ebook img

Towards photostatistics from photon-number - The Adami Lab PDF

12 Pages·2004·0.13 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Towards photostatistics from photon-number - The Adami Lab

journal of modern optics, 15 june–10 july 2004 vol. 51, no. 9–10, 1517–1528 Towards photostatistics from photon-number discriminating detectors HWANG LEEy, ULVI YURTSEVERy, PIETER KOKyz, GEORGE M. HOCKNEYy, CHRISTOPH ADAMIy, SAMUEL L. BRAUNSTEIN} and JONATHAN P. DOWLINGy yQuantum Computing Technologies Group, Section 367, Jet Propulsion Laboratory, California Institute of Technology, MS 126-347, 4800 Oak Grove Drive, CA 91109-8099, USA zHewlett Packard Laboratories, Bristol BS34 8QZ, UK }Computer Science, University of York, York YO10 5DD, UK (Received27October 2003) Abstract. We study the properties of a photodetector that has a number- resolving capability. In the absence of dark counts, due to its finite quantum efficiency,photodetectionwithsuchadetectorcanonlyeliminatethepossibility thattheincidentfieldcorrespondstoanumberofphotonslessthanthedetected photon number. We show that such a non-photon number-discriminating detector, however, provides a useful tool in the reconstruction of the photon number distribution of theincident fieldeven inthepresence ofdark counts. 1. Introduction With the recent advent of linear optical quantum computation (LOQC), interest in photon number-resolving detectors has been growing. The ability to discriminate the number of incoming photons plays an essential role in the realizationofnonlinearquantumgatesinLOQC[1–4]aswellasinquantumstate preparation[5–9].Suchdetectorsareusedtopost-selectparticularquantumstates of a superposition and consequently produce the desired nonlinear interactions of LOQC.Inaddition,themostprobingattacksaneavesdroppercanlaunchagainsta typicalquantumcryptographysystemexploitphoton-numberresolvingcapability [10, 11]. Recent efforts in the development of such photon number-resolving detectors include the visible light photon counter [12, 13], fibre-loop detectors [14–18], and superconducting transition edge sensors [19]. Standard photodetec- tors can measure only the presence or absence of light (single-photon sensitivity), andgenerallydonothavethecapabilityofdiscriminatingthenumberofincoming photons (single-photon number resolution). There have been suggestions for accomplishingsingle-photonresolutionusingmanysingle-photonsensitivedetec- tors arranged in a detector array or detector cascade [20–23]. For example, the VLPC (visible light photon counter) is based on a confined avalanche breakdown in a small portion of the total detection area and, hence, can be modelled as a detector cascade [24]. Then again, fibre-loop detectors may be regarded as a detector cascade in the time domain. JournalofModernOpticsISSN0950–0340print/ISSN1362–3044online#2004Taylor&FrancisLtd http://www.tandf.co.uk/journals DOI:10.1080/09500340410001670901 1518 H. Lee et al. Let us suppose a number-resolving detector detects two photons in a given time interval. If the quantum efficiency of the detector is one, we can be certain that two photons came from the incident light during that time interval. If the quantum efficiency is, say, 0.2, what can we say about the incident light? In the absenceofdarkcounts,wecansafelysayonlythattherewereatleasttwophotons intheincidentbeam.Inthiscase,photodetectionratherhasanon-photonnumber discriminating feature, since the conditional probability of not having zero, or one photonintheincidentpulseiszero.Canwesaymorethanthat?Forexample,what is the probability that the incident pulse actually corresponds to two photons, or three?Inthispaperweattempttoanswerthesequestionsanddiscusstheeffectof quantum efficiency on the photon-number resolving capability. The semiclassical treatment of photon counting statistics was first derived by Mandel [25]. He showed that the photon counting distribution is Poissonian if the intensity of the incident light beam is constant in time. The full quantum mechanical description has been formulated by Kelley and Kleiner [26], and Glauber [27]. Here, we start from a formula based on the photon-number state representation, which was first derived by Scully and Lamb [28]. For a given quantum efficiency of the photodetector, the photon counting distribution can be written as X1 (cid:1)n(cid:2) PðkÞ¼ (cid:1)kð1(cid:1)(cid:1)Þn(cid:1)kSðnÞ; ð1Þ k n¼k where P(k) is the probability of detecting k photons, (cid:1) is the quantum efficiency and S(n) is the probability that the source (incident light) corresponded to n photons. It is assumed that the quantum efficiency (cid:1) does not depend on the intensityoflight.ThisdistributionthenleadstoanaturaldefinitionofP(k|n),i.e. as the conditional probability of detecting k photons given n photons being in the source, as (cid:1)n(cid:2) PðkjnÞ¼ (cid:1)kð1(cid:1)(cid:1)Þn(cid:1)k: ð2Þ k Contrariwise, suppose we detect k photons. If the detector has a quantum efficiency much less than unity, very little can be said about the actual photon number associated with the source. We may, however, deduce that the source actually corresponded to n photons with a certain probability. This probability should be high if the quantum efficiency is high. In the following, we will use Bayes’s theorem in order to extract information about the source from the conditional photon-counting statistics. Armed with the conditional probability P(k|n), we may write PðkjnÞSðnÞ PðkjnÞSðnÞ QðnjkÞ¼ ¼ : ð3Þ P PðkÞ PðkjiÞSðiÞ i Now we will identify the conditional probability, Q(n|k), with the probability that an ideal detector would have detected n photons, given that k photons are detected by the imperfect detector. It is given as the product of the probability of thesourcecorrespondingtonphotonsbeforethemeasurementandtheprobability of k-photon detection, given that n photons are in the source, normalized to the probability of having k photons detected. Towards photostatistics from photon-number discriminating detectors 1519 perfect detector mode 1 mode 2 imperfect detector Figure 1. Schematic ofthe heralded-photon set-up. It is evident that generally we cannot find Q(n|k) if we do not know S(n)—the distribution of the source. The exceptions are: (i) if n<k, then PðkjnÞ¼0. Therefore, QðnjkÞ¼0 for n<k no matter what S(n) is. (ii) If the quantum efficiency is unity, that is (cid:1)¼1, then PðkjnÞ¼0 when n6¼k, and Pðkjn¼kÞ¼1. Since PðkÞ¼Sðn¼kÞ, it leads to Qðn¼kjkÞ¼1 and Qðn6¼kjkÞ¼0. Hence, without knowing S(n), we only have information about Q(n|k) when n<k or (cid:1)¼1. Such a detector, then, readily excludes certain numbers of photons in the source rather than determining them. Without a prior knowledge of the source, if (cid:1)6¼1, we call this a non-photon number-discriminating detector. Obviously,thedifficultyintalkingaboutthisconditionalprobabilityliesinthe fact that the photodetection is destructive. We will elaborate this subtlety using aheralded-photonset-updepictedinfigure1.TheunitaryoperatorUU^ transforms H the input state in the following way: 1 1 X X UU^ (cid:2) jni¼ (cid:2) jni jni : ð4Þ H n n 1 2 n¼0 n¼0 Suppose now the photon number in mode 1 is measured by the perfect detector andthephotonnumberinmode2ismeasuredbytheimperfectdetector.Itisnow simple to understand the conditional probability of finding n photons in mode 1, given that k photons are detected in mode 2. In this way, we can have an operational definition of the conditional probability Q(n|k). It is easy to show that the conditional probability of finding n photons in mode 1 given that k photons are detected in mode 2 is given by precisely the same expression for the quantity Qðn;kÞ as equation (3). (Note that the transformation given by UU^ is far H fromquantumcloning,sincetheresultofUisfarfrombeingequaltothecloning transformation ðP1 (cid:2) jni Þ(cid:2)ðP1 (cid:2) jni Þ.) n¼0 n 1 n¼0 n 2 2. POVM description Suppose an initial quantum state j i of incident light is represented in the in photon number states as 1 X j i¼ (cid:2) jni: ð5Þ in n n¼0 1520 H. Lee et al. The probability of finding n photons is then given by SðnÞ¼jhnj ij2 ¼j(cid:2) j2. n We may describe the detection process of the perfect detector by entangling the source with the detector ancilla j0i X j i¼UU^j ij0i¼ (cid:2) jni ja i ; ð6Þ f in n Q n M n where X 1 UU^ ¼ jjihjj(cid:2) ðaa^yÞj ð7Þ ðj!Þ1=2 j is the unitary operator implementing the von Neumann measurement, and we identified the quantum (Q) and measurement device (M) Hilbert spaces with appropriate subscripts. The probability to detect k photons, P(k), can now be calculated as PðkÞ¼Tr ðja i ha jj ih jÞ¼j(cid:2) j2; ð8Þ Q k M k f fi k as expected. Of course, the same result would have been obtained as PðkÞ¼TrðM (cid:3) Þ using a positive operator valued operator measure (POVM) k in M ¼jkihkj: ð9Þ k The present unitary treatment, however, is more transparent and allows a straightforward extension to imperfect detectors. For an imperfect photodetector with a finite quantum efficiency, the POVM can be written as 1 X M ¼ PðkjnÞja iha j; ð10Þ k n n n¼k with P(k|n) given by equation (2). We can verify this by calculating as before j i¼Uj ij0i j0i ; ð11Þ f in M E except that the entanglement operator U now acts on the joint Hilbert space of the quantum system, a measurement device and an environment j0i , which is E where our lost photons go. Thus, the unitary operator can be given by UU^ ¼Xjjihjj(cid:2) 1 (cid:1)(cid:1)1=2aa^y(cid:2)1þexpði’Þð1(cid:1)(cid:1)Þ1=21(cid:2)bb^y(cid:2)j; ðj!Þ1=2 j where’isanarbitraryphaseandby isthecreationoperatorfortherelevantmode of the environment. Similar to equation (8), we now have (cid:3) (cid:4) PðkÞ¼Tr ja i ha jj ih j QE k M k fi fi X1 (cid:1)n(cid:2) ¼ j(cid:2) j2 (cid:1)kð1(cid:1)(cid:1)Þn(cid:1)k: ð12Þ n k n¼k supporting our equation (10) as the correct POVM. Note that any relative phase information between the photon number states disappeared. ThePOVMeffectivelytransferstheinitialquantumstateoflightintoacertain quantum state (as a result of k-photon detection) by the transformation (cid:3) (cid:3)M (cid:3) ; ð13Þ f(cid:1)k k in Towards photostatistics from photon-number discriminating detectors 1521 where (cid:3) ¼j ih j. The probability that k photons are detected can then also in in in be written as (cid:3) (cid:4) PðkÞ¼Tr (cid:3) f(cid:1)k ¼TrðM j ih jÞ k in in X ¼ PðkjnÞSðnÞ: ð14Þ n On the other hand, the joint probability that the source state is jni and k photons are detected can be expressed by the probability Pðn;kÞ¼Tr ðhn;a j ih jn;a iÞ E k f f k ¼hnj(cid:3) jni¼PðkjnÞSðnÞ; ð15Þ f(cid:1)k which defines the quantity Q(n|k) as ha j(cid:3) ja i QðnjkÞ(cid:3) n f(cid:1)k n : ð16Þ (cid:3) (cid:4) Tr (cid:3) f(cid:1)k In particular, the quantity Qðn¼kjkÞ corresponds to the confidence of the state preparation in a situation that we have more than two modes for the input and the photon number in one of the modes is measured [22]. 3. An example Letustakeasimpleexampleofaninputcoherentstateinordertoseehowthe valueofQ(n|k)behaves.Indoingthis,weassumethatthedarkcountrateissmall enough that we can ignore it, and the quantum efficiency is independent of the number of incident photons. A coherent state j(cid:2)i is expressed in terms of number state as j(cid:2)i¼expð(cid:1)j(cid:2)j2=2ÞP ð(cid:2)n=ðn!Þ1=2Þjni [29]. The probability of finding n photons in n j(cid:2)i is then given by a Poisson distribution: expð(cid:1)nn(cid:1)Þnn(cid:1)n SðnÞ¼jhnj(cid:2)ij2 ¼ ; ð17Þ n! where nn(cid:1) ¼j(cid:2)j2. The probability of detection of k photons with an imperfect detector is well known to be [28] X1 (cid:1)n(cid:2) expð(cid:1)nn(cid:1)Þnn(cid:1)n PðkÞ¼ (cid:1)kð1(cid:1)(cid:1)Þn(cid:1)k k n! n¼k ðnn(cid:1)(cid:1)Þk ¼ expð(cid:1)nn(cid:1)(cid:1)Þ: ð18Þ k! That is, the photon counting statistics are simply a Poisson distribution where the average number of photons detected is the average number of incident photons multiplied by the quantum efficiency of the detector—given by kk(cid:1) ¼nn(cid:1)(cid:1) as expected. 1522 H. Lee et al. 1 Q(n|1) 0.8 0.6 n=1 0.4 (a) (b) 0.2 2 3 4 n Figure 2. The conditional probability Q(n|k¼1) for a coherent state with nn(cid:1) ¼1 as a functionof n.(a) Fordetector efficiency (cid:1)¼0.2,Qð1j1Þ¼0:45,Qð2j1Þ¼0:36.(b) For detector efficiency(cid:1)¼0.9, Qð1j1Þ¼0:90,Qð2j1Þ¼0:09. NowletusconsidertheconditionalprobabilityQ(n|k).Usingequation(3),itis simply found as (cid:1)n(cid:2) (cid:1)kð1(cid:1)(cid:1)Þn(cid:1)k expð(cid:1)nn(cid:1)Þnn(cid:1)n=n! k QðnjkÞ¼ expð(cid:1)nn(cid:1)(cid:1)Þðnn(cid:1)(cid:1)Þk=k! 1 ¼ ð1(cid:1)(cid:1)Þn(cid:1)k exp½(cid:1)nn(cid:1)ð1(cid:1)(cid:1)Þ(cid:4)nn(cid:1)n(cid:1)k ðn(cid:1)kÞ! ¼expð(cid:1)(cid:1)llÞ(cid:1)lll; ð19Þ l! where l¼n(cid:1)k and (cid:1)ll¼nn(cid:1)ð1(cid:1)(cid:1)Þ. Therefore, the form of Q(n|k) is again a Poisson distribution as a function of n(cid:1)k with the average number of photons that fail to register given as nn(cid:1)ð1(cid:1)(cid:1)Þ. Figure 2 shows the behaviour of Q(n|k) as a function of n, for the case of coherent state input with nn(cid:1) ¼1, when one photon is detected. We can see that if thequantumefficiencyofthedetectoris0.2,withthedetectionofonephoton,then the chances are more than 50% that the input contains more than two photons (Qð1j1Þ¼0:45). Obviously, Qð1j1Þ increases for higher values of the quantum efficiency (Qð1j1Þ¼0:9 for (cid:1)¼0:9). On the other hand, for a given quantum efficiency of the detector, Qð1j1Þ gets large as the average number of photons in the input decreases. For nn(cid:1) ¼0:2, Qð1j1Þ¼0:85 even if the detector efficiency is 0.2, as depicted in figure 3. In this case, with the detection of one photon the chances are 85% that the detection result corresponds to the correct value. 4. Reconstruction of photon-number distribution So far we have seen that we cannot say much about Q(n|k) if we do not know atleasttheform ofS(n).Inotherwords,using animperfectdetector,even aftera detection of k photons, we do not gain much information about the number of incident photons without a prior knowledge of the incident field. Hence, the Towards photostatistics from photon-number discriminating detectors 1523 1 Q(n|1) 0.8 0.6 n=0.2 0.4 (a) 0.2 (b) 2 3 4 n Figure 3. The conditional probability Q(n|k¼1) for a coherent state with nn(cid:1) ¼0:2 as a function of n. (a) For detector efficiency (cid:1)¼0:2, Qð1j1Þ¼0:85, Qð2j1Þ¼0:14. (b) Fordetector efficiency (cid:1)¼0:9,Qð1j1Þ¼0:98,Qð2j1Þ¼0:02. problem comes back to reconstruction of photon number distribution of the incident field, S(n), from the photon counting probability, P(k) by a sequence of identical inputs and measurements. For this purpose, we may write equation (1) in a matrix form as P¼PS; ð20Þ where 0 1 0 1 Pð0Þ Sð0Þ BPð1ÞC BSð1ÞC B C B C B C B C P¼BPð2ÞC; S¼BSð2ÞC ð21Þ B C B C B (cid:5) C B (cid:5) C @ A @ A (cid:5) (cid:5) and each component of the matrix P represents the conditional probability P(k|n) as (cid:1)n(cid:2) P (cid:3)PðkjnÞ¼ (cid:1)kð1(cid:1)(cid:1)Þn(cid:1)k: ð22Þ kn k Then the inverse of P can be found explicitly; it is given by (cid:1)m(cid:2) ðP(cid:1)1Þ ¼ (cid:1)(cid:1)mð(cid:1)(cid:1)1Þm(cid:1)n; ð23Þ nm n which,in the absence of dark counts, determines the photon-number distribution of the source from the photon-counting distribution. Note, however, that for a fixed n there are alternating signs in ðP(cid:1)1Þ as the index m changes, so that this mn inversion formula can yield unphysical results depending on the probability distribution P(k). Put another way, for small efficiencies the matrix P is ill- conditioned and P(cid:1)1 will contain matrix elements that can become very large in absolute value and can have either sign. This means that even if the calculated S(n)’s are all non-negative, a good inversion is still not guaranteed, since the inversion formula (23) is highly sensitive to small changes of P(k), amplifying the statistical and other noise which is inevitably present in any actual measurement. 1524 H. Lee et al. 0.2 (a) y t0.15 i l i b a b 0.1 o r P (b) 0.05 10 20 30 40 Number of photons Figure 4. Reconstruction of the input-state photon-number distribution: (a) photon counting distribution from the NIST experiment [19]; (b) reconstruction of the inputstatephotondistribution(nn(cid:1) (cid:7)20:15),assumingaquantumefficiencyof0.2and nodark counts. ThefactthattheinversematrixP(cid:1)1 isalsoanuppertriangularmatriximpliesthat S(n)isdeterminedbyP(k)withk(cid:6)n.Inotherwords,thetailofthemeasurement result, P(k), where the statistical noise can be expected to be greatest, affects the inversion significantly. The validity conditions for the inversion formula shall be discussed elsewhere. Instead, we can carry out the following recipe. (1) Seek an underlying distri- bution S for which PS matches the observed frequencies according to a (cid:4)2 goodness-of-fit test. (2) Then, of all such distributions, seek the one with maxi- mum entropy; i.e. pick the most likely underlying distribution, which is statisti- callyconsistentwiththedata.Wenotethatentropyisthelogarithmofthenumber of underlying configurations corresponding to a given distribution. Thus, such a strategy seeks the ‘most-likely’ distribution satisfying the constraint that, when filtered by imperfect detection, the resulting distribution is consistent within the statisticalnoiseoftheactualdata[30].Thisapproachhasbeenusedformanyyears in the imaging community for deconvolution [31]. Here we ensure a global maximumentropydistributionisfoundbyimplementingthisstrategyasagenetic algorithm, which we will discuss in a forthcoming paper [32]. Using this method, the reconstruction of S with a data set from an actual experiment is depicted in figure 4. The plot of figure 4, curve (a) shows the probability distribution of the detected photons. The raw count data was normalized by the total number of counts (data provided courtesy of NIST [19]). The result of reconstruc- tion is shown by the plot of figure 4, curve (b) and yields a nn(cid:1) ¼20:15 for the incidentcoherentstate—agoodagreementwiththedetectedexperimentalvalueof nn(cid:1) ¼4:02 [19]. exp 5. Dark counts In addition to the finite quantum efficiency, the dark counts further make the results extracted from the experiment obscure. For a constant dark-count rate, however, a generalization of the inversion may be handled analytically. Towards photostatistics from photon-number discriminating detectors 1525 Suppose in general that the probability of d dark counts is D(d), a discrete probability distribution, which satisfies the normalization condition 1 X DðdÞ¼1: ð24Þ d¼0 We assume that the distribution D(d) represents probabilities of dark counts in a fixedintervaloftime(cid:5)(durationofasinglephoton-countingexperiment),andthat these probabilities are independent of all other relevant observables such as incoming photon number, number of detected photons, etc. Therefore in the presence of dark counts, the conditional probability of detecting k photons, given n photons in the incident field equation (2), is modified accordingly: Xk (cid:1) n (cid:2) P ðkjnÞ¼ DðdÞ (cid:1)k(cid:1)dð1(cid:1)(cid:1)Þn(cid:1)kþd; ð25Þ D k(cid:1)d d¼0 which expresses the probability P ðkjnÞ as a sum (over all possible d) of the D probability of registering k(cid:1)d photons from the incident field via true detection while the remaining d are registered via dark-count events. As long as the physical time scale for the emergence of a dark-count event is much smaller than (cid:5) and the events arrive at a constant rate (cid:6), the probability distributionD(d)(ofthearrivalofddark-counteventsinthefixedtimeinterval(cid:5)) has a Poissonian-distribution limit: ð(cid:6)(cid:5)Þd DðdÞ¼expð(cid:1)(cid:6)(cid:5)Þ ; ð26Þ d! where the average number of dark count is dd(cid:1) ¼(cid:6)(cid:5). Since after interchanging the summation index d with k(cid:1)d we can rewrite equation (25) as Xk (cid:1)n(cid:2) P ðkjnÞ¼ Dðk(cid:1)dÞ (cid:1)dð1(cid:1)(cid:1)Þn(cid:1)d; ð27Þ D d d¼0 where the summation over d is in the form of a matrix multiplication (equation (22)), X ðP Þ ¼ D P ; ð28Þ D kn kd dn d or in matrix notation, P ¼DP; ð29Þ D where ðP Þ (cid:3)P ðkjnÞ, and D kn D D (cid:3)Dðk(cid:1)dÞ kd ð(cid:6)(cid:5)Þk(cid:1)d ¼expð(cid:1)(cid:6)(cid:5)Þ : ð30Þ ðk(cid:1)dÞ! Throughout the paper, we adopt the convention that k!¼(cid:2)ðkþ1Þ(cid:3)1 for (cid:1)a(cid:2) negative integer k, and correspondingly (cid:3)0 whenever b>a. In equation b (30), this convention implies that D ¼0 for d>k. kd 1526 H. Lee et al. Now by inverting equation (27) in a matrix form, from the fundamental equation, X PðkÞ¼ P ðkjnÞSðnÞ; ð31Þ D n wecanreconstructthephotonnumberdistributionS(n)oftheincidentfieldgiven thevectorofobservedphoton-countprobabilitiesP(k),whoserelationshiptoS(n) (givenbyequation(31)),canagainbewritteninmatrixformasP¼P S,justasin D equations (20) and (21). The inverse of the coefficient matrix P ðkjnÞ¼ðP Þ is D D kn found from equation (29), P(cid:1)1 ¼P(cid:1)1D(cid:1)1: ð32Þ D Let us first consider the inverse of the matrix D defined as in equation (30). It is not difficult to prove that the inverse of D is given by ð(cid:1)(cid:6)(cid:5)Þi(cid:1)j ðD(cid:1)1Þ ¼expð(cid:6)(cid:5)Þ ij ði(cid:1)jÞ! ð(cid:6)(cid:5)Þi(cid:1)j ¼expð(cid:6)(cid:5)Þð(cid:1)1Þi(cid:1)j : ð33Þ ði(cid:1)jÞ! Combining equation (33) with equation (32) and equation (23), we find X(cid:5)i(cid:6) ð(cid:1)(cid:6)(cid:5)Þi(cid:1)n ðP (cid:1)1Þ ¼expð(cid:6)(cid:5)Þ (cid:1)(cid:1)ið(cid:1)(cid:1)1Þi(cid:1)k D kn k ði(cid:1)nÞ! i ð(cid:1)1Þkþn expð(cid:6)(cid:5)ÞX1 (cid:8)(cid:5)i(cid:6)(cid:7) (cid:9)(cid:8)ð1(cid:1)(cid:1)Þ(cid:6)(cid:5)(cid:9)i ¼ ði(cid:1)nÞ! : ð34Þ ð1(cid:1)(cid:1)Þkð(cid:6)(cid:5)Þn k (cid:1) i¼0 The infinite power series over i in equation (34) is absolutely convergent for all values of the argument ð1(cid:1)(cid:1)Þ(cid:6)(cid:5) z(cid:3) ; ð35Þ (cid:1) andcanbeexpressedintermsoftheconfluenthypergeometricfunction F ða;b;zÞ 1 1 (whichisanentirefunctionoftheargumentzforallparametervaluesa,b),giving a compact expression for the inverse matrix elements in the form: expð(cid:6)(cid:5)Þ ðP (cid:1)1Þ ¼ð(cid:1)1Þkþn D kn ð1(cid:1)(cid:1)Þkð(cid:6)(cid:5)Þn 8 n! >><k!ðn(cid:1)kÞ!zn1F1ðnþ1;n(cid:1)kþ1;zÞ; if k(cid:9)n; (cid:8) ð36Þ 1 >>:ðk(cid:1)nÞ!zk1F1ðkþ1;k(cid:1)nþ1;zÞ; if k>n: The inversion of equation (32) can now be effected by substituting equations (35) and (36) into the matrix formula X SðkÞ¼ ðP (cid:1)1Þ PðnÞ: ð37Þ D kn n Thissuggeststhatthereconstructionofthephotonstatisticsofthesourcemaybe possiblewithanygivenquantumefficiencyofthedetector,eveninthepresenceof

Description:
[14–18], and superconducting transition edge sensors [19]. Standard Journal of Modern Optics ISSN 0950–0340 print/ISSN 1362–3044 online © 2004 Taylor & Francis Ltd .. a compact expression for the inverse matrix elements in the form: ًPD. ہ1قkn ¼ ًہ1قk‏n We wish to thank G. S. Ag
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.