Experimental demonstration of random walk by probability chaos using single photons

In our former work (Sci. Rep. 4: 6039, 2014), we theoretically and numerically demonstrated that chaotic oscillation can be induced in a nanoscale system consisting of quantum dots between which energy transfer occurs via optical near-field interactions. Furthermore, in addition to the nanoscale implementation of oscillators, it is intriguing that the chaotic behavior is associated with probability derived via a density matrix formalism. Indeed, in our previous work (Sci. Rep. 6: 38634, 2016) we examined such oscillating probabilities via diffusivity analysis by constructing random walkers driven by chaotically driven bias. In this study, we experimentally implemented the concept of probability chaos using a single-photon source that was chaotically modulated by an external electro-optical modulator that directly yielded random walkers via single-photon observations after a polarization beam splitter. An evident signature was observed in the resulting ensemble average of the time-averaged mean square displacement. Although the experiment involved a scaled-up, proof-of-concept model of a genuine nanoscale oscillator, the experimental observations clearly validate the concept of oscillating probability, paving the way toward future ideal nanoscale systems.

Energy transfer based on optical near-field interactions has been intensively studied theoretically 1) and experimentally. 2,3) The optical excitation generated in the energy level of the smaller quantum dot, denoted by S1 in Fig. 1(a), can be transferred to the upper energy level of the larger dot L2 through optical near-field interactions. 4) The excitation on L2 could then be shifted to the lower energy level L1 via energy dissipation. Such near-field processes have been utilized in lighting and energy applications. 5) Additionally, fundamental studies toward computing and intelligent functions such as solution searching 6) , decision making 7) , and computing 8) are emerging to benefit from the novel optical mechanisms inherent in the subwavelength scale. In the context of functional systems, the generation of periodic signals is one of the most fundamental aspects of the functional systems observed in nature, such as heart beats, and artificially constructed devices, such as clock signals for microprocessors.
Indeed, Naruse et al. theoretically demonstrated optical pulsation using energy transfer among quantum dots interacted via near-field interactions 9) by incorporating a delay mechanism as schematically indicated by D in Fig. 1(a).
Furthermore, combined with additional external delay, indicated by DC in Fig. 1(a), chaotic oscillations and random number generations have been demonstrated. 10) The solid curves in Figs. 1(b,i), 1(b,ii), 1(b,iii), and 1(b,iv) show the time evolution of the population associated with the lower energy level of the larger quantum dot L2 when the coupling strength aC between the external delay system and the original energy transfer system is given by 30 × 10 -3 , 180 × 10 -3 , 182 × 10 -3 , and 207 × 10 -3 , respectively. 10) In Fig. 1(b), periodically oscillating evolution of the population in (i), chaotically evolving populations in (ii) and (iii), and a quasiperiodically evolving population in (iv) are evident.
Let aC be a control parameter, and the red circular and the blue x marks in Fig. 1(c) represent local maxima and local minima of the populations, respectively. Bifurcations are clearly observed, which were adapted from Fig. 3 of our former work. 10) In addition to the nanoscale implementation of such oscillatory behavior, it is intriguing that the where means the displacement of the walker and denotes an ensemble average. One of the major results therein was that the ETMSD underwent large deviations from linear scaling in short time intervals in the case of chaotic oscillations, exhibiting a superdiffusion-like behavior. Such large diffusivity implies benefits in applications such as decision making; for example, the chaotic time series generated by a semiconductor laser provide efficient solutions to the multi-armed bandit problem. 12) Based on these earlier works, we experimentally implemented the concept of dynamically changing probability, including chaotic changes, in this study by using a single-photon source modulated by an external electrooptic modulator. The single photon was then subjected to a polarization beam splitter (PBS) and detected by either of the two avalanche photodiodes (APDs) corresponding to horizontal or vertical polarization. As explained in detail below, the identity of the APD corresponds directly to the direction of the random walker. Consequently, the diffusivity is evaluated directly from the photon measurement sequences. Although the experiment was a scaled-up, proof-of-concept model of the genuine nanoscale oscillators, the experimental observations clearly validate the concept of oscillating probability, paving the way for future nanoscale systems.
The experimental setup is schematically illustrated in Fig. 2. A single photon is emitted from a single individual nitrogen-vacancy (NV) color center, 13) which has a broadband emission spectrum in the visible range (640-700 nm), [14][15][16] excited by a green laser (CNI Lasers, diode-pumped solid-state laser; Wavelength: 532 nm). After passing through a polarizer (denoted by Pol in Fig. 2) and a zeroorder half-wave plate (l/2), the emitted photon impinges on an electro-optical phase modulator (Thorlabs, EO-PM-NR-C1) driven by an external system through an amplifier. The modulated signal is subjected to the PBS, followed by single-photon detection by either APDH or APDV (Excelitas Technologies, SPCM-AQRH-16-FC), which correspond to horizontal and vertical polarization, respectively. The signal is then sent to a time-correlated single-photon-counting system (PicoQuant, PicoHarp 300).
Assume that there is no modulation by the modulator and that the polarization of the input single photon is 45° with respect to the horizontal axis; the photon is detected by APDH or APDV with 50:50 probability. Next, the single photons are modulated by a periodically oscillating signal, which is shown in Fig. 1(b,i), using the electro-optical phase modulator. In this study, the time scale of the nanoscale simulations in Fig. 1(b) was stretched by about a factor of 10 7 in the driving signal in the experiment taking into account the experimentally available single-photon rate (which is on the order of 10 4 /s).
The red and green traces in Fig. 3(a) show the number of photons per 2-ms-duration time bin detected by APDH and APDV, respectively. The total duration of the measurement was about 12.7 s. Figure 3 is a magnified view of Fig. 3(a) corresponding to the initial 1 s. From these images, we can observe that the polarization of the incoming single photons is modulated by the periodic signals, while photons are detected simultaneously at either APDH or APDV, especially when the periodical modulation signal crosses the zero point. That is, the probabilistic attributes of single photons are also observed.
As is well known, when a random walker moves toward the plus and minus directions with probabilities p and q, respectively, the mean displacement is given by , while the variance is given by . Therefore, if the probability of single-photon detection by APDH and APDV is 50:50, the mean square displacement at time t is equal to t since p = q = 1/2, which is called normal diffusion. The black solid line in Fig. 4(a) depicts normal diffusion, meaning that the ETMSD is equal to the time difference t. The magenta marks in Fig. 4(a) whereas its diffusing space is rather limited with the other modulations. These observations are consistent with the numerical studies described in Ref. 11.
To summarize, we experimentally examined the concept of dynamically changing probabilistic behavior by combining single photons emitted from a NV center in a nanodiamond and an electrooptical phase modulator. The single photon measurements made by either of the two photodetectors corresponding to horizontally and vertically polarized light directly generated random walkers, whose diffusivities were analyzed using the mean square displacement as well as by examining the trajectories of the walkers in a phase space. Super-diffusion-like behavior was observed on a short time scale by chaotic modulation of single photons. The results agree well with the theoretical and numerical findings reported in Ref. 11. Although this is a scaled-up, proof-of-concept study, it is noteworthy that large diffusivity was experimentally achievable using a rather simple experimental setup, motivating future implementation to nanoscale optical energy transfer, as originally discussed in Ref. 10. A relevant theoretical examination of nano-scale oscillator is the principle using superradiance proposed by Shojiguchi et al. 17) . The experimental realization would in either method, however, require breakthroughs in device and material levels; exploiting intrinsic attributes in molecular level, such as photochromism, 18) could be one interesting resource. Meanwhile, chaotically oscillating dynamics have been utilized in recent intelligent functions such as decision making 12) , Monte-Carlo computation 19) , and artificial data generation through generative adversarial networks 20) ; these applications indicate the possibility of directly utilizing random physical processes in nature.