JP6260896B2 - Ising model quantum computing device - Google Patents

Description

  The present invention provides a quantum computing device that can easily solve an NP complete problem and the like mapped to an Ising model by easily solving the Ising model.

  The Ising model has been originally studied as a model of a magnetic material, but has recently attracted attention as a model mapped from the NP complete problem. However, it is very difficult to solve the Ising model when the number of sites is large. Therefore, quantum annealing machines and quantum adiabatic machines that implement the Ising model have been proposed.

  The quantum annealing machine solves the Ising model by physically implementing the Ising interaction and Zeeman energy, then cooling the system sufficiently to realize the ground state and observing the ground state. However, when the number of sites is large, the system is trapped in a metastable state in the course of cooling, and the number of metastable states increases exponentially with the number of sites, so the system goes from the metastable state to the ground state. There was a problem that it was difficult to alleviate.

  In the quantum adiabatic machine, the transverse magnetic field Zeeman energy is physically implemented, and then the system is sufficiently cooled to realize a ground state of only the transverse magnetic field Zeeman energy. Then, gradually reduce the transverse magnetic field Zeeman energy and gradually implement the Ising interaction physically, realize the ground state of the system including the Ising interaction and the longitudinal magnetic field Zeeman energy, and observe the ground state By doing so, the Ising model is solved. However, when the number of sites is large, there is a problem that the transverse magnetic field Zeeman energy must be gradually decreased and the speed at which the Ising interaction is gradually physically implemented must be exponentially slowed with respect to the number of sites. It was.

  When mapping an NP complete problem etc. to an Ising model and implementing the Ising model with a physical spin system, the Ising interaction between physically close sites is large, and between physically distant sites The natural law that the Ising interaction is small is a problem. In an artificial Ising model that maps the NP-complete problem, the Ising interaction between physically close sites may be small, and the Ising interaction between physically remote sites may be large. Because. The difficulty of mapping to this natural spin system also makes it difficult to easily solve the NP complete problem and the like.

International Publication No. 2012/118064

  A first conventional technique (see Patent Document 1) and a second conventional technique for solving the above problem will be described. The NP complete problem can be replaced with a magnetic Ising model, and the magnetic Ising model can be replaced with a network of lasers.

  Here, in the Ising model of the magnetic material, the spin direction is reversed (in the case of antiferromagnetic interaction) or in the same direction (strong) so that the energy of the spin arrangement is minimized in the interacting atom pair. Try to point in the case of magnetic interaction).

  On the other hand, in the laser network, oscillation polarization (in the case of the first prior art) or phase (in the case of the second prior art) so that the threshold gain of the oscillation mode is minimized in the interacting laser pair. ) Tries to point in the reverse rotation or phase (for antiferromagnetic interactions) or the same rotation or phase (for ferromagnetic interactions).

  That is, in a system consisting of one laser pair, the polarization or phase of oscillation can be optimized so that the threshold gain of the oscillation mode is minimized. And in a system with many laser pairs, if you try to optimize the polarization or phase of oscillation in one “laser” pair, you cannot optimize the polarization or phase of oscillation in “other” laser pairs. The “total” of the network will be searched for a “compromise point” of polarization or phase of oscillation.

  However, when optimizing the polarization or phase of oscillation across the entire network of lasers, instead of launching a separate oscillation mode for each laser pair, each oscillation mode is launched so that one oscillation mode is launched across the entire network of lasers. It is necessary to synchronize between the lasers.

  As described above, in the first and second prior arts, the pumping current is gradually increased and controlled by each laser, and one oscillation mode in which the threshold gain is the lowest in the entire laser network is started. The polarization or phase is measured, and the spin direction of each atom is measured. Therefore, the problem of trapping in the metastable state in the quantum annealing machine and the mounting speed problem of the Ising interaction in the quantum adiabatic machine can be solved.

  In the first prior art and the second prior art, as described later with reference to FIGS. 1 and 2, not only the magnitude of Ising interaction between physically close sites but also physical The size of the Ising interaction between sites located far away can be freely controlled. Therefore, an artificial Ising model mapped from the NP complete problem or the like can be solved regardless of the physical distance between the sites.

  FIG. 1 shows an outline of the quantum computing device of the Ising model of the first prior art. An outline of the Ising model quantum computing device of the second prior art is shown in FIG.

Let the Ising Hamiltonian be

The Ising interaction mounting unit I12 controls the amplitude and phase of light exchanged between the two surface emitting lasers V1 and V2, thereby simulating the pseudo Ising interaction between the two surface emitting lasers V1 and V2. implementing the magnitude and sign of J _12.

The Ising interaction mounting unit I13 controls the amplitude and phase of light exchanged between the two surface emitting lasers V1 and V3, thereby simulating the pseudo Ising interaction between the two surface emitting lasers V1 and V3. implementing the magnitude and sign of J _13.

The Ising interaction mounting unit I23 controls the amplitude and phase of light exchanged between the two surface emitting lasers V2 and V3, thereby simulating the pseudo Ising interaction between the two surface emitting lasers V2 and V3. implementing the magnitude and sign of J _23.

  The master laser M performs injection locking with respect to the surface emitting lasers V1, V2, and V3, and aligns the oscillation frequencies of the surface emitting lasers V1, V2, and V3 to the same frequency. By optimizing the polarization or phase of oscillation in the entire network of surface emitting lasers V1, V2, and V3 by synchronizing the surface emitting lasers V1, V2, and V3, a network of surface emitting lasers V1, V2, and V3 is used. One oscillation mode can be started up in total.

In the first prior art, the Ising spin measurement unit (not shown) has reached the steady state in the process in which the surface emitting lasers V1, V2, and V3 exchange light, and then the oscillation circles of the surface emitting lasers V1, V2, and V3. By measuring the counterclockwise / clockwise polarization, the upward / downward pseudo Ising spins σ ₁ , σ ₂ , σ ₃ of the surface emitting lasers V 1, V 2, V ₃ are measured.

  However, since the surface-emitting laser V has in-plane anisotropy, it is difficult to oscillate at the same frequency and the same threshold gain for both counterclockwise / clockwise circularly polarized light. Therefore, a certain surface emitting laser V oscillates light having a counterclockwise (or clockwise) circularly polarized light as a single laser rather than oscillating light having a clockwise (or counterclockwise) circularly polarized light. It can be easy. Then, the surface emitting laser V oscillates light having clockwise (or counterclockwise) circularly polarized light as a whole for the entire laser network, but the counterclockwise (or clockwise) circularly polarized light is obtained. It may cause a wrong answer to oscillate the light it has.

In the second prior art, the Ising spin measuring unit (not shown) has a straight line of oscillation of the surface emitting lasers V1, V2, and V3 after the surface emitting lasers V1, V2, and V3 reach a steady state in the process of exchanging light. By measuring the advance / lag of the polarization phase, the upward / downward direction of the pseudo Ising spins σ ₁ , σ ₂ , σ ₃ of the surface emitting lasers V 1, V 2, V ₃ is measured.

  Here, counterclockwise / clockwise circularly polarized light is obtained by superposing horizontally polarized light and vertically polarized light with the same weight with a phase difference of ± π / 2. In other words, the upward / downward information of the Ising spin can be obtained by measuring the phase advance / lag of the vertical polarization without measuring the left / right rotation of the circular polarization and without measuring the horizontal polarization. You can get it. Therefore, the problem of in-plane anisotropy of the surface emitting laser V in the first prior art can be solved.

The principle of the Ising model quantum computing device of the second prior art is shown in FIG. The oscillation phase 0 of the linearly polarized light of the master laser M does not change from the initial state to the steady state. The oscillation phase φ _V (t) of the linearly polarized light of each surface emitting laser V is ideally 0 in the initial state, which is the same as the oscillation phase 0 of the linearly polarized light of the master laser M, and in the steady state, ± π / 2 deviated from the oscillation phase 0 of the linearly polarized light of the master laser M. Φ _V (steady state) = ± π / 2 in the steady state corresponds to σ = ± 1 (in the same order as the compound codes).

For each pair of surface emitting lasers V, when the Ising interaction J _ij is positive, it is energetically advantageous that the pseudo spins σ of the two surface emitting lasers V have different signs. Therefore, each Ising interaction mounting unit makes it easy to start an oscillation mode in which the oscillation phases φ _V (stationary) of the two surface emitting lasers V have different signs and the shift is π.

For each pair of surface emitting lasers V, when the Ising interaction J _ij is negative, it is advantageous in terms of energy that the pseudo spins σ of the two surface emitting lasers V have the same sign. Therefore, each Ising interaction mounting unit makes it easy to start an oscillation mode in which the oscillation phases φ _V (stationary) of the two surface emitting lasers V have the same sign and the deviation is zero.

  However, in the whole Ising model quantum computing device, one oscillation mode is started up as a whole, and in each pair of surface-emitting lasers V, the above-described oscillation mode may actually start up, but it does not necessarily start up. Sometimes not.

By the way, it is desirable that the oscillation phase φ _V (t) of the linearly polarized light of each surface emitting laser V is ideally the same 0 as the oscillation phase 0 of the linearly polarized light of the master laser M in the initial state. Is slightly shifted from the oscillation phase 0 of the linearly polarized light of the master laser M.

The oscillation phase φ _V (t = 0) in the initial state of the linearly polarized light of each surface emitting laser V includes the free-running frequency ω of each surface emitting laser V, the oscillation frequency ω ₀ of the master laser M, and the injection locking width Δω _L (ω (How close can ω ₀ be to injection locking)?

That is, if the free-running frequency ω of each surface-emitting laser V can be matched with the oscillation frequency ω ₀ of the master laser M, the oscillation phase φ _V (t = 0) in the initial state of the linearly polarized light of each surface-emitting laser V ) Becomes 0 which is the same as the oscillation phase 0 of the linearly polarized light of the master laser M. However, since it is difficult to align the free-running frequency ω of each surface emitting laser V with the oscillation frequency ω ₀ of the master laser M, the oscillation phase φ _V (t = 0) slightly deviates from the oscillation phase 0 of the linearly polarized light of the master laser M.

  Therefore, a certain surface-emitting laser V as a single laser oscillates light having an oscillation phase delayed (or advanced) from the oscillation phase 0 of the linear polarization of the master laser M rather than the linear polarization of the master laser M. It may be easy to oscillate light having an oscillation phase that is advanced (or delayed) from the oscillation phase 0. The surface emitting laser V, as the whole laser network, oscillates light having an oscillation phase delayed (or advanced) from the oscillation phase 0 of the linearly polarized light of the master laser M. An erroneous answer may be generated that oscillates light having an oscillation phase that is advanced (or delayed) from the oscillation phase 0 of the linearly polarized light of the laser M.

Further, the steady-state oscillation phase φ _V (steady state) of the linearly polarized light of each surface emitting laser V is measured using homodyne detection with reference to the oscillation phase 0 of the linearly polarized light of the master laser M. However, fluctuations in the optical path length from each surface-emitting laser V / master laser M to a detector (not shown) cause an error in reading the steady-state oscillation phase φ _V (steady state) of the linearly polarized light of each surface-emitting laser V. obtain. And it is difficult to suppress fluctuations in the optical path length from each surface emitting laser V / master laser M to a detector (not shown).

  In FIG. 2, when the number of Ising sites is M, M surface emitting lasers V are required, and M (M-1) / 2 are required for the Ising interaction mounting portions. When the number of Ising sites increases, the Ising model quantum computing device becomes large and complex.

  Accordingly, in order to solve the above-described problems, an object of the present invention is to prevent a read error and simplify a circuit configuration in an Ising model quantum computing device.

  In order to achieve the above object, a plurality of laser pulses having the same oscillation frequency are oscillated using harmonic mode-locked oscillation, and the amplitude and phase of light exchanged between the plurality of laser pulses are controlled. Measuring the spin direction of multiple Ising sites by implementing the magnitude and phase of the Ising interaction between multiple laser pulses and measuring the steady-state oscillation phase of the multiple laser pulses; did.

  Specifically, the present invention relates to a harmonic mode-locked laser that oscillates a plurality of laser pulses having the same oscillation frequency corresponding to a plurality of sites of the Ising model, and two pairs of the plurality of laser pulses. An Ising interaction implementation that implements the magnitude and sign of the pseudo Ising interaction between the two laser pulses by controlling the amplitude and phase of the light exchanged between the two laser pulses; An Ising spin measurement unit that measures pseudo Ising spins of the plurality of laser pulses by measuring an oscillation phase of the plurality of laser pulses after a plurality of laser pulses reach a steady state in a process of exchanging light. And an Ising model quantum computation device.

  According to this configuration, since the plurality of laser pulses have the same oscillation frequency, the oscillation phase in the initial state of each laser pulse changes to one of the two oscillation phases in the steady state of each laser pulse. Is not near to the other phase. Therefore, a read error can be prevented in the Ising model quantum computing device.

  The first and second conventional techniques require a master laser, whereas the present invention can eliminate the need for a master laser. Furthermore, while the first and second conventional techniques require as many surface emitting lasers as the number of Ising sites, in the present invention, only one harmonic mode-locked laser needs to be prepared. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  Further, according to the present invention, the Ising interaction implementation unit generates a delay time equal to a time interval between the two laser pulses, and a part of one of the two laser pulses is input. For implementation of Ising interaction that amplitude-phase modulates a part of the input one laser pulse, and combines the amplitude-phase modulated part of the one laser pulse with the other laser pulse of the two laser pulses A delay modulation unit, wherein the Ising interaction implementation delay modulation unit includes a single set of delay lines and amplitude phase modulators for a plurality of pairs of laser pulses having the same time interval among the plurality of laser pulses. Is an Ising model quantum computing device characterized in that

  According to this configuration, it is sufficient to share the Ising interaction mounting unit for a plurality of pairs of laser pulses having the same time interval, instead of preparing the Ising interaction mounting unit for each pair of laser pulses. That is, when the number of Ising sites is M, the first and second prior arts require M (M−1) / 2 Ising interaction implementation units, whereas in the present invention, the Ising mutual implementation unit. There is no need for M (M−1) / 2 action mounting sections.

  In particular, when M laser pulses corresponding to M Ising sites are arranged at equal intervals (time interval τ) in the ring resonator of the harmonic mode-locked laser, the time interval between different laser pulses is , Τ,..., (M−1) τ, and (M−1) types, it is sufficient to prepare only the (M−1) types of Ising interaction mounting units. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  The present invention also provides a plurality of laser pulses corresponding to a plurality of sites of the Ising model and having the same oscillation frequency, and a single laser pulse corresponding to the DC magnetic field of the Ising model and having the same oscillation frequency. And for each pair of the plurality of laser pulses, by controlling the amplitude and phase of the light exchanged between the two laser pulses, between the two laser pulses. Controlling the amplitude and phase of light injected from the single laser pulse with respect to each of the plurality of laser pulses. To implement a pseudo Zeeman energy magnitude and sign for each of the plurality of laser pulses. And a plurality of laser pulses after reaching a steady state in the process in which the plurality of laser pulses are exchanged and injected, by measuring the oscillation phase of the plurality of laser pulses, An Ising model quantum computation device comprising: an Ising spin measurement unit that measures Ising spins.

  According to this configuration, since the plurality of laser pulses have the same oscillation frequency, the oscillation phase in the initial state of each laser pulse changes to one of the two oscillation phases in the steady state of each laser pulse. Is not near to the other phase. Therefore, a read error can be prevented in the Ising model quantum computing device.

  In the first and second prior arts, the master laser is used for the purpose of injection locking to the surface emitting laser, whereas in the present invention, the direct current of the Ising model is used for the purpose of mounting Zeeman energy. A master pulse corresponding to the magnetic field is used. Furthermore, while the first and second conventional techniques require as many surface emitting lasers as the number of Ising sites, in the present invention, only one harmonic mode-locked laser needs to be prepared. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  Further, according to the present invention, the Ising interaction implementation unit generates a delay time equal to a time interval between the two laser pulses, and a part of one of the two laser pulses is input. For implementation of Ising interaction that amplitude-phase modulates a part of the input one laser pulse, and combines the amplitude-phase modulated part of the one laser pulse with the other laser pulse of the two laser pulses A delay modulation unit, wherein the Zeeman energy implementation unit generates a delay time equal to a time interval between each of the single laser pulse and the plurality of laser pulses, and a part of the single laser pulse. A part of the single laser pulse that is amplitude-phase modulated and a part of the single laser pulse that is amplitude-phase modulated. A Zeeman energy mounting delay modulation unit that multiplexes each of the plurality of laser pulses, wherein the Ising interaction mounting delay modulation unit and the Zeeman energy mounting delay modulation unit include the plurality of laser pulses and the single laser pulse. The Ising model quantum computing device is characterized by sharing a single set of delay line and amplitude phase modulator for a plurality of pairs of laser pulses having the same time interval among laser pulses.

  According to this configuration, it is sufficient to share the Ising interaction mounting unit for a plurality of pairs of laser pulses having the same time interval, instead of preparing the Ising interaction mounting unit for each pair of laser pulses. That is, when the number of Ising sites is M, the first and second prior arts require M (M−1) / 2 Ising interaction implementation units, whereas in the present invention, the Ising mutual implementation unit. There is no need for M (M−1) / 2 action mounting sections. In addition, it is sufficient to prepare the Zeeman energy mounting unit in common with the Ising interaction mounting unit, instead of preparing the Zeeman energy mounting unit separately from the Ising interaction mounting unit.

  In particular, M laser pulses corresponding to M Ising sites and one master pulse corresponding to the DC magnetic field of the Ising model are equally spaced (time interval τ) in the ring resonator of the harmonic mode-locked laser. Since the time intervals between different pulses are M types of τ,..., Mτ, the Ising interaction implementation unit and the Zeeman energy implementation unit (these need only be shared) It is sufficient to prepare only M types. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  In the present invention, the Ising spin measurement unit may detect the different laser pulses in a pseudo manner by delay-detecting whether the oscillation phase of different laser pulses is in phase or out of phase among the plurality of laser pulses. The Ising model quantum computing device is characterized by measuring whether Ising spin is in the same direction or in the opposite direction.

  According to this configuration, it is possible to eliminate the problem of fluctuation of the optical path length from each surface emitting laser / master laser to the detector in the second prior art. Therefore, a read error can be prevented in the Ising model quantum computing device.

  In the second conventional technique, one homodyne detection device is required for each surface emitting laser, and as many homodyne detection devices as the number of Ising sites are required. It is sufficient to prepare one delay detector for the laser pulse. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  Further, the present invention is the Ising model quantum computation device, wherein the harmonic mode-locked laser is a passive mode-locked laser.

  According to this configuration, a saturable absorber in the ring resonator of the harmonic mode-locked laser can be used to oscillate a series of laser pulses without requiring external control.

  Further, the present invention is the Ising model quantum computation device, wherein the harmonic mode-locked laser is a forced mode-locked laser.

  According to this configuration, it is possible to oscillate a series of a plurality of laser pulses with a high repetition frequency using the modulator in the ring resonator of the harmonic mode-locked laser.

  The harmonic mode-locked laser may be a mode-locked fiber laser, and the Ising interaction mounting unit may include a fiber delay line.

According to this configuration, since an optical fiber is used as a transmission path for a plurality of laser pulses, it is possible to provide an Ising model quantum computing device with low optical loss, high optical intensity, and a simple circuit configuration. Here, the time during which the optical path length of the optical fiber is stably maintained without being influenced by mechanical vibration or thermal expansion is 10 ⁻⁴ s, and the Ising model quantum computer stabilizes the Ising spin. Since the time required for this is 10 ⁻⁹ s, the Ising spin measurement result is not influenced by mechanical vibration or thermal expansion of the optical fiber.

  The present invention can prevent a read error and simplify a circuit configuration in an Ising model quantum computing device.

It is a figure which shows the outline | summary of the quantum computing device of the Ising model of the 1st prior art. It is a figure which shows the outline | summary of the quantum computer of the 2nd prior art Ising model. It is a figure which shows the principle of the quantum computing device of the Ising model of the 2nd prior art. It is a figure which shows the outline | summary of the quantum calculation apparatus of the Ising model of 1st Embodiment. It is a figure which shows the principle of the quantum calculation apparatus of the Ising model of 1st Embodiment. It is a figure which shows the structure of the quantum calculation apparatus of the Ising model of 1st Embodiment. It is a figure which shows the structure of the quantum calculation apparatus of the Ising model of 1st Embodiment. It is a figure which shows the process of the quantum calculation apparatus of the Ising model of 1st Embodiment. It is a figure which shows the process of the quantum calculation apparatus of the Ising model of 1st Embodiment. It is a figure which shows the outline | summary of the quantum calculation apparatus of the Ising model of 2nd Embodiment. It is a figure which shows the principle of the quantum calculation apparatus of the Ising model of 2nd Embodiment. It is a figure which shows the structure of the quantum calculation apparatus of the Ising model of 2nd Embodiment. It is a figure which shows the structure of the quantum calculation apparatus of the Ising model of 2nd Embodiment. It is a figure which shows the process of the quantum calculation apparatus of the Ising model of 2nd Embodiment. It is a figure which shows the process of the quantum calculation apparatus of the Ising model of 2nd Embodiment. It is a figure which shows the process of the quantum calculation apparatus of the Ising model of 2nd Embodiment.

  Embodiments of the present invention will be described with reference to the accompanying drawings. The embodiments described below are examples of the present invention, and the present invention is not limited to the following embodiments. In the present specification and drawings, the same reference numerals denote the same components.

(First embodiment)
An outline of the Ising model quantum computation device of the first embodiment is shown in FIG. In the first embodiment, the Ising Hamiltonian is expressed by Equation 3.

  A harmonic mode-locked laser (not shown) corresponds to a plurality of sites of the Ising model and oscillates a plurality of laser pulses P1, P2, and P3 having the same oscillation frequency.

The Ising interaction implementation I12 controls the amplitude and phase of the light exchanged between the two laser pulses P1 and P2, thereby controlling the pseudo Ising interaction J ₁₂ between the two laser pulses P1 and P2. Implement the size and sign of.

Ising interaction mounting portion I13, by controlling the two laser pulses P1, P3 light amplitude and phase exchanged between the two pseudo Ising interaction J ₁₃ between the laser pulses P1, P3 Implement the size and sign of.

Ising interaction mounting portion I23, by controlling the two laser pulses P2, P3 amplitude and phase of light exchanged between the two laser pulses P2, P3 pseudo Ising interaction J ₂₃ between Implement the size and sign of.

The Ising spin measurement unit (not shown) measures the advance / lag of the oscillation phase of the laser pulses P1, P2, and P3 after the laser pulses P1, P2, and P3 reach a steady state in the process of exchanging light. The upward / downward direction of the pseudo Ising spins σ ₁ , σ ₂ , and σ ₃ of the laser pulses P1, P2, and P3 is measured.

  In this way, the pumping current is gradually increased and controlled by a harmonic mode-locked laser (not shown), and one oscillation mode in which the threshold gain is the lowest in the entire network of the laser pulses P1, P2, and P3 is started, and the laser pulses P1, P2 are started. , P3 oscillation phase is measured, and the spin direction of each atom corresponding to the laser pulses P1, P2, P3 is measured.

The principle of the Ising model quantum computing device of the first embodiment is shown in FIG. The oscillation phase φ _V (t) of each laser pulse P is 0 in the initial state, and is ± π / 2 deviated from the oscillation phase 0 in the initial state in the steady state. Φ _V (steady state) = ± π / 2 in the steady state corresponds to σ = ± 1 (in the same order as the compound codes).

For each pair of laser pulses P, it is energetically advantageous that the pseudo spins σ of the two laser pulses P are of opposite signs when the Ising interaction J _ij is positive. Therefore, each Ising interaction mounting unit makes it easy to start an oscillation mode in which the oscillation phase φ _V (steady state) of the two laser pulses P has different signs and the deviation is π.

For each pair of laser pulses P, it is energetically advantageous that the pseudo spins σ of the two laser pulses P have the same sign when the Ising interaction J _ij is negative. Therefore, each Ising interaction mounting unit makes it easy to start an oscillation mode in which the oscillation phase φ _V (stationary) of the two laser pulses P has the same sign and the deviation is zero.

  However, in the whole Ising model quantum computing device, one oscillation mode is caused to rise as a whole, and in each pair of laser pulses P, the above-described oscillation mode may actually rise or not necessarily rise. Sometimes.

The calculation principle shown in FIGS. 4 and 5 will be described in detail. For each laser pulse P1, P2, and P3, the rate equation for the oscillation intensity A _Vi (t), the oscillation phase φ _Vi (t), and the carrier inversion distribution number difference N _Ci (t) is as shown in Equation 4-7: Become.

ω is the oscillation frequency. Q is the resonator Q value of each laser pulse P. P is the number of electrons injected per second for each laser pulse P to realize the population inversion, that is, the pumping rate. -(1/2) (ω / Q) A _Vi (t) in Expression 4 indicates a rate at which the oscillation intensity A _Vi (t) decreases with time due to the resonator loss.

τ _sp is the electron lifetime due to spontaneous emission to other oscillation modes other than the laser oscillation mode. β is a coupling constant to the laser oscillation mode in the total spontaneous emission light. (1/2) E _CVi (t) A _Vi (t) in Equation 4 indicates a rate at which the oscillation intensity A _Vi (t) increases with time due to stimulated emission. E _CVi (t) in Equation 4 indicates a rate at which the oscillation intensity A _Vi (t) increases with time due to spontaneous emission.

The term related to ξ _{ij in} Equation 4 is a term related to the mutual injection light between the laser pulses P for implementing the Ising interaction. − (Ω / Q) (1/2) ξ _ij A _Vj (t) cos {φ _Vj (t) −φ _Vi (t)} in Expression 4 indicates that light is transmitted from the j-th site to the i-th site. It shows the rate at which the oscillation intensity A _Vi (t) at the i-th site changes over time when injected. Σ (j ≠ i) in Expression 4 indicates contributions from all the sites other than the i-th site (j-th) in the i-th site.

The term related to ξ _{ij in} Formula 5 is a term related to the mutual injection light between the laser pulses P for implementing the Ising interaction. -(1 / A _Vi (t)) (ω / Q) (1/2) in Formula 5 ξ _ij A _Vj (t) sin {φ _Vj (t) −φ _Vi (t)} is the j-th site This shows the rate at which the oscillation phase φ _Vi (t) at the i-th site changes with time when light is injected from the i-th site to the i-th site. Σ (j ≠ i) in Expression 5 indicates contributions from all sites (jth) other than the ith site in the ith site.

F _AV , F _φV, and F _N indicate noise with respect to the oscillation intensity A _Vi (t), the oscillation phase φ _Vi (t), and the carrier inversion distribution number difference N _Ci (t) at the i-th site, respectively.

Ｆ_ＡＶを無視して、数式８を変形すると、数式９のようになる。

In the steady state, Equation 4 becomes Equation 8.

If _FAV is ignored and Equation 8 is transformed, Equation 9 is obtained.

Here, σ _i of the Ising model takes −1 or +1, and sin φ _Vi can take from −1 to +1. Therefore, paying attention to the similarity between the Ising model and the laser system, if σ _i = sinφ _Vi is set, Equation 9 becomes Equation 10.

When Equation 10 is added to all M sites, Equation 11 is obtained, which is expressed as the overall threshold gain ΣE _CVi of the laser system.

Here, since mutual injection is performed between the laser pulses P, A _Vi (t) to A _Vj (t) can be set. Then, in the above definition of σ _i = sin φ _Vi , since σ _i = ± 1, φ _Vi to φ _Vj to ± π / 2 can be set. At this time, Expression 11 becomes Expression 12.

Here, when the harmonic mode-locked laser has a uniform medium, the oscillation phase state {σ _i } that realizes the minimum threshold gain ΣE _CVi is selected as the entire laser system. That is, one specific oscillation mode is selected for the entire laser system. Then, due to the competition between the oscillation modes, one specific oscillation mode suppresses the other oscillation modes.

That is, ΣE _CVi in Expression 12 is minimized for the entire laser system. On the other hand, (ω / Q) M in Expression 12 is constant for the entire laser system. Therefore, as a whole laser system, Σξ _ij σ _i σ _{j in} Expression 12 is minimized.

putting the ξ _ij = _{J ij,} the entire laser system and Σξ _ij σ _i σ _j is minimized, ΣJ _{ij σ i σ j} is also minimized. That is, the ground state that minimizes the Ising Hamiltonian of Equation 3 is realized.

In order to improve the calculation accuracy, the difference between the minimum threshold gain and the next smallest threshold gain for the laser oscillation mode of the entire laser system is determined by the saturation gain E _CV and the photon attenuation rate ω determined by the spontaneous emission rate. It is necessary to make it sufficiently larger than β (ω / Q) (1 / R) which is the difference of / Q. Here, R = I / I _th −1 is a normalized pump rate, and I and I _th are an injection current and a laser oscillation threshold value, respectively. Therefore, calculation accuracy can be improved by reducing β and increasing R.

As described with reference to FIGS. 4 and 5, since the plurality of laser pulses P have the same oscillation frequency, the oscillation phase φ _Vi (t = 0) in the initial state of each laser pulse P is equal to each laser pulse. Of the two oscillation phases φ _Vi (steady state) = ± π / 2 in the steady state of P, there is no such thing as being close to one phase and far from the other phase. Therefore, a read error can be prevented in the Ising model quantum computing device.

  In the first and second prior arts, the master laser M is required, whereas in the present embodiment, the master laser M can be omitted. Furthermore, in the first and second prior arts, as many surface emitting lasers V as the number of Ising sites are required, in this embodiment, it is sufficient to prepare only one harmonic mode-locked laser. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  FIG. 6 shows a first configuration of the Ising model quantum computation device according to the first embodiment. In the configuration of FIG. 6, the harmonic mode-locked laser is a passively mode-locked fiber laser.

  The passive mode-locked fiber laser is composed of a pump laser 1, a WDM (Wavelength Division Multiplexing) coupler 2, an erbium-doped fiber 3, a quarter-wave plate 4, a half-wave plate 5, a beam splitter 6 and a quarter-wave plate 7. Composed. The quarter wavelength plate 4, the half wavelength plate 5, and the quarter wavelength plate 7 constitute a saturable absorber using NPR (Nonlinear Polarization Rotation). A passively mode-locked fiber laser can oscillate a series of laser pulses P using a saturable absorber in a ring resonator without requiring external control.

  FIG. 7 shows a second configuration of the Ising model quantum computation device according to the first embodiment. In the configuration of FIG. 7, the harmonic mode-locked laser is a forced mode-locked fiber laser.

  The forced mode-locked fiber laser includes a pump laser 1, a WDM coupler 2, an erbium-doped fiber 3, a dispersion shift fiber 8, a coupler 9, an isolator 10, a modulator 11, a filter 12, a coupler 13, a clock extractor 14, and a phase controller 15. And an amplifier 16. The modulator 11 sets the repetition frequency of the series of the plurality of laser pulses P to an arbitrary integer multiple of the frequency c / nL (c is the speed of light and n is the refractive index) determined by the ring resonator length L. The forced mode-locked fiber laser can oscillate a series of a plurality of laser pulses P having a high repetition frequency by using the modulator 11 in the ring resonator.

  6 and 7, the method of oscillating a series of a plurality of laser pulses P is different, but the method of implementing Ising interaction and the method of measuring Ising spin are the same. Here, it is assumed that M laser pulses P corresponding to M Ising sites are arranged at equal intervals (time interval τ) in the ring resonator of the harmonic mode-locked laser.

  The Ising interaction implementation unit includes amplitude phase modulators AP1, AP2,..., AP (M-1) and fiber delay lines DL1, DL2,.

  The set of the amplitude phase modulator AP1 and the fiber delay line DL1 generates a delay time τ, and one of the two laser pulse P pairs whose time interval is τ is input as a part of the input. A part of the one laser pulse P that has been subjected to amplitude-phase modulation, and a part of the one laser pulse P that has undergone amplitude-phase modulation is replaced with the other laser pulse of the pair of two laser pulses P whose time interval is τ. Combine with P.

Here, the set of the amplitude phase modulator AP1 and the fiber delay line DL1 is shared for any two pairs of laser pulses P whose time interval is τ. Thus, the set of amplitude phase modulator AP1 and fiber delay line DL1 implements the magnitude and sign of a pseudo Ising interaction J _ij between two laser pulses P whose time interval is τ.

  The set of the amplitude phase modulator AP2 and the fiber delay line DL2 generates a delay time 2τ, and a part of one laser pulse P of two laser pulse P pairs whose time interval is 2τ is input and input. A part of the one laser pulse P that has been subjected to amplitude phase modulation, and a part of the one laser pulse P that has been subjected to amplitude phase modulation is replaced with the other laser pulse of the pair of two laser pulses P whose time interval is 2τ. Combine with P.

Here, the set of the amplitude phase modulator AP2 and the fiber delay line DL2 is shared for any two pairs of laser pulses P whose time interval is 2τ. Thus, a set of amplitude and phase modulators AP2 and fiber delay line DL2 implements magnitude and sign of the pseudo Ising interaction J _ij between two laser pulses P time interval is 2.tau.

  As for the set of the amplitude phase modulator AP (M-1) and the fiber delay line DL (M-1), the set of the amplitude phase modulator AP1 and the fiber delay line DL1, and the set of the amplitude phase modulator AP2 and the fiber delay line DL2. The same is true.

  The processing of the Ising model quantum computation device of the first embodiment is shown in FIGS. Here, it is assumed that three laser pulses P corresponding to three Ising sites are arranged at equal intervals (time interval τ) in the ring resonator of the harmonic mode-locked laser.

-Arg _{(J ij),} if any positive _{J ij} is, it represents a reverse-phase _modulation, if _{J ij} is negative, represents the phase modulation. That is, if J _ij is positive, an oscillation mode in which the oscillation phase of the two laser pulses P is shifted by π is likely to rise, and if J _ij is negative, the oscillation phase of the two laser pulses P is An oscillation mode in which the deviation is 0 is made to rise easily.

First, in the upper part of FIG. 8, the laser pulse P1 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Ising interactions J ₁₂ and J ₁₃ are mounted.

A part of the laser pulse P1 is input to the fiber delay line DL1, and is subjected to amplitude modulation proportional to | J ₁₂ | and phase modulation of −arg (J ₁₂ ) by the amplitude phase modulator AP1, and the fiber delay line DL1. The signal is output from the fiber delay line DL1 after a delay time τ. A part of the laser pulse P1 output from the fiber delay line DL1 is combined with the laser pulse P2 being input to the fiber delay lines DL1 and DL2. Thus, the magnitude and sign of the Ising interaction J ₁₂ is mounted.

A part of the laser pulse P1 is input to the fiber delay line DL2, is subjected to amplitude modulation proportional to | J ₁₃ | and phase modulation of −arg (J ₁₃ ) by the amplitude phase modulator AP2, and is transmitted by the fiber delay line DL2. The signal is output from the fiber delay line DL2 after a delay time 2τ. A part of the laser pulse P1 output from the fiber delay line DL2 is combined with the laser pulse P3 being input to the fiber delay lines DL1 and DL2. Thus, the magnitude and sign of the Ising interaction J ₁₃ is mounted.

Next, in the middle part of FIG. 8, the laser pulse P2 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Ising interactions J ₂₃ and J ₂₁ are implemented.

A part of the laser pulse P2 is input to the fiber delay line DL1, and is subjected to amplitude modulation proportional to | J ₂₃ | and phase modulation of −arg (J ₂₃ ) by the amplitude phase modulator AP1, and the fiber delay line DL1. The signal is output from the fiber delay line DL1 after a delay time τ. Part of the laser pulse P2 output from the fiber delay line DL1 is combined with the laser pulse P3 being input to the fiber delay lines DL1 and DL2. Thus, the magnitude and sign of the Ising interaction J ₂₃ are mounted.

A part of the laser pulse P2 is input to the fiber delay line DL2, and is subjected to amplitude modulation proportional to | J ₂₁ | and phase modulation of −arg (J ₂₁ ) by the amplitude phase modulator AP2, and is transmitted by the fiber delay line DL2. The signal is output from the fiber delay line DL2 after a delay time 2τ. A part of the laser pulse P2 output from the fiber delay line DL2 is combined with the laser pulse P1 being input to the fiber delay lines DL1 and DL2. Thus, the magnitude and sign of the Ising interaction J ₂₁ is mounted.

Next, in the lower part of FIG. 8, the laser pulse P3 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Ising interactions J ₃₁ and J ₃₂ are implemented.

A part of the laser pulse P3 is input to the fiber delay line DL1, and is subjected to amplitude modulation proportional to | J ₃₁ | and phase modulation of −arg (J ₃₁ ) by the amplitude phase modulator AP1, and the fiber delay line DL1. The signal is output from the fiber delay line DL1 after a delay time τ. A part of the laser pulse P3 output from the fiber delay line DL1 is combined with the laser pulse P1 being input to the fiber delay lines DL1 and DL2. Thus, the magnitude and sign of the Ising interaction J ₃₁ is mounted.

A part of the laser pulse P3 is input to the fiber delay line DL2, is subjected to amplitude modulation proportional to | J ₃₂ | and phase modulation of −arg (J ₃₂ ) by the amplitude phase modulator AP2, and is received by the fiber delay line DL2. The signal is output from the fiber delay line DL2 after a delay time 2τ. A part of the laser pulse P3 output from the fiber delay line DL2 is combined with the laser pulse P2 being input to the fiber delay lines DL1 and DL2. Thus, the magnitude and sign of the Ising interaction J ₃₂ are mounted.

  8 are repeated in the above order until the laser pulses P1, P2, and P3 reach a steady state in the process of exchanging light.

  As described with reference to FIGS. 8 and 9, for each pair of laser pulses P, an Ising interaction mounting unit is not prepared, but an Ising interaction mounting is performed for a plurality of pairs of laser pulses P having the same time interval. It is enough to share the department. That is, when the number of Ising sites is M, the first and second prior arts require M (M-1) / 2 Ising interaction mounting units, whereas in this embodiment, the Ising site is Ising sites. It is not necessary to have M (M−1) / 2 interaction mounting units.

  In particular, when M laser pulses P corresponding to M Ising sites are arranged at equal intervals (time interval τ) in the ring resonator of the harmonic mode-locked laser, the time between different laser pulses P Since the intervals are (M−1) types of τ,..., (M−1) τ, it is sufficient to prepare only the (M−1) types of Ising interaction mounting units. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

In FIGS. 8 and 9, Ising interactions J _ij and J _ji are mounted at different timings. Therefore, in this embodiment, not only a normal Ising model in which J _ij = J _ji but also a general Ising model in which J _ij ≠ J _ji can be solved.

  6 and 7, the Ising spin measurement unit includes beam splitters B1 and B2, reflectors R1 and R2, and detectors DI and DO.

  The Ising spin measurement unit delay-detects whether the oscillation phase of different laser pulses P among the plurality of laser pulses P is in-phase or anti-phase, so that pseudo Ising spins of different laser pulses P are in the same direction. And the reverse direction is measured.

  In the present embodiment, the Ising spin measurement unit performs delayed detection between different laser pulses P separated by one pulse. As a modification, the Ising spin measurement unit may perform delayed detection between different laser pulses P separated by several pulses.

  The preceding laser pulse P is output from the beam splitter 6 in FIG. 6 or the coupler 13 in FIG. 7, passes through the reflection at the beam splitter B1, the reflection at the reflection mirror R1, and the reflection at the reflection mirror R2, and then enters the beam splitter B2. To reach. The subsequent laser pulse P is output from the beam splitter 6 in FIG. 6 or the coupler 13 in FIG. 7, passes through the beam splitter B1, and reaches the beam splitter B2.

  Here, the difference between the optical path length of the beam splitter B1 to the reflecting mirrors R1 and R2 to the beam splitter B2 and the optical path length of the beam splitter B1 to the beam splitter B2 is the difference between the preceding laser pulse P and the subsequent laser pulse P. Equal to the interval. Therefore, when the preceding laser pulse P and the subsequent laser pulse P reach the beam splitter B2, interference between the preceding laser pulse P and the subsequent laser pulse P occurs.

  When the oscillation phase between the preceding laser pulse P and the succeeding laser pulse P is 0 (in phase), the preceding laser pulse P reflected by the beam splitter B2 and the succeeding light transmitted through the beam splitter B2 are transmitted. The laser pulse P of FIG. 1 generates constructive interference, and a high light intensity is detected in the detector DI (I is In-phase).

  When a high light intensity is detected by the detector DI, the pseudo Ising spins of the preceding laser pulse P and the succeeding laser pulse P are measured in the same direction.

  When the oscillation phase between the preceding laser pulse P and the succeeding laser pulse P is π (reverse phase), the preceding laser pulse P that has passed through the beam splitter B2 and the beam splitter B2 are reflected. Subsequent laser pulses P generate constructive interference, and high light intensity is detected at the detector DO (O is out-of-phase).

  When high light intensity is detected by the detector DO, the pseudo Ising spin of the preceding laser pulse P and the succeeding laser pulse P is measured in the reverse direction.

  As described with reference to FIGS. 6 and 7, the problem of fluctuation in the optical path length from each surface emitting laser V / master laser M to the detector in the second prior art can be eliminated. Therefore, a read error can be prevented in the Ising model quantum computing device.

  In the second prior art, one homodyne detection device is required for each surface-emitting laser V, and as many homodyne detection devices as the number of Ising sites are required. It is sufficient to prepare one delay detection device for all the laser pulses P. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

As described with reference to FIGS. 6 and 7, since an optical fiber is used as a transmission path for a plurality of laser pulses P, an Ising model quantum computing device with low optical loss, high optical intensity, and a simple circuit configuration is provided. Can be provided. Here, the time during which the optical path length of the optical fiber is stably maintained without being affected by mechanical vibration or thermal expansion is 10 ⁻⁴ s, and the quantum computer of the Ising model can measure Ising spin. Since the required time is 10 ⁻⁹ s, the measurement result of Ising spin is not affected by mechanical vibration or thermal expansion of the optical fiber.

(Second Embodiment)
An overview of the Ising model quantum computation device of the second embodiment is shown in FIG. In the second embodiment, the Ising Hamiltonian is expressed by Equation 13.

  A harmonic mode-locked laser (not shown) corresponds to a plurality of sites of the Ising model, corresponds to a plurality of laser pulses P1, P2, P3 having the same oscillation frequency, and a DC magnetic field of the Ising model, and has the same oscillation. A single master pulse P0 having a frequency is oscillated.

  The Ising interaction mounting units I12, I13, and I23 perform the same processing as in FIG.

The Zeeman energy mounting unit Z1 mounts the magnitude and sign of the pseudo Zeeman energy λ _{1 in} the laser pulse P1 by controlling the amplitude and phase of the light injected from the master pulse P0 for the laser pulse P1. .

The Zeeman energy mounting unit Z2 controls the magnitude and sign of the pseudo Zeeman energy λ _{2 in} the laser pulse P2 by controlling the amplitude and phase of light injected from the master pulse P0 for the laser pulse P2. .

The Zeeman energy mounting unit Z3 controls the magnitude and sign of the pseudo Zeeman energy λ _{3 in} the laser pulse P3 by controlling the amplitude and phase of the light injected from the master pulse P0 for the laser pulse P3. .

An Ising spin measuring unit (not shown) measures the advance / lag of the oscillation phase of the laser pulses P1, P2, and P3 after reaching a steady state in the process in which the laser pulses P1, P2, and P3 are exchanged and injected. Thus, the upward / downward direction of the pseudo Ising spins σ ₁ , σ ₂ , and σ ₃ of the laser pulses P1, P2, and P3 is measured.

  In this way, the pumping current is gradually increased and controlled by a harmonic mode-locked laser (not shown), and one oscillation mode in which the threshold gain is the lowest in the entire network of laser pulses P1, P2, and P3 is started, and laser pulses P1, P2 are started. , P3 oscillation phase is measured, and the spin direction of each atom corresponding to the laser pulses P1, P2, P3 is measured.

The principle of the Ising model quantum computing device of the second embodiment is shown in FIG. The oscillation phase 0 of the master pulse P0 does not change from the initial state to the steady state. The oscillation phase φ _V (t) of each laser pulse P is 0 in the initial state, and is ± π / 2 deviated from the oscillation phase 0 in the initial state in the steady state. Φ _V (steady state) = ± π / 2 in the steady state corresponds to σ = ± 1 (in the same order as the compound codes).

  Each Ising interaction implementation unit performs the same processing as in FIG.

For each laser pulse P, when the Zeeman energy λ _i is positive, it is energetically advantageous that the pseudo spin σ of the laser pulse P is −1. Therefore, the Zeeman energy mounting portion, the oscillation phase phi _V of the laser pulses P are such that - [pi] / 2, so that the oscillation mode is likely to rise.

For each laser pulse P, when the Zeeman energy λ _i is negative, it is energetically advantageous that the pseudo spin σ of the laser pulse P is +1. Therefore, the Zeeman energy mounting portion, the oscillation phase phi _V of the laser pulses P is such that + [pi / 2, so that the oscillation mode is likely to rise.

  However, in the whole Ising model quantum computing device, one oscillation mode is caused to rise as a whole, and in each pair of laser pulses P, the above-described oscillation mode may actually rise or not necessarily rise. Sometimes.

The calculation principle shown in FIGS. 10 and 11 will be described in detail. For each laser pulse P1, P2, P3, the rate equation for the oscillation intensity A _Vi (t), the oscillation phase φ _Vi (t) and the carrier inversion distribution number difference N _Ci (t) is as shown in Equation 14-17. Become.

ω is the oscillation frequency. Q is the resonator Q value of the master pulse P0 and each laser pulse P. n _M is the number of photons from the master pulse P0. P is the same as Equation 6. -(1/2) (ω / Q) A _Vi (t) in Expression 14 is the same as Expression 4.

τ _sp is the same as Equations 6 and 7. β is the same as Equation 7. (1/2) E _CVi (t) A _Vi (t) in Expression 14 and E _CVi (t) in Expression 14 are the same as Expression 4.

Term xi] _ij terms and Equation 15 xi] _ij of equation (14) involving involving is similar to equation 4 and 5. F _AV , F _φV and F _N are the same as those in Expression 4-6.

For the light injected into each laser pulse P, the intensity of the component having a phase shifted by π / 2 from the oscillation phase 0 of the master pulse P0 is proportional to η _i .

The term related to η _{i in} Equation 14 is a term related to Zeeman energy. (Ω / Q) √n _M {−η _i sinφ _Vi (t)} in Expression 14 is the oscillation intensity A _Vi at the i-th site when light is injected from the master pulse P0 to the i-th site. (T) shows a rate that changes over time.

The term related to η _{i in} Expression 15 is a term related to Zeeman energy. (1 / A _Vi (t)) (ω / Q) √n _M {−η _i cos φ _Vi (t)} in Expression 15 is obtained when light is injected from the master pulse P0 to the i-th site. It shows the rate at which the oscillation phase φ _Vi (t) at the i-th site changes over time.

Ｆ_ＡＶを無視して、数式１８を変形すると、数式１９のようになる。

In the steady state, Equation 14 becomes Equation 18.

Ignoring F _AV, by transforming Equation 18, it becomes as Equation 19.

Here, σ _i of the Ising model takes −1 or +1, and sin φ _Vi can take from −1 to +1. Therefore, paying attention to the similarity between the Ising model and the laser system, if σ _i = sin φ _Vi is set, Formula 19 becomes Formula 20:

When Equation 20 is added to all M sites, Equation 21 is obtained, which is expressed as the overall threshold gain ΣE _CVi of the laser system.

Here, since mutual injection is performed between the laser pulses P, and one injection is performed from the master pulse P0 to each laser pulse P, A _Vi (t) to A _Vj (t) to √n _M You can. Then, in the above definition of σ _i = sin φ _Vi , since σ _i = ± 1, φ _Vi to φ _Vj to ± π / 2 can be set. At this time, Expression 21 becomes Expression 22.

Here, when the harmonic mode-locked laser has a uniform medium, the oscillation phase state {σ _i } that realizes the minimum threshold gain ΣE _CVi is selected as the entire laser system. That is, one specific oscillation mode is selected for the entire laser system. Then, due to the competition between the oscillation modes, one specific oscillation mode suppresses the other oscillation modes.

That is, as a whole laser system, ΣE _CVi of Expression 22 is minimized. On the other hand, (ω / Q) M in Formula 22 is constant for the entire laser system. Therefore, Ση _i σ _i + Σξ _ij σ _i σ _j in Expression 22 is minimized for the entire laser system.

When ξ _ij = J _ij and η _i = λ _i , as a whole laser system, when Ση _i σ _i + Σξ _ij σ _i σ _j is minimized, Σλ _i σ _i + ΣJ _ij σ _i σ _j is also minimized. Is done. That is, the ground state that minimizes the Ising Hamiltonian of Expression 13 is realized.

In order to improve the calculation accuracy, the difference between the minimum threshold gain and the next smallest threshold gain for the laser oscillation mode of the entire laser system is determined by the saturation gain E _CV and the photon attenuation rate ω determined by the spontaneous emission rate. It is necessary to make it sufficiently larger than β (ω / Q) (1 / R) which is the difference of / Q. Here, R = I / I _th −1 is a normalized pump rate, and I and I _th are an injection current and a laser oscillation threshold value, respectively. Therefore, calculation accuracy can be improved by reducing β and increasing R.

As described with reference to FIGS. 10 and 11, since the plurality of laser pulses P have the same oscillation frequency, the initial oscillation phase φ _Vi (t = 0) of each laser pulse P is determined by each laser pulse. Of the two oscillation phases φ _Vi (steady state) = ± π / 2 in the steady state of P, there is no such thing as being close to one phase and far from the other phase. Therefore, a read error can be prevented in the Ising model quantum computing device.

  In the first and second prior arts, the master laser M is used for the purpose of injection locking to the surface emitting laser V, whereas in this embodiment, Ising is used for the purpose of mounting Zeeman energy. A master pulse P0 corresponding to the DC magnetic field of the model is used. Furthermore, in the first and second prior arts, as many surface emitting lasers V as the number of Ising sites are required, in this embodiment, it is sufficient to prepare only one harmonic mode-locked laser. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

  FIG. 12 shows a first configuration of the Ising model quantum computation device according to the second embodiment. In the configuration of FIG. 12, the harmonic mode-locked laser is a passively mode-locked fiber laser.

  The passive mode-locked fiber laser is the same as in FIG.

  FIG. 13 shows a second configuration of the Ising model quantum computation device according to the second embodiment. In the configuration of FIG. 13, the harmonic mode-locked laser is a forced mode-locked fiber laser.

  The forced mode-locked fiber laser is the same as in FIG.

  12 and 13, the method of oscillating a series of a plurality of laser pulses P is different, but the method of implementing Ising interaction and the method of measuring Ising spin are the same. Here, it is assumed that (M + 1) laser pulses P corresponding to M Ising sites and DC magnetic fields are arranged at equal intervals (time interval τ) in the ring resonator of the harmonic mode-locked laser.

  The Ising interaction mounting unit and the Zeeman energy mounting unit are configured by amplitude phase modulators AP1, AP2,..., AP (M) and fiber delay lines DL1, DL2,.

The set of the amplitude phase modulator AP1 and the fiber delay line DL1 is similar to FIGS. 6 and 7 in that the magnitude and sign of the pseudo Ising interaction J _ij between two laser pulses P whose time interval is τ. And the magnitude and sign of the pseudo Zeeman energy λ _i in the laser pulse P whose time interval with the master pulse P0 is τ.

  The set of the amplitude phase modulator AP1 and the fiber delay line DL1 generates a delay time τ, and a part of the master pulse P0 is inputted, and a part of the inputted master pulse P0 is amplitude phase modulated and amplitude phase modulated. A part of the master pulse P0 is combined with the laser pulse P whose time interval with the master pulse P0 is τ.

Here, the set of the amplitude phase modulator AP1 and the fiber delay line DL1 is shared for any two pairs of laser pulses P having a time interval of τ and the pair of master pulses P0 and laser pulses P. In this way, the set of the amplitude phase modulator AP1 and the fiber delay line DL1 implements the magnitude and sign of the pseudo Zeeman energy λ _i in the laser pulse P whose time interval with the master pulse P0 is τ.

The set of the amplitude phase modulator AP2 and the fiber delay line DL2 is similar to FIGS. 6 and 7 in that the magnitude and sign of the pseudo Ising interaction J _ij between two laser pulses P whose time interval is 2τ. And the magnitude and sign of the pseudo Zeeman energy λ _i in the laser pulse P whose time interval with the master pulse P0 is 2τ.

  A set of the amplitude phase modulator AP2 and the fiber delay line DL2 generates a delay time 2τ, and a part of the master pulse P0 is inputted, and a part of the inputted master pulse P0 is amplitude phase modulated and amplitude phase modulated. A part of the master pulse P0 is combined with the laser pulse P whose time interval with the master pulse P0 is 2τ.

Here, the set of the amplitude phase modulator AP2 and the fiber delay line DL2 is shared for any two pairs of laser pulses P and a pair of master pulses P0 and laser pulses P whose time interval is 2τ. In this way, the set of the amplitude phase modulator AP2 and the fiber delay line DL2 implements the magnitude and sign of the pseudo Zeeman energy λ _i in the laser pulse P whose time interval with the master pulse P0 is 2τ.

  The set of the amplitude phase modulator AP (M) and the fiber delay line DL (M) is the same as the set of the amplitude phase modulator AP1 and the fiber delay line DL1 and the set of the amplitude phase modulator AP2 and the fiber delay line DL2. Holds.

  The processing of the Ising model quantum computation device of the second embodiment is shown in FIGS. Here, it is assumed that four laser pulses P corresponding to three Ising sites and a DC magnetic field are arranged at equal intervals (time interval τ) in the ring resonator of the harmonic mode-locked laser.

-Arg _{(J ij)} is the same as FIG. 9.

-Arg (λ _j ) represents phase lag modulation if λ _j is positive and represents phase advance modulation if λ _j is negative. That is, if λ _j is positive, an oscillation mode in which the oscillation phase of each laser pulse P is −π / 2 is likely to rise, and if λ _j is negative, the oscillation phase of each laser pulse P is + π. The oscillation mode that is / 2 is made to rise easily.

First, in the upper part of FIG. 14, the master pulse P0 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Zeeman energies λ ₁ , λ ₂ , and λ ₃ are mounted.

A part of the master pulse P0 is input to the fiber delay line DL1, is subjected to amplitude modulation proportional to | λ ₁ | and phase modulation of −arg (λ ₁ ) by the amplitude phase modulator AP1, and is received by the fiber delay line DL1. The signal is output from the fiber delay line DL1 after a delay time τ. Part of the master pulse P0 output from the fiber delay line DL1 is combined with the laser pulse P1 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Zeeman energy λ ₁ is implemented.

A part of the master pulse P0 is input to the fiber delay line DL2, is subjected to amplitude modulation proportional to | λ ₂ | and phase modulation of −arg (λ ₂ ) by the amplitude phase modulator AP2, and is received by the fiber delay line DL2. The signal is output from the fiber delay line DL2 after a delay time 2τ. A part of the master pulse P0 output from the fiber delay line DL2 is combined with the laser pulse P2 being input to the fiber delay lines DL1, DL2, and DL3. In this way, the magnitude and sign of the Zeeman energy λ ₂ is implemented.

A part of the master pulse P0 is input to the fiber delay line DL3, and is subjected to amplitude modulation proportional to | λ ₃ | and phase modulation of −arg (λ ₃ ) by the amplitude phase modulator AP3, and is transmitted by the fiber delay line DL3. The signal is output from the fiber delay line DL3 after a delay time 3τ. Part of the master pulse P0 output from the fiber delay line DL3 is combined with the laser pulse P3 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Zeeman energy λ ₃ is implemented.

Next, in the lower part of FIG. 14, the laser pulse P1 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Ising interactions J ₁₂ and J ₁₃ are implemented.

A part of the laser pulse P1 is input to the fiber delay line DL1, and is subjected to amplitude modulation proportional to | J ₁₂ | and phase modulation of −arg (J ₁₂ ) by the amplitude phase modulator AP1, and the fiber delay line DL1. The signal is output from the fiber delay line DL1 after a delay time τ. Part of the laser pulse P1 output from the fiber delay line DL1 is combined with the laser pulse P2 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Ising interaction J ₁₂ is mounted.

A part of the laser pulse P1 is input to the fiber delay line DL2, is subjected to amplitude modulation proportional to | J ₁₃ | and phase modulation of −arg (J ₁₃ ) by the amplitude phase modulator AP2, and is transmitted by the fiber delay line DL2. The signal is output from the fiber delay line DL2 after a delay time 2τ. A part of the laser pulse P1 output from the fiber delay line DL2 is combined with the laser pulse P3 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Ising interaction J ₁₃ is mounted.

In the lower part of FIG. 14, the amplitude / phase modulator AP3 blocks light. In order to mount the pseudo Zeeman energy λ ₁ with the laser pulse P1, it is sufficient to inject light from the master pulse P0 to the laser pulse P1, and light is injected from the laser pulse P1 to the master pulse P0. It is not necessary.

Next, in the upper part of FIG. 15, the laser pulse P2 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Ising interactions J ₂₃ and J ₂₁ are implemented.

A part of the laser pulse P2 is input to the fiber delay line DL1, and is subjected to amplitude modulation proportional to | J ₂₃ | and phase modulation of −arg (J ₂₃ ) by the amplitude phase modulator AP1, and the fiber delay line DL1. The signal is output from the fiber delay line DL1 after a delay time τ. A part of the laser pulse P2 output from the fiber delay line DL1 is combined with the laser pulse P3 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Ising interaction J ₂₃ are mounted.

A part of the laser pulse P2 is input to the fiber delay line DL3, and is subjected to amplitude modulation proportional to | J ₂₁ | and phase modulation of −arg (J ₂₁ ) by the amplitude phase modulator AP3, and is transmitted by the fiber delay line DL3. The signal is output from the fiber delay line DL3 after a delay time 3τ. Part of the laser pulse P2 output from the fiber delay line DL3 is combined with the laser pulse P1 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Ising interaction J ₂₁ is mounted.

In the upper part of FIG. 15, the amplitude / phase modulator AP2 blocks light. In order to implement the pseudo Zeeman energy λ ₂ with the laser pulse P2, it is only necessary to inject light from the master pulse P0 to the laser pulse P2, and light is injected from the laser pulse P2 to the master pulse P0. It is not necessary.

Next, in the lower part of FIG. 15, the laser pulse P3 is input to the amplitude phase modulator and the fiber delay line, and the magnitudes and signs of the Ising interactions J ₃₁ and J ₃₂ are implemented.

A part of the laser pulse P3 is input to the fiber delay line DL2, and is subjected to amplitude modulation proportional to | J ₃₁ | and phase modulation of −arg (J ₃₁ ) by the amplitude phase modulator AP2, and is transmitted by the fiber delay line DL2. The signal is output from the fiber delay line DL2 after a delay time 2τ. A part of the laser pulse P3 output from the fiber delay line DL2 is combined with the laser pulse P1 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Ising interaction J ₃₁ is mounted.

A part of the laser pulse P3 is input to the fiber delay line DL3, and is subjected to amplitude modulation proportional to | J ₃₂ | and phase modulation of −arg (J ₃₂ ) by the amplitude phase modulator AP3. The signal is output from the fiber delay line DL3 after a delay time 3τ. A part of the laser pulse P3 output from the fiber delay line DL3 is combined with the laser pulse P2 being input to the fiber delay lines DL1, DL2, and DL3. Thus, the magnitude and sign of the Ising interaction J ₃₂ are mounted.

In the lower part of FIG. 15, the amplitude / phase modulator AP1 blocks light. In order to implement the pseudo Zeeman energy λ ₃ with the laser pulse P3, it is only necessary to inject light from the master pulse P0 to the laser pulse P3, and light is injected from the laser pulse P3 to the master pulse P0. It is not necessary.

  The processes in the upper and lower stages in FIG. 14 and the processes in the upper and lower stages in FIG. 15 are repeated in the above-described order until a steady state is reached in the process in which the laser pulses P1, P2, and P3 are exchanged and injected.

  As described with reference to FIGS. 14 to 16, instead of preparing an Ising interaction mounting unit for each pair of laser pulses P, Ising interaction mounting is performed for a plurality of pairs of laser pulses P having the same time interval. It is enough to share the department. That is, when the number of Ising sites is M, the first and second prior arts require M (M-1) / 2 Ising interaction mounting units, whereas in this embodiment, the Ising site is Ising sites. It is not necessary to have M (M−1) / 2 interaction mounting units. In addition, it is sufficient to prepare the Zeeman energy mounting unit in common with the Ising interaction mounting unit, instead of preparing the Zeeman energy mounting unit separately from the Ising interaction mounting unit.

  In particular, M laser pulses P corresponding to M Ising sites and one master pulse P0 corresponding to the DC magnetic field of the Ising model are equally spaced (time intervals) in the ring resonator of the harmonic mode-locked laser. τ), the time intervals between different pulses P are M types of τ,..., Mτ, so the Ising interaction implementation unit and the Zeeman energy implementation unit (if these are prepared in common) It is sufficient to prepare only M types. Therefore, the circuit configuration can be simplified in the Ising model quantum computing device.

In FIGS. 14 to 16, Ising interactions J _ij and J _ji are mounted at different timings. Therefore, in this embodiment, not only a normal Ising model in which J _ij = J _ji but also a general Ising model in which J _ij ≠ J _ji can be solved.

  12 and 13, the Ising spin measurement unit is the same as in FIGS. 6 and 7.

As described with reference to FIGS. 12 and 13, since an optical fiber is used as a transmission path for a plurality of laser pulses P, an Ising model quantum computing device with low optical loss, high optical intensity, and a simple circuit configuration is provided. Can be provided. Here, the time during which the optical path length of the optical fiber is stably maintained without being affected by mechanical vibration or thermal expansion is 10 ⁻⁴ s, and the quantum computer of the Ising model can measure Ising spin. Since the required time is 10 ⁻⁹ s, the measurement result of Ising spin is not affected by mechanical vibration or thermal expansion of the optical fiber.

  The Ising model quantum computing device of the present invention is suitable for solving a NP complete problem mapped to the Ising model at high speed and easily.

M: Master lasers V, V1, V2, V3: Surface emitting lasers P0: Master pulses P, P1, P2, P3: Laser pulses I12, I13, I23: Ising interaction mounting parts Z1, Z2, Z3: Zeeman energy mounting parts AP1, AP2, AP3, AP (M-1), AP (M): Amplitude / phase modulators DL1, DL2, DL3, DL (M-1), DL (M): Fiber delay lines B1, B2: Beam splitter R1 R2: Reflector DI, DO: Detector 1: Pump laser 2: WDM coupler 3: Erbium-doped fiber 4: 1/4 wavelength plate 5: 1/2 wavelength plate 6: Beam splitter 7: 1/4 wavelength plate 8 : Dispersion shift fiber 9: coupler 10: isolator 11: modulator 12: filter 13: coupler 14: clock extractor 15: phase controller 6: amplifier

Claims (8)

A harmonic mode-locked laser that oscillates multiple laser pulses with the same oscillation frequency, corresponding to multiple sites of the Ising model,
By controlling the amplitude and phase of light exchanged between two laser pulses for each pair of the plurality of laser pulses, the magnitude and sign of the pseudo Ising interaction between the two laser pulses An Ising interaction implementation that implements
Ising spin measurement that measures the pseudo Ising spin of the plurality of laser pulses by measuring the oscillation phase of the plurality of laser pulses after the laser pulses reach a steady state in the process of exchanging light. And
An Ising model quantum computing device comprising: The Ising interaction implementation unit generates a delay time equal to a time interval between the two laser pulses, and receives a part of one of the two laser pulses as an input, A delay modulation unit for mounting an Ising interaction that amplitude-phase modulates a part of the laser pulse and combines a part of the one laser pulse that has been amplitude-phase modulated with the other laser pulse of the two laser pulses. ,
The delay modulation unit for mounting the Ising interaction shares a single set of delay lines and amplitude phase modulators for a plurality of pairs of laser pulses having the same time interval among the plurality of laser pulses. The Ising model quantum computation device according to claim 1. A harmonic that oscillates a plurality of laser pulses having the same oscillation frequency corresponding to a plurality of sites of the Ising model and a single laser pulse corresponding to the DC magnetic field of the Ising model and having the same oscillation frequency. A wave mode-locked laser,
By controlling the amplitude and phase of light exchanged between two laser pulses for each pair of the plurality of laser pulses, the magnitude and sign of the pseudo Ising interaction between the two laser pulses An Ising interaction implementation that implements
For each of the plurality of laser pulses, by controlling the amplitude and phase of light injected from the single laser pulse, the magnitude and sign of the pseudo Zeeman energy in each of the plurality of laser pulses can be obtained. The Zeeman Energy Implementation Department to implement,
The pseudo Ising spin of the plurality of laser pulses is measured by measuring the oscillation phase of the plurality of laser pulses after reaching a steady state in the process in which the plurality of laser pulses are exchanged and injected. Ising spin measurement unit,
An Ising model quantum computing device comprising: The Ising interaction implementation unit generates a delay time equal to a time interval between the two laser pulses, and receives a part of one of the two laser pulses as an input, A delay modulation unit for mounting an Ising interaction that amplitude-phase modulates a part of the laser pulse and combines a part of the one laser pulse that has been amplitude-phase modulated with the other laser pulse of the two laser pulses. ,
The Zeeman energy implementation produces a delay time equal to the time interval between each of the single laser pulse and the plurality of laser pulses, and a portion of the single laser pulse is input and input A Zeeman energy mounting delay modulation unit that amplitude-phase modulates a part of the single laser pulse and combines the amplitude-phase modulated part of the single laser pulse with each of the plurality of laser pulses; ,
The Ising interaction mounting delay modulation unit and the Zeeman energy mounting delay modulation unit include a delay line and a plurality of pairs of laser pulses having the same time interval among the plurality of laser pulses and the single laser pulse. 4. The Ising model quantum computation device according to claim 3, wherein a single set of amplitude and phase modulators is shared.   The Ising spin measurement unit delay-detects whether the oscillation phase of different laser pulses among the plurality of laser pulses is in-phase or anti-phase, so that pseudo Ising spins of the different laser pulses are in the same direction. The Ising model quantum computation device according to any one of claims 1 to 4, wherein the quantum computation device measures either the reverse direction or the reverse direction.   6. The Ising model quantum computation device according to claim 1, wherein the harmonic mode-locked laser is a passive mode-locked laser.   6. The Ising model quantum computation device according to claim 1, wherein the harmonic mode-locked laser is a forced mode-locked laser. The harmonic mode-locked laser is a mode-locked fiber laser,
8. The Ising model quantum computation device according to claim 1, wherein the Ising interaction mounting unit includes a fiber delay line. 9.

Images