 kgk 2
p
for theSTFTinGrö henig'sdenition, the
p−
norm oftheSTFT anbe estimatedasV g C f (x, ω)
p
p = 1 2π
p−2 p
V g G f (x, ω) p
p ≤ 1 2π
p−2 p 2 p
kf k 2 kgk 2
p
.
Theseinequalities allow tostate the adapted strongun ertainty prin iple:
Theorem4.6(Thestrongun ertaintyprin iple-adapted). Supposethat
kf k 2 = kgk 2 = 1
. IfU ⊂ R 2
and0 ≤ ε ≤ 1
are su h thatZ Z
U |V g f (x, ω)| 2 dx dω ≥ 1 − ε,
then
|U| ≥ 2π(1 − ε) p−2 p p 2
p−2 2
for all p > 2.
(19)In parti ular,
|U| ≥ sup
p>2 2π(1 − ε) p−2 p p 2
p−2 2
≥ 4π(1 − ε) 2 .
(20)Proof. This proof is the adaptation of the proof given by Grö henig in [Grö01 ℄. From
theassumtionit followsthat:
1 − ε ≤ Z Z
U |V g f (x, ω)| 2 dx dω
= Z Z
R 2
χ U (x, ω)|V g f (x, ω)| 2 dx dω
= kχ U (x, ω)|V g f (x, ω)| 2 k 1 .
(21)Now, Hölder's inequality 2.1 an be applied to (21 ). The hoi e of parameters is
a = p 2
and
b = p−2 p
. These parameters fullltherequirements:1 a + 1
b = 2
p + p − 2
p = 1.
Moreover, ithasto beshown that
|V g f (x, ω)| 2 ∈ L 2 p (R 2 )
andχ U (x, ω) ∈ L p−2 p (R 2 )
:Z Z
R 2 |V g f (x, ω)| 2· p 2 dx dω 2/p
= kV g f (x, ω)k 2 p .
Hen e it follows from (18 ) that
|V g f (x, ω)| 2 ∈ L p 2 (R 2 )
. The fun tionχ U (x, ω)
belongsto
L p−2 p
aslongasU
hasnitemeasure. For anyU
with innitemeasuretheinequality(19 )is trivial.
In the ase of
|U| < ∞
applying Hölder's inequality to (21) and then Lieb's inequality(4.5 ) to (22)gives:
1 − ε ≤ kχ U (x, ω)|V g f (x, ω)| 2 k 1
≤ kχ U (x, ω)k p/(p−2) k|V g f (x, ω)| 2 k p/2
= Z Z
R 2 |χ U (x, ω)| p−2 p p−2 p Z Z
R 2 |V g f (x, ω)| 2· p 2 dx dω 2 p
= |U| p−2 p Z Z
R 2 |V g f (x, ω)| p dx dω 2 p
(22)
≤ |U| p−2 p 1 2π
p−2 2 2 p
kf k 2 kgk 2 p 2 p
= 1
2π |U| p−2 p 2 p
2 p
.
(23)For
p > 2
the inequality(23) an be reformulated as:|U| ≥ 2π(1 − ε) p−2 p p 2
p−2 2
and inparti ular for
p = 4
|U| ≥ 4π(1 − ε) 2
whi h proves the theorem.
4.2 Non-optimality of the strong un ertainty prin iple
The role of Gaussian fun tions is oftena spe ial one. They areoptimal withregard to
Heisenberg'sun ertaintyprin iple inthesensethattheyaretheonlyfun tionsforwhi h
equalityin(11) holds. The question ofinterest is now whetheror not this optimality is
preserved for Grö henig's un ertainty prin iples. This se tion is devoted to theanswer
ofthat question.
4.2.1 Gaussian fun tions
Consider theGaussian fun tion
f : R → R
,x 7→ 2 π
1/4
c −1/2 e −((x−b)/c) 2
and thewindowg : R → R
,x 7→ π 2 1/4
d −1/2 e −((x−b)/d) 2
with onstantsb, c, d ∈ R
wherec
andd
an beassumedpositive. The fun tionsarenormalized sothat
kf k 2 = kgk 2 = 1
. First,thesetU ⊂ R 2
whereRR
U |V g f (x, ω)| 2 dx dω ≥ 1 − ε
holdsforsomeε ≥ 0
hastobedetermined:I :=
Z Z
U |V g f (x, ω)| 2 dx dω
= 1 π 2 (cd) −1
Z Z
U
Z
R
e − (t−b)/c
2
e − (t−b−x)/d 2
e −itω dt
2 dx dω.
The hange of variables
t − b = cds
,dt = cd ds
yields:I = 1 π 2 cd
Z Z
U
Z
R
e −(ds) 2 e −(cs−x/d) 2 e −iω(cds+b) ds
2 dx dω
= 1 π 2 cd
Z Z
U
Z
R
e − c 2 +d 2
s 2 +(cds)(iω−2x/d 2 )/(c 2 +d 2 )
− x/d 2 ds
2 dx dω.
Completing the squareintheexponent
I = 1 π 2 cd
Z Z
U
e c
2 d 2 /(4(c 2 +d 2 ))
iω−2x/d 2 2
− x/d 2
2 × . . .
. . . × Z
R
e − c 2 +d 2
s+(cd)(iω−2x/d 2 )/(2(c 2 +d 2 )) 2
ds
2 dx dω
and the hange of variables
s = s + (cd)(iω − 2x/d ˜ 2 )/(2(c 2 + d 2 ))
gives:I = 1 π 2 cd
Z Z
U
e c
2 d 2 /(4(c 2 +d 2 ))
iω−2x/d 2 2
− x/d 2
2
Z
R
e − (c 2 +d 2 ) 1/2 ˜ s
2
d˜ s
2 dx dω.
Using
Re[ c 2 d 2
2(c 2 + d 2 ) iω − 2x d 2
2
− 2 x d
2
]
= − c 2 d 2 ω 2
2(c 2 + d 2 ) + 2c 2 x 2
d 2 (c 2 + d 2 ) − 2c 2 x 2
d 2 (c 2 + d 2 ) − 2x 2 c 2 + d 2
= − c 2 d 2 ω 2
2(c 2 + d 2 ) − 2x 2 c 2 + d 2
theintegral simplies to
I = 1 π 2 cd
Z Z
U
e − (cdω)/ √
2(c 2 +d 2 ) 2
− ( √ 2x)/( √
c 2 +d 2 ) 2 √
√ π c 2 + d 2
2
dx dω.
A hange of variables shows that hoosing
U ′
as a ir le (U
then be omes an ellipse)allowsanalyti integration of theGaussian fun tion. Let
u = cdω
p 2(c 2 + d 2 ) , v =
√ 2x
√ c 2 + d 2
withthe orrespondingdeterminant of theJa obian
∂(x, ω)
∂(u, v) =
c 2 + d 2
cd .
I = 1 π
Z Z
U ′
e −(u 2 +v 2 ) du dv = 1 π
Z R 0
Z 2π 0
e −r 2 r dϕ dr = 1 − e −R 2
(24)where
U ′
isthenewdomainofintegration. AsU ′
isa ir leitsmeasure anbe omputedas
|U ′ | = πR 2
.The measures of the original domain of integration
U
andU ′
are related through thedeterminant ofthe Ja obian:
|U| = c 2 + d 2
cd |U ′ | = c 2 + d 2 cd πR 2 .
Choosing
ε = e −R 2
the onditionsoftheorem4.6arefullledandtheSTFTisessentially bounded onU
. Thestrong un ertaintyprin iple (inequality(19)) thus states:c 2 + d 2
cd πR 2 ≥ 2π(1 − e −R 2 ) p−2 p p 2
2
p−2 .
(25)To provethatGaussian fun tionsarenotoptimal inthestrong un ertaintyprin iple for
theSTFT it hasto be shownthat inequality (25 )is a stri t one. First,it an be noted
that
min
c,d
c 2 + d 2
2cd = 1
sothatitis su ient to onsider theinequalityR 2 ≥ (1 − e −R 2 ) p−2 p p
2
2
p−2 .
(26)Denethefun tions
l(R) = R 2 and r(R) = (1 − e −R 2 ) p−2 p p 2
2
p−2 .
Then
l ′ (R) = 2R and r ′ (R) = p 2
2
p−2 p
p − 2 (1 − e −R 2 ) p−2 2 (2Re −R 2 ).
Sin e
l(0) = r(0) = 0
showing thatl ′ (R) ≥ r ′ (R)
forR > 0
is su ient to prove that(26 )is astri t inequality. Toshow that
l ′ (R) ≥ r ′ (R) ⇔ 1 ≥ p 2
2
p−2 p
p − 2 1 − e −R 2 2
p−2 e −R 2
(27)holds for allp onsider thederivative oftheright-handside. Lettheright-handside be
r 2 (R)
. Thenr 2 ′ (R) = p 2
p−2 2 p
p − 2 2Re −R 2 1 − e −R 2 4−p p−2 h p
p − 2 e −R 2 − 1 i .
Hen e,
r 2 (R)
is maximal forp
p−2 e −R 2 − 1
= 0
, equivalentlye −R 2 M = p−2 p
whereR M
denotesthe
x
-valuewherethemaximumistakenon. Sin er 2 (R M ) = 1
andfurthermorer 2 (0) = lim R→∞ r 2 (R) = 0
, inequality (27 )is satised for allR
and equality isattainedonly for
e −R 2 = p−2 p
. From this it an be on luded that inequality (25) is a stri t onefor all
R > 0
. The bound given bythestrong un ertainty prin iple istherefore not the best possible for Gaussian fun tions.Moreover, it an be observed that
lim R→∞ r(R) = p 2 p−2 2
whereas
lim R→∞ l(R) = ∞
.Thismeansthatthelowerboundisagoodestimationwhen onsideringsmallper entages
of the total energy of the STFT (
ε
lose to1
). However, it looses its meaning as anestimation for the area that overs high per entages of the total energy (
ε
small). Forsmall
ε
the STFT is mu h less on entrated than the bound inthe strong un ertainty prin iple suggests. Figure2 illustratesthis fa tfor the aseofp = 4
.0 0.5 1 1.5 2 2.5
0 1 2 3 4 5 6 7
R
l(R)=R 2
r(R)=2(1−e −R
2
) 2
Figure2: Qualityof thelowerbound for
p = 4
To investigatethe inuen e ofthedilationfa tors
c
andd
intheoriginalinequality(25)itis usefulto rewritethe inequality
c 2 + d 2
2cd ≥ 1 ⇔ (c − d) 2 ≥ 0.
This expression allows the observation that merely the dieren e of the dilation fa -
tors matters. This orrelates with the expe tations. The STFT of a rather spread out
Gaussian fun tion with respe t to a narrow window (thus,
c
large andd
small) an beinterpreted as the Fourier transform of a fun tion with small essential support. As a
onsequen ethe essentialsupportofthe STFThasto be large. Butfor largedomainsof
integration the lowerboundis not agoodestimation.
The reversed ase (
d
large andc
small) orrespondsto the STFT of a narrow fun tion with respe t to a broad window. Su h a window fun tion does not alter the fun tionsigni antly. Thus, like in the rst ase, the STFT an be interpreted as the Fourier
transform of a fun tion withsmall essential support. However, this is not a reasonable
hoi eofthe pair
(f, g)
asg
doesnotemphasizethefun tionf
onsubsetsofthedomain.Chirp fun tions are fun tions where the frequen y in reases or de reases with time. A
spe ial ase onsidered in this se tion are Gaussian fun tions with an additional imag-
inary term in the exponent. This term introdu es os illations whose amplitude is en-
velopedbytheGaussianfun tionwithpurelyrealexponent(seegure3). Thefrequen y
ofthe os illationsin reases for
|x| → ∞
.−2 −1.5 −1 −0.5 0 0.5 1 1.5 2
−1
−0.8
−0.6
−0.4
−0.2 0 0.2 0.4 0.6 0.8 1
x
Gaussian
Chirp
Figure 3:AGaussian fun tion ompared withthereal partof a hirpfun tion
The obje tive of this se tion is to study the role of hirp fun tions inthe strong un er-
taintyprin ipleintrodu edinse tion3. Theinuen eofthefrequen y
ϕ(x)
onthevalueofthe estimationisto be investigated.
Letthe fun tion be
f : R → C, x 7→ ( π 2 ) 1 4 e −x 2 −iαx 2
and the windowbeg : R → R, x 7→
( π 2 ) 1 4 e −x 2
. Itissu ient to onsiderGaussian fun tions entered atx = 0
be auseshiftsalongthe
x
-axisdon'thaveanyinuen eonthe result,asseeninsubse tion4.2.1. Sin ethe inequality (25) was sharpest for the dilation fa tors
c = d = 1
, these onstants areusedinthe following omputations, aswell.
I :=
Z Z
U |V g f (x, ω)| 2 dx dω
= 1
π 2 Z Z
U
Z
R
e −t 2 −iαt 2 e −(t−x) 2 e −itω dt
2 dx dω
= 1
π 2 Z Z
U
e −2x 2 Z
R
e −t 2 (2+iα)−t(iω−2x) dt
2 dx dω
= 1
π 2 Z Z
U
e −2x 2
e (iω−2x) 2 /(2(2+iα))
Z
R
e − 2+iα
t+(iω−2x)/(2(2+iα)) 2
dt
2 dx dω.
Using the hange ofvariable
˜ t = t + (iω − 2x)/(2(2 + iα))
and thefa tthatRe( 1
2(2 + iα) (iω − 2x) 2 ) = 1
4 + α 2 (−ω 2 − 2αxω + 4x 2 )
theintegral transformsinto
I = 1 π 2
Z Z
U
e −1/(4+α 2 )(2x 2 (2+α 2 )+ω 2 +2αxω) Z
R
e −(2+iα)˜t 2 d˜ t
2 dx dω.
Dening
A := 2(2 + α 2 )
4 + α 2 , B := 1
4 + α 2 , C := 2α 4 + α 2
theintegral an bewritten as
I = 1 π 2
Z Z
U
e −(Ax 2 +Bω 2 +Cxω) π
(4 + α 2 ) 1/2 dx dω.
Changingvariables to
u = √
Ax + C 2 √
A ω, v = B − C 2 4A
1 2 ω
withthe orrespondingdeterminant of theJa obian
∂(x, ω)
∂(u, v) = (AB − C 2
4 ) − 1 2 = (4 + α 2 ) 1 2
allowsto solve theintegral
I = 1
π(4 + α 2 ) 1/2 Z Z
U ′
e −(u 2 +v 2 ) (4 + α 2 ) 1 2 du dv = 1 π
Z R 0
Z 2π 0
e −r 2 r dϕ dr = 1 − e −R 2 .
Themeasure ofthe domain of integration is
|U| = ∂(x, ω)
∂(u, v) |U ′ | = (4 + α 2 ) 1 2 πR 2 .
(4 + α 2 ) 1 2
2 R 2 ≥ (1 − e −R 2 ) p−2 p p 2
2
p−2
(28)
holds.
For
α = 0
this redu es to the ase studied in se tion 4.2.1 . Sin e(4+α 2 2 ) 1/2 ≥ 1
forall
α
, inequality (28 ) is always a stri t inequality. Moreover, it an be noted that theestimation for thesize ofintegration be omesweaker for largervalues of
α
. The STFToffun tionswithstrongos illationsisthusmu hless on entrated thanthelowerbound
suggests. It is an analogous observation to the one made in the se tion 4.2.1 for the
STFT offun tionswithlarge dilation fa tors.
The importan e of Cohen's lass of joint time-frequen y distributions naturally implies
thestudy of un ertainty prin iplesfor su hrepresentations. The entral element of this
hapteris aninequalityfor the Wignerdistribution thatisdue to A.J.E.M. Janssen.
5.1 Janssen's un ertainty prin iple
Theresultsfound byJanssenand Carypiswill rstbepresentedintheir originalversion
using the alternative denition of the Wigner distribution and the Fourier transform.
Let
W f J (x, ω)
denote the denition a ording to (10 ) andW f (x, ω)
the version in (9).The alternative denition of the Fourier transform is denoted by
f ˆ J (ω)
. Janssenstatesin[Jan89 ℄
Theorem 5.1(Janssen). Let
M
be a positive denite2 × 2
matrix anddenoteits deter-minant by
|M|
. ThenZ Z
R 2
z T M zW f J (z) dz ≥ |M| 1/2
2π kf k 2 2
(29)where
z = (x, ω) T
.Carypis shows in [Car14 ℄ that for
M = I
equality is attained ifand only iff
is of theform
f (x) = Ce −πx 2
,C ∈ C
. This is a simple onsequen e of the original version of Heisenberg'sun ertaintyprin iple forf
andf ˆ
asshowninthefollowing.Using theelementary inequality
a 2 + b 2 ≥ 2ab
witha = Z
R
x 2 |f (x)| 2 dx 1/2
and b = Z
R
ω 2 | ˆ f J (ω)| 2 dω 1/2
Heisenberg'sun ertaintyprin iple
Z
R
x 2 |f (x)| 2 dx 1/2 Z
R
ω 2 | ˆ f J (ω)| 2 dω 1/2
≥ 1 4π kf k 2 2
an be rewritten as
Z
R
x 2 |f (x)| 2 dx + Z
R
ω 2 | ˆ f J (ω)| 2 dω ≥ 1 2π kf k 2 2 .
Equalityisonlyattainedif
a = b
. ThisonlyholdsforthoseGaussiansfun tionsthatareinvariant underFourier transformation,namelythe fun tions
f (x) = Ce −πx 2
,C ∈ C
.Inserting the marginal onditions of the Wigner distribution gives Janssen's result for
the ase
M = I
:Z Z
R 2
(x 2 + ω 2 )W f J (x, ω) dx dω ≥ 1 2π kf k 2 2 .
Equality isattainedinthe same ases asbefore.
To transform the statement to our denition of theWigner distribution we observe the
relation
W f J ( x
√ 2π , ω
√ 2π ) = 1
√ 2π Z
R
f 1
√ 2π (x + y 2 )
f 1
√ 2π (x − y 2 )
e −iωy dy
= √
2πW f 1
√ 2π
(x, ω)
(30)with
f √ 1 2π
(x) = f ( √ 1
2π x)
. A hange ofvariables in(29)givesZ Z
R 2
√ x 2π , ω
√ 2π
M x
√ 2π , ω
√ 2π
T
W f J ( x
√ 2π , ω
√ 2π ) dx dω ≥ |M| 1 2 kf k 2 2 .
Using relation(30) this an berewritten as
Z Z
R 2
(x, ω)M (x, ω) T W f 1
√ 2π
(x, ω) dx dω ≥ √
2π|M| 1 2 kf k 2 2 .
With
kf √ 1 2π k 2 2 = Z
R |f ( x
√ 2π )| 2 dx = Z
R |f (x)| 2 √
2π dx = √ 2πkf k 2 2
and thesubstitution
g(x) = f √ 1 2π
(x)
Janssen's un ertaintyprin iple inthetransformed versionreadsZ Z
R 2
(x, ω)M (x, ω) T W g (x, ω) dx dω ≥ |M| 1 2 kgk 2 2
(31)withequalityfor
M = I
ifandonly iff (x) = Ce −πx 2
or equivalentlyg(x) = f ( √ 1 2π x) = Ce − x2 2
.5.2 Equality in Janssen's prin iple
In the following the inequality will be onsidered for a general, positive denite
2 × 2
-matrix. The question to be answered is whi h fun tions an attain thelower bound in
themore general setting. Itwill beshown that hirpfun tions an yieldequalityunlike
intheun ertaintyprin iples fortheSTFT.
Janssenstatesin[Jan89℄thatWignerdistributions oflinearlytransformedfun tionsare
againWigner distributions:
Theorem5.2 (Janssen). Forany lineartransformation
z ∈ R 2 7→ Az ∈ R 2
with|A| = 1
of the phase plane there is a unitary operator
T
ofL 2 (R)
su h thatW f (Az) = W T f (z)
,f ∈ L 2 (R)
.The proof requiresWeyl al ulus of pseudodierential operators and group theory and
isbeyond the s ope ofthis report.
DeBruijn derives in[DB73℄the relationbetween thetransformation inthe phaseplane
and theunitaryoperator
T
.Theorem 5.3 (De Bruijn). Consider the transformation
(x, ω) 7→ (x ′ , ω ′ ) = (αx + βω, γx + δω)
withαδ − βγ = 1
. ThenW T f J (x ′ , ω ′ ) = W f J (x, ω)
(32)where
T f (x) = α − 1 2 f x α
exp iπγx 2 α
(33)
for
β = 0
andT f (x) = (iβ) − 1 2 Z
R exp − π
iβ (δx 2 − 2xt + αt 2 )
f (t) dt
(34)for
β 6= 0
.Thisrelation holdsinthe aseofJanssen'sdenitionoftheWignerdistribution. Forthe
fun tion
g
in(31 )we have to usetherelationg(x) = f ( √ 1
2π x)
andequation(30 ). LetD
be the s aling operator
Df (x) = f ( √ x
2π ) = g(x)
andD −1
its inverse. Applying (30)toboth sides ofequation(32) gives
√ 2πW DT f ( √
2πx ′ , √
2πω ′ ) = √
2πW Df ( √ 2πx, √
sπω)
or equivalently
W DT f (x ′ , ω ′ ) = W Df (x, ω).
With
f = D −1 g
andDf = g
we an on ludethattherelationbetweentheunitaryoper-ator
T
inDeBruijn's theoremwiththe alternative denitionof theWigner distribution and the operatorT ′
forour denition ofthedistribution readsT ′ = DT D −1
.The operator
T ′
an thus be obtained from the operatorT
dened intheorem 5.3. Forthe ase
β = 0
we getT ′ g(x) = (DT D −1 g)(x) = T (D −1 g)( x
√ 2π ) = α − 1 2 (D −1 g)( x α √
2π ) exp( iγx 2 2α )
= α − 1 2 g( x
α ) exp( iγx 2 2α ).
For
β 6= 0
equation(34 )givesT ′ g(x) = (DT D −1 g)(x) = T (D −1 g)( x
√ 2π )
= (iβ) − 1 2 Z
R
exp( πi β (δ x 2
2π − 2 xt
√ 2π + αt 2 ))(D −1 g)(t) dt.
With
(D −1 g)(t) = g( √
2πt)
and the hange of variablesu = √
2πt
the operatorT ′
forβ 6= 0
is givenbyT ′ g(x) = (2πiβ) − 1 2 Z
R
exp( i
2β (δx 2 − 2xu + αu 2 ))g(u) du.
To summarizede Bruijn's theorem an berestated as: