TECHNICAL FIELD

The embodiments relate generally to computing and more specifically to optical computing.

BACKGROUND

Recent years have seen interest in the possibilities in optical computing. An optical computer is a computer which performs its computation with photons produced by lasers or diodes, as opposed to with the electrons as in conventional computers. Despite of the interest, a general-purpose optical computer has re- mained an elusive idea. One of the challenges has been the nature of photons that do not interact with each other in vacuum but require a medium of some kind. Also, many optical nonlinear phenomena, such as four-wave mixing, require a high power pump to control a low power idler, which is an exact opposite to the oper- ating principle of a transistor. Moreover, in many realizations of an optical logic the wavelength of the input differs from the output, which complicates design of cas- caded systems.

One of the components required for realizing an all-optical computer is an all-optical memory. There is a need for realizing such an all-optical memory which would be both fast and easily integrable (that is, realizable also as a photonic integrated circuit).

BRIEF DESCRIPTION

According to an aspect, there is provided the subject matter of the in- dependent claims. Embodiments are defined in the dependent claims. The scope of protection sought for various embodiments is set out by the independent claims.

The embodiments and features, if any, described in this specification that do not fall under the scope of the independent claims are to be interpreted as examples useful for understanding various embodiments of the invention.

BRIEF DESCRIPTION OF DRAWINGS

In the following, example embodiments will be described in greater de- tail with reference to the attached drawings, in which

Figure 1A shows a schematic drawing of the basic operating principle of a laser-based optical memory unit according to embodiments; Figure 1B illustrates a process of writing data to an optical memory unit according to an embodiment;

Figures 2A, 2B and 2C illustrate a process of writing data to an optical memory unit according to embodiments;

Figure 3 illustrates a process of writing data to an optical memory unit according to embodiments;

Figure 4 illustrates exemplary simulation results for a process of writ- ing data to an optical memory unit according to embodiments;

Figures 5 A, 5B, 6, 7, 8 and 9 illustrate different implementations of the optical memory unit according to embodiments; and

Figures 10A and 10B illustrate a process of writing data to an optical memory unit according to an alternative, polarization-based embodiment.

DETAILED DESCRIPTION OF SOME EMBODIMENTS

The following embodiments are only presented as examples. Although the specification may refer to "an", "one”, or "some" embodiment(s) and/or exam- ple(s) in several locations of the text, this does not necessarily mean that each ref- erence is made to the same embodiment(s) or example (s), or that a particular fea- ture only applies to a single embodiment and/or example. Single features of differ- ent embodiments and/or examples may also be combined to provide other embod- iments and/or examples.

In the following, the terms "amplitude" and "power" (or "optical power” or "intensity") are used sometimes in the same context It should be understood that the power P is proportional to a square norm of the complex electric field am- plitude E, such that

The optical memory units and optical memories to be discussed below comprise multiple lasers. The embodiments are not limited to any particular laser type or types unless explicitly stated otherwise. For example, any of the lasers dis- cussed in connection with embodiments may be lasers of one or more of the fol- lowing types: a semiconductor laser (e.g., a vertical cavity surface-emitting laser (VCSEL) or an edge emitting laser), a gas laser (e.g., HeNe laser), a liquid laser, a solid-state laser (e.g., Nd:YAG laser), a plasmonic laser , a Raman laser, a laser using chemical pumping, a laser using electrical pumping, a linear-cavity laser (i.e., a Fabry-Perot cavity laser), a ring laser, a disk laser and a nano-scale laser other than a plasmonic laser. A plasmonic laser is a type of laser which aims to confine light at a subwavelength scale far below the Rayleigh's diffraction limit of light, by storing some of the light energy through electron oscillations called surface plasmon po- laritons. Due to the role of the surface plasma excitations in their operation, they are sometimes called SPASERs (Surface Plasmon Amplification by Stimulated Emission of Radiation).

In the Figures 5A, 5B and 6 to 8 to be discussed below in detail, dashed line is used for indicating optional features such as an optional optical input or out- put

Figure 1A illustrates an optical memory unit 100 according to embodi- ments. The optical memory unit serves as the fundamental building block of an op- tical memory. The optical memory unit may, thus, be equally called an optical memory cell. Figure 1B illustrates a simplistic flow chart for a process of writing data to the optical memory unit 100.

The optical memory unit 100 comprises a first laser 101 ("L1") and a second laser 102 ("L2"). The first and second lasers 101, 102 may be specifically first and second slave lasers. A slave laser is a laser which emits on an optical fre- quency which is dictated by an external master laser (or a seed laser) via injection locking. Typically (but not always), the master laser is a low-noise single-frequency laser, and the slave laser is a high-power laser (i.e., providing higher power at least relative to the master laser). The free-running frequency (or one of the free-run- ning frequencies) of the slave laser should be relatively close to the frequency of the master laser (i.e., within certain pre-defined limits relative to the frequency of the master laser) to allow for injection (or frequency) locking. A free-running fre- quency is defined as a frequency of the slave laser when it is used without injection locking. Injection locking makes it easier to achieve good noise properties, compar- ing with the attempt of directly obtaining low-noise performance of a high-power laser.

During injection locking, the frequency and phase of the injection locked slave laser is locked to the frequency and phase of the master laser, but the ampli- tude of the injection locked slave laser is not. The output of an injection-locked slave laser may be written (at least approximately) as where x and y are complex amplitudes of the input electric field of the slave laser originating from the master laser and the output electric field of the slave laser, respectively, p is a dimensionless amplification factor determined by the proper- ties of the particular slave laser and the notation (... )° denotes the normalization operation. Thus, the slave laser may be interpreted as performing, during the in- jection locking, a normalization operation (or at least an approximation of a nor- malization operation) to the electric field received from the master laser.

Specifically, the first slave laser 101 is configured to output a first laser beam having a first frequency when injection-locked with a first (external) injec- tion laser beam 103 having the first frequency. In other words, the first slave laser 101 has a free-running frequency of which is sufficiently close to said first fre- quency to enable injection locking at that frequency. The phase of the first laser beam is also locked to the phase of the first injection laser beam (due to the funda- mental nature of injection locking). Both the first laser beam and the first injection laser beam 103 may have the same (first) polarization, at least in some embodi- ments. In other embodiments, the first laser beam and the first injection laser beam 103 may not necessarily always have the same polarization (e.g., in the polariza- tion-based embodiments discussed in connection with Figures 10A and 10B). The first frequency may be specifically an optical frequency and/or the first polariza- tion may be a linear polarization (or circular or elliptical polarization). The first injection laser beam 103 may be provided by a (first) reference master laser (not shown in Figure 1A). Said reference master laser may or may not form a part of the optical memory unit. As described above, a single reference master laser may be used, in some embodiments, for driving multiple optical memory units (i.e., multi- ple pairs of first and second slave lasers) simultaneously.

The second slave laser 102 is configured to output a second laser beam having the first frequency when injection-locked with a second (external) injection laser beam 104 having the first frequency. The phase of the second laser beam is also locked to the phase of the second injection laser beam (due to the fundamental nature of injection locking). The second injection laser beam 104 may be provided by said first reference master laser or by a second reference master laser (not shown in Figure 1A). In some embodiments, the second laser beam and the second injection laser beam 104 may also have said first polarization. In other embodi- ments, the second laser beam and the second injection laser beam 104 may not necessarily always have the same polarization (e.g., in the polarization-based em- bodiments discussed in connection with Figure 10A and 10B).

In some embodiments, the second slave laser 102 may be configured to output a second laser beam having the first frequency and a second polarization non-orthogonal to the first polarization when injection-locked with a second (ex- ternal) injection laser beam 104 having the first frequency and the first (or second) polarization. In other words, the second laser beam has at least a non-zero compo- nent corresponding to the first polarization to enable interaction between the first and second slave lasers 101, 102.

In some embodiments, the first and second slave laser 101, 102 may be specifically lasers of the same type and/or the same make and/or the same model. The first and second slave lasers 101, 102 may have the same (physical) properties. The first and second slave lasers 101, 102 may operate, in the optical memory unit, in a symmetrical manner (assuming symmetrical injection).

The first and second slave lasers 101, 102 are configured to be mutually coupled (when injection locked). In other words, the first (output) laser beam pro- vided by the first slave laser 101 is coupled to the second slave laser 102 and the second (output) laser beam provided by the second slave laser 102 is coupled to the first slave laser 101. It should be noted that while the beams outputted by the first and second slave lasers 101, 102 are shown as two spatially separated beams for clarity of presentation, they may be substantially coincident The coupling strength between the first and second slave lasers 101, 102 may be the same as the coupling strength between the first slave laser 101 and the first reference master laser and/or the second slave laser and the first reference master laser or the sec- ond slave laser and the second reference master laser (whichever is driving the second slave laser).

The first and second slave lasers 101, 102 may be mutually coupled to each other during the injection locking so that the electromagnetic fields inside the two laser cavities of the first and second slave lasers 101, 102 are coupled such that their phases are locked. That is, the difference of the phases does not change. In some alternative embodiments, the electromagnetic fields inside the two laser cav- ities of the first and second slave lasers 101, 102 are coupled such that the phase difference varies periodically. The phase difference is not chaotic, however, as in that case the first and second slave lasers 101, 102 would not be locked at all.

The mutual coupling may be achieved in many different alternative ways such as coupling over air or free space (optionally via lenses and/or other optical components), coupling via evanescent waves in a waveguide, coupling via multimode interference filter or coupling via a small space between the two laser cavities that allows the intracavity evanescent fields to interact with laser. It is also known in the art that adjacent laser media are able share part of the intracavity fields with the neighboring laser. Some examples of coupling mechanisms are dis- cussed in more detail in relation to Figures 5A, 5B and 6 to 9. The optical memory unit 100 is configured so that an optical path length between the first and second slave lasers 101, 102 corresponds to a phase shift (φ ) substantially equal to it (i.e., 180°) at the first frequency (i.e., the injection-locking frequency). Specifically, said optical path length may correspond (substantially) toφ = π + 2πn, where n may be any integer. Such a configuration may be realized in a variety of different ways. In some embodiments, the first and second slave lasers 101, 102 may be simply arranged in the optical memory unit 100 to have a spacing resulting in said desired phase shift. Additionally or alternatively, at least one (tun- able) phase-shifting element (e.g., a phase shifter) may be arranged in an optical path between the first and second slave lasers 101, 102 for enabling accurate con- trol of the phase shift φ (so as to accurately implement said phase shift substan- tially equal to it at the first frequency). By changing the phase shift induced by the at least one tunable phase-shifting element, the phases of the first and second laser beams outputted by the first and second slave lasers 101, 102 (and thus the data maintained in the optical memory unit) may be changed. Additionally or alterna- tively, at least one phase shifting element may be integrated into the laser cavity of the first and/or second slave lasers 101, 102.

In the following, the operation of the optical memory unit 100 as a phase-based memory is discussed briefly in connection with Figure 1B. A more de- tailed description of one exemplary method of writing data to the optical memory unit 100 is provided in connection with Figures 2A, 2B and 2C. In general, the op- eration is analogous with a flip-flop circuit as commonly used in electronics. Spe- cifically, operation of a phase-based optical memory unit (i.e., an optical memory unit where data is written to the phases of the first and second laser beams output- ted by the first and second slave lasers 101, 102) is discussed in the following. An alternative embodiment describing a polarization-based optical memory unit (i.e., an optical memory unit where data is written to the polarizations of the first and second laser beams outputted by the first and second slave lasers 101, 102) is dis- cussed prior to and in connection with Figures 10A and 10B.

The first and second injection laser beams 103, 104 are, in block 111, injected to the firstand second slave lasers 101, 102 for achieving injection locking. Preferably, the first and second injection laser beams 103, 104 have, at the respec- tive first and second slave lasers 101, 102 (i.e., inside their respective cavities), equal amplitude and phase (assuming that the first and second slave lasers have the same or similar physical properties which is usually but not necessarily always the case). This enables maintaining, in block 111, at a time, one of two possible steady states of the optical memory unit 100. Specifically, the two steady states comprise a first steady state associated with a first pair of values of phases of the first and second slave lasers 101, 102 and a second steady state associated with a second pair of values of the phases of the first and second slave lasers 101, 102. The first pair of values of the phases of the first and second slave lasers 101, 102 (i.e., the first steady state) corresponds to (+π/3, -π/3) and the second pair of val- ues of the phases of the first and second slave lasers 101, 102 (i.e., the second state state) corresponds to (+π/3, -π/3 ). In other words, the electric fields outputted by the first and second slave lasers 101, 102 have the forms E ₁ = e ^iπ/3 E _ref and E ₂ = e ^-iπ/3 E _ref in the first steady state and forms E ₁ = e ^-iπ/3 E _ref and E ₂ = e ^iπ/3 E _ref in the second steady state, respectively, where E _ref,1 and E _ref,2 are electric field vectors being in phase with the first and second injection laser beams 103, 104, respectively, and E ₁ and E ₂ are the electric field vectors outputted by the first and second slave lasers 101, 102. The optical memory unit 100 may, thus, be called bistable.

It should be emphasized that the optical memory unit 100 having the two steady states (+π/3, -π/3 ) and (+π/3, -π/3 ) is a fundamental physical prop- erty of an optical memory unit 100 operating in a symmetrical manner resulting from the fact the phase shift between the mutually coupled first and second slave lasers 101, 102 is substantially equal to π and that the amplitudes, phases (and also polarizations assuming no elements affecting the polarization are arranged in the optical path between the first and second slave lasers 101, 102) of the first and second laser beams outputted by the first and second slave laser 101, 102 are equal (i.e., it is not an arbitrary selection). The result that the two steady states are spe- cifically (+π/3, -π/3 ) and (+π/3, -π/3 ) may be derived by solving the following set of equations describing the optical memory unit 100 of Figure 1A for E ₁ and E ₂ for steady state operation (with E _ref,1 = E _ref,2 = E _ref ):

Here, E ₁ , E ₂ , E _ref,1 , and E _ref,2 have fixed phase (for sake of simplicity but without loss of generality assumed to be set to zero here) and the same amplitude. Thus, only the phases of E ₁ and E ₂ (relative to E _ref,1 / E _ref,2 ) are unknown in said equa- tions.

With that said, in some embodiments, the optical memory unit 100 may not satisfy all of said conditions for the bistable operation, at least not exactly or fully, resulting in imperfect but still functional operation. Also, in some embodi- ments, the first and second injection laser beams 103, 104 during steady state op- eration may not necessarily have, at the respective first and second slave lasers 101, 102, equal amplitude and phase as described above, e.g., due to differing phys- ical properties of the first and second slave laser 101, 102 or other asymmetry in the optical memory unit 100. In general, said injection-locking of the first and sec- ond slave lasers 101, 102 with the first and second injection laser beams 103, 104 may enable maintaining, at a time, one of two steady states of the optical memory unit when 1) phases of the first and second injection laser beams are locked to each other (i.e., E _ref,1 , and E _ref,2 have the same phase) and 2) amplitudes of the first and second injection laser beams 103, 104 are selected so that amplitudes of the first and second injection laser beams at the first and second slave lasers 101, 102 (i.e., inside the cavity of the first and second slave laser 101, 102), respectively, are equal to the amplitudes of the second and first laser beams at the first and second slave lasers. In other words, in the condition 2), the amplitude of the first injection laser beam at the first slave laser 101 should be equal to the amplitude of the second laser beam (which is outputted by the second slave laser 102) at the first slave laser 101 (i.e., the two amplitudes are evaluated at the same location) and correspond- ingly the amplitude of the second injection laser beam at the second slave laser 102 should be equal to the amplitude of the first laser beam (which is outputted by the first slave laser 101) at the second slave laser 102. In such a case, the switching between the two steady states may be still enabled by introducing a perturbation component to at least one of the first and second injection laser beams during steady state operation, as discussed above. In general, the perturbation component may correspond to a perturbation of at least one of amplitude, phase and polariza- tion.

Moreover, it should be noted that above it was assumed that the phases outputted by the first and the second slave laser 101, 102 stay locked in respect to each other. Also, the state-of-polarization of the outputted beam from the first slave laser 101 is the same as polarization outputted by the second slave laser 102 when the second laser beam is observed at the face of the first slave laser 101.

The memory may be “read" by measuring the phases of output beams of the first and second slave lasers 101, 102 or phase of at least one of them. In some embodiments, at least one optical element such as a beam splitter/combiner may be provided in the optical path between the first and second slave lasers 101, 102 for guiding a part of the first and second laser beam(s) outputted by the first and/or second slave lasers 101, 102 away from the optical memory unit 100 and thus enabling the memory to be read more easily. Switching between the two steady states (i.e., a write event) is enabled by introducing or adding, in block 112, a perturbation component (or equally a per- turbation signal or laser beam) to at least one of the first and second injection laser beams 103, 104 (that is, to at least one laser beam for maintaining a steady state). In other words, an amplitude and phase perturbation laser beam may be combined with at least one of the first and second injection laser beams 103, 104 before in- jection to the first and/or second slave laser 101, 102. Here, the perturbation com- ponent may be specifically an amplitude and phase perturbation component though other types of perturbations may also be employed in other embodiments as mentioned above (see also discussion in connection with Figure 10A).

The amplitude and phase perturbation component may have a certain pre-defined duration. The amplitude and phase perturbation component may be specifically a short high-intensity laser beam originating from an external (laser) source. The peak power of the amplitude and phase perturbation component may be (much) higher than the power of the first and second injection beams 103, 104 for maintaining a steady state. In general, higher the peak power of the amplitude and phase perturbation component, the faster the write event may be completed (though still being dependent on laser settling time). For example, the peak power of the amplitude and phase perturbation component may be equal to or higher than the power of the first and second injection beams 103, 104 for maintaining a steady state times two, preferably times three, even more preferably times five.

The duration of the amplitude and phase perturbation component may be at least longer than the settling time of the optical memory unit 100. The settling time of the optical memory unit 100 is primarily dependent on laser properties and the distance the light must travel between the first and second slave lasers 101, 102. Assuming semiconductor lasers with approximately 1 ns settling time and a few millimeter distance between them, it may be beneficial, for example, to have the perturbation longer than a few (e.g., two, three or four) nanoseconds (or larger than 10 ns to be on the safe side). If some other type of lasers with substantially longer settling time or much longer distance between the lasers is used, then the settling time would consequently be longer.

In some embodiments, the amplitude and phase perturbation compo- nent is introduced to both of the first and second injection laser beams 103, 104. In such embodiments, the amplitude and phase perturbation component (or specifi- cally the associated electric field) combined with one of the first and second injec- tion laser beams 103, 104 may have a complex-conjugated form relative to the am- plitude and phase perturbation component combined with the other one of the first and second injection laser beams 103, 104 (i.e., the signs of the phases of the elec- tric fields of the amplitude and phase perturbation components injected to the first and second slave lasers 101, 102 may be opposite to each other). This embodiment provides the benefit that the memory will settle faster and more reliably.

The first slave laser 101 may have a first free-running frequency while the second slave laser 102 may also have said first free-running frequency or a sec- ond free-running frequency different from the first free-running frequency. In gen- eral, a free-running frequency is defined as a frequency at which a normally driven oscillator (here, a slave laser) operates in the absence of a driving signal (here, a laser beam outputted by a master laser). The first and/or second free-running fre- quencies may correspond substantially (or approximately) to the first frequency used by the first and/or second reference master laser. Alternatively, a frequency offset may exist between the first frequency and the first free-running frequency and/or between the first frequency and the second free-running frequency. De- pending on the parameters of the first and second slave lasers 101, 102 different stable operation regions (i.e., frequency ranges or locking ranges) may be identi- fied. The first and second free-running frequencies may be chosen specifically to enable operation at one such a stable operation region.

In some embodiments, the bandwidth of the first and/or second refer- ence master laser is narrower than the bandwidth of the first and/or second slave lasers 101, 102. Especially, if semiconductor slave lasers (e.g., VCSEL) are em- ployed, bandwidth of such a semiconductor laser is relatively wide which will add noise to the signal and thus complicate potential following signal processing steps. This can be mitigated by using a narrow linewidth reference laser, such as an ex- ternal cavity diode laser, or some other laser type with narrow linewidth (as known in the art).

In addition to the elements 101, 102 illustrated in Figure 1A, the optical memory unit 100 may comprise one or more further elements. For example, the optical memory unit 100 may comprise one or more optical elements for guiding the first injection laser beam 103 to the first slave laser 101 and/or the second in- jection laser beam 104 to the second slave laser 102 (and/or outputting the first and/or second laser beams from the optical memory unit 100 for enabling reading of data of the optical memory unit 100, i.e., reading of phase of at least one of the first and second laser beams, as described above). The optical elements may be equally called optical components. Said one or more optical elements may com- prise, e.g., one or more beam splitter/combiner elements, one or more mirrors and/or one or more focusing lenses. It is obvious to any skilled person that multiple different sets of optical elements in multiple different configurations and/or ar- rangements may be employed for implementing this feature.

Additionally or alternatively, the one or more optical elements of the optical memory unit 100 may comprise, e.g., one or more optical elements selected from the following:

- polarization controlling elements: polarization controllers, polarizers, polariza- tion beam splitters, Faraday mirrors,

- phase controlling elements: phase shifters, liquid crystal spatial modulators, holographic elements,

- power controlling elements: optical attenuators, circulators, optical isolators, light traps and terminators,

- wavelength controlling elements: optical filters (bandpass, notch etc.), dichroic mirrors, diffraction gratings,

- focusing elements: lenses, prisms,

- light sources: additional lasers,

- light detecting elements: photodetectors,

- light guiding elements: optical fibers, optical waveguide structures, mirrors.

In some embodiments, the one or more optical elements are configured (and/or arranged) to form the first and second injection laser beams 103, 104 from a first input laser beam of the optical memory unit 100. This may be achieved, for example, using a beam splitter/combiner element for splitting said first input laser beam to the first and second injection laser beams 103, 104 (which may be subse- quently guided to the first and second slave laser 101, 102 possibly using one or more further optical elements). Two examples of this configuration are illustrated in and discussed in connection with Figures 5A and SB. In waveguide-based em- bodiments (see, e.g., Figures 6 and 7), a directional coupler (or other optical cou- pling means) may be used instead of a beam splitter/combiner element.

In some embodiments, the one or more optical elements are configured (and/or arranged), additionally or alternatively, to guide at least one second input laser beam of the optical memory unit 100 corresponding to at least one perturba- tion component to respective at least one of the first and second slave lasers 101, 102. Specifically, said at least one second input laser beam may be guided so as to combine it with the first and/or second injection laser beams 103, 104 (in steady state) deriving from at least one reference master laser (see, e.g., Figures 5A and 5B). In such embodiments, said one or more optical elements may comprise at least one beam splitter/combiner element arranged in an optical path of the first and second laser beams provided by the first and second slave lasers 101, 102 so as to guide said at least one second input laser beam to the first and/or second slave lasers 101, 102. Such at least one beam splitter/combiner element may also simul- taneously enable outputting of the first and second laser beams (or at least one of them) for reading data of the optical memory unit 100, as shown in Figures 5A and SB to be discussed in detail below. In other embodiments, the optical memory unit may be configured to receive at least one laser beam corresponding directly to a combination of a laser beam originating from a reference master laser and a per- turbation component

While, in some embodiments, the optical memory unit 100 may be im- plemented by relying fully on free-space transmission of light (and optionally the aforementioned one or more optical elements), in other embodiments, the connec- tions between different parts (e.g., the first and second slave lasers 101, 102) of the optical memory unit 100 may be implemented using optical waveguiding means. The optical waveguiding means may comprise a plurality of optical waveguides such as optical fibers and possibly also one or more optical components integrated into and/or connected between the plurality of optical elements. Thus, the optical memory unit 100 may also comprise one or more optical waveguides. Some exam- ples of these additional elements and their arrangement and configuration within the optical memory unit 100 are discussed in connection with exemplary imple- mentations of the optical memory unit shown in Figures 5 A, 5B, 6, 7, 8 and 9.

In some embodiments, the optical memory unit 100 may comprise at least one reference master laser for driving the first and second slave lasers 101, 102 (i.e., for outputting the first and second injection laser beams 103, 104).

In some embodiments, there is provided an optical memory comprising one or more optical memory units 100 (preferably, a plurality of optical memory units 100) and at least one reference master laser for driving said one or more op- tical memory units 100 (or said plurality of optical memory units 100). Said at least one reference master laser may be specifically used for providing the first and sec- ond injection laser beams for operating the optical memory unit(s) 100 of the op- tical memory at the steady state (i.e., in either of the two steady states). An external amplitude and phase perturbation component (or signal or beam) may be com- bined to this steady state injection laser beam using, e.g., a beam splitter. Said ex- ternal amplitude and phase perturbation component (or beam) may originate from outside the optical memory. In some embodiments, said at least one reference mas- ter laser may be isolated from the rest of the optical memory by an optical isolator to prevent first and/or second laser beams and/or the external amplitude and phase perturbation beam coupling to it. In some alternative embodiments, the optical memory may further comprise at least one perturbation laser for outputting an amplitude and phase perturbation beam corresponding to the amplitude and phase perturbation com- ponent. In such embodiments, the optical memory may be configured so that am- plitude and phase perturbation component is introduced to the first and second injection laser beams in a phase-conjugated form relative to each other.

Figures 2A, 2B and 2C illustrate a process of writing data to an optical memory unit according to embodiments. Said optical memory unit may correspond to any optical memory unit according to embodiments discussed in connection with Figure 1A (and/or to be discussed below). Specifically, Figure 2A illustrates the process as a flow chart, Figure 2B illustrates, in a schematic manner, power and phase (or phase offset relative to steady state operation) of a first injection laser beam used for driving a first slave laser of the optical memory unit during said pro- cess as a function of time and Figure 2C illustrates, in a schematic manner, complex values of a normalized electric field of the injection laser beam in the complex plane during said process. It should be noted that the phase in Figure 2B (and also in other parts of the application) corresponds specifically to time-invariant phase or phase offset (relative to phase of the electric field in a steady state), not to instan- taneous phase of the electric or magnetic field which constantly changes over time and value of which depends on frequency. In other words, the phase corresponds to a constant phase term φ , not to a sum ω t + φ , where ω is angular frequency and t is time. Moreover, the circle in Figure 2C correspond to a unit circle. In the follow- ing, Figures 2A, 2B and 2C are discussed in parallel as they describe the same pro- cess.

It should be emphasized that the process of writing data illustrated in Figures 2A, 2B and 2C is only one possible example of a process of writing data to the optical memory unit. It is obvious to any skilled person in the art that the switching between two steady states may be achieved in a variety of different ways (i.e., manipulating the injected laser beams in a variety of different ways).

Referring to Figure 2A, the optical memory unit is operated, initially, in block 201 of Figure 2A, at a first steady state (which may correspond to either of the two steady states of the optical memory unit described above). As described above, operation at the first steady state may entail injecting the first slave laser with a first injection laser beam having an amplitude with a first amplitude value and a phase with a first phase value and by injecting the second laser with the sec- ond injection laser beam having an amplitude with said first amplitude value and a phase with said first phase value. The first and second injection laser beams also have the same (linear) polarization.

Referring to Figures 2B, the operating in the first steady state indicated with time interval 201 is associated with a certain a constant power level and con- stant phase (chosen in this particular example to correspond to zero). In regard to Figure 2C, this means that the point (complex number) 201 corresponding to the first steady state in the complex plane has a certain positive real part and a zero imaginary part It should be noted that the unit circle of Figure 2C, where also the point 201 lies, is defined specifically to correspond to the steady state amplitude (that is, specifically to correspond to amplitude of the external injection at steady state operation).

The actual data writing process may be initiated by erasing, in block 202 of Figure 2A, the optical memory unit by increasing the amplitude of a first injection laser beam to match a second amplitude value. The amplitude of the first injection laser beam may, in practice, be increased by adding an amplitude and phase per- turbation component or at least a first part of the amplitude and phase perturba- tion component to the first injection laser beam, as described above. Said erasing in block 202 causes the optical memory unit to move away from the first steady state. The second amplitude value may be, e.g., equal to or larger than the first am- plitude value multiplied by two, preferably by three, even more preferably by four.

The increase in the amplitude may be specifically a stepwise increase, i.e., an increase akin to applying a step function (not a smooth continuous increase over certain time interval). This is illustrated in both Figures 2B and 2C in connec- tion with the time interval 202 and the point 202. Referring to Figure 2B, the power increases to a second power level in a stepwise manner and is maintained in said second power level for the duration of the second time interval 202 while the phase is not changed. Similarly, in Figure 2C, the real part of the complex value of the electric field increases when moving from point 201 (steady state) to a perturbed state following the arrow 202. In other embodiments, the increase in the amplitude may be carried out over a certain time interval of a certain pre-defined length in continuous manner.

Once the optical memory unit has been moved away from the first steady state, the optical memory is switched, in block 203 of Figure 2A, to the sec- ond steady state by applying a first phase shift to the first injection laser beam while reducing said amplitude of the first injection laser beam back to the first am- plitude value. In other words, data (i.e., a value of a bit) is written to the optical memory unit in block 203. The first phase shift may be specifically a phase shift of π/3. In some more general embodiments, the first phase shift may be selected to be within a certain pre-defined range comprising π/3 and defined to be close enough to π/3 so that the optical memory unit stabilizes to the π/3 operation point during the cool down period (e.g., from π/3 to π/2 or from π/6 to π/2). The change in the amplitude and phase of the first injection laser beam in block 203 may, in practice, be caused by a second part of the amplitude and phase perturbation com- ponent added to the first injection laser beam, as described above. The aforemen- tioned change in the amplitude and phase is reflected in elements 203 of Figures 2B and 2C. Specifically, Figures 2B and 2C illustrate an example where the change in the amplitude and phase is carried out over a first time interval of a first pre- defined length in continuous manner (in other embodiments, this change may be, e.g., step-like). Namely, the power level decreases and the phase increases both in a continuous manner during the "Write" stage 203 (i.e., the first pre-defined time interval mentioned above) illustrated in Figure 2B. In Figure 2C, it is shown with the arrow 203 how the complex value of the electric field is moved back to the unit circle (i.e., back to the amplitude level associated with the steady state).

Finally, the optical memory unit is "cooled down", in block 204 of Figure 2 A, by applying a second phase shift to the first injection laser beam. The second phase shift is an opposite phase shift compared to the first phase shift. For example, if the first phase shift was selected to be π/3, the second phase shift may be, thus, specifically a phase shift of -π/3. The change in the phase of the first injection laser beam in block 204 may, in practice, be caused by a third part of the amplitude and phase perturbation component added to the first injection laser beam, as described above. Specifically, Figures 2B and 2C illustrate an example where the change in the amplitude and phase is carried out over a second time interval of a second pre- defined length in continuous manner (in other embodiments, this change may be, e.g., step-like).

The first and second pre-defined lengths of the first and second time intervals may be equal. This also the case which is depicted in Figure 2C. Namely, in Figure 2B during the "Cool" stage 204 (i.e., the second pre-defined time interval mentioned above), the power level is not changed but the phase is shifted in a con- tinuous manner so as to return it back to the initial steady state phase (i.e., zero). The returning to the steady state (though to a different one of the two steady states) is best illustrated in Figure 2C, where the complex value of the electric field returns back to the initial point 201 as indicated by the arrow 204.

It should be noted that, in some embodiments, the exact form of the in- jected laser beam (i.e., its amplitude and/or phase and their variation over time) during the switching from one steady state to another may differ from the one il- lustrated in Figure 2B. For example, the phase may not necessarily be shifted line- arly in the write/ cool stages 203, 204 and/or the amplitude may not be shifted lin- early in the write stage 203.

It should also be noted that when the optical memory unit is used for the very first time after powering the optical memory unit up, the optical memory unit may be in an indeterminate, oscillatory, or chaotic state despite the external injection (not shown in Figures 2A, 2B and 2C). In such a case, a steady state may also be reached by applying an amplitude and phase perturbation signal. As described above, faster and more reliable settling of the memory may be enabled by introducing the amplitude and phase perturbation component to both of the first and second injection laser beams in phase-conjugated forms relative to each other. Figure 3 illustrates, using a flowchart similar to Figure 2A, a process accord- ing to embodiments for writing of data to the optical memory unit according to such an alternative scheme.

The basic operating principle of the optical memory unit as illustrated in a schematic manner and discussed in connection with Figures 2B and 2C applies, mutatis mutandis, also for the process of Figure 3 though obviously as faster and more reliable settling of the memory is enabled in this case, the time intervals as- sociated with blocks 302, 304 may be reduced compared to the time intervals as- sociated with elements 203, 204 of Figures 2B and 2C. Moreover, Figures 2B and 2C may specifically correspond to one of the first and second injection laser beams. The other of the first and second injection laser beams may correspond to Figure 2B but with the sign of the phase reversed (multiplied by -1) and to Figure 2C but with the sign of the imaginary part reversed.

Initially, the optical memory unit is operating, in block 301, at a first steady state. Block 301 may correspond fully to block 201 of Figure 2A.

Then, the optical memory unit is erased, in block 302, by increasing the amplitude of the first injection laser beam as well as the amplitude of the second injection laser beam to match a second amplitude value. The erasing in block 302 causes the optical memory unit to move away from the first steady state.

The optical memory is switched, in block 303, (from the perturbed state) to the second steady state by applying, over a first time interval of a first pre- defined length in continuous manner, a first phase shift to the first injection laser beam while reducing said amplitude of the first injection laser beam back to the first amplitude value and applying, over the first time interval of the first pre-de- fined length also in a continuous manner, a second phase shift to the second injec- tion laser beam while reducing said amplitude of the second injection laser beam back to the first amplitude value. The second phase shift may be specifically an op- posite phase shift to the first phase shift. Namely, if the first phase shift is a phase shift of π/3, the second phase shift is a phase shift of -π/3 and vice versa.

The optical memory unit is cooled down, in block 304, by applying, over a second time interval of a second pre-defined length in a continuous manner, the second phase shift to the first injection laser beam and the first phase shift to the second injection laser beam. This way the phases of both the first and second injec- tion laser beams are returned to the steady state phase.

In some embodiments, the process of Figures 2A, 2B and 2C or of Figure 3 may further comprise a step of reading data from the optical memory unit by measuring phase of at least one of the first and second laser beams outputted by the first and second slave lasers. This step may precede and/or follow the steps of Figure 2A or 3.

Figure 4 illustrates exemplary simulation results for an execution of the process of Figure 3 carried out for an optical memory unit of Figure 1A. Specifically, Figure 4 illustrates input signal in terms of power and phase for one of the first and second injection laser beams (top subfigure), for a first laser beam (i.e., output) of the first slave laser (middle subfigure) and for a second laser beam (i.e., an output) of the second slave laser (bottom subfigure). The power trace is shown normalized to a pre-defined value (exact power values being irrelevant for the following dis- cussion) and the phase trace is given in radians. The simulation results of Figure 4 have been calculated using equations (2), (3), (4), and (5) provided and described in the Supplementary information of journal publication "von Lerber, T., Lassas, M., Lyubopytov, V.S. et al. All-optical majority gate based on an injection-locked laser. SciRep 9, 14576 (2019). https://doi.org/10.1038/s41598-019-51025-y". The sim- ulation results illustrated in Figure 4 correspond to parameters defined in the table below.

In regard to the table, it should be especially noted that, in this simula- tion, the system is simplified such that 1) linewidth enhancement factor a is zero, 2) the first and second slave lasers L1 and L2 have the same free-running frequency and 3) the external optical injection occurs exactly at the free-running frequency of the two lasers. In practical implementations, one or more of said three simplifica- tions may not hold.

In the case illustrated in Figure 4, the high-intensity perturbation is in- jected into first slave laser L1 and with opposite sign to second slave laser L2. Fig- ure 4 shows the same four stages which were illustrated schematically also in Fig- ure 2B (namely, "Erase", "Write", "Cool" and "Hold"). Before the Erase stage, the optical memory unit is operated in a steady state with phases of the first and second slave lasers (L1 and L2) corresponding to π/3 and -π/3. Then, at the Erase stage, the memory is unstabilized (see the dashed phase curve of L1 and L2 that goes down to zero). Then, the writing and subsequently the cooldown stages follow.

With the simulation parameter values shown in the table above, the sta- ble locking can be reached although free-running frequencies of the first and sec- ond slave lasers L1 and L2 differ from the reference light frequency up to about ±0.9 GHz which is the stable locking region of the optical memory unit. Obviously, the optical memory unit should be operated in the stable locking region for ena- bling it to function as a memory according to embodiments. In other words, the optical memory unit should be operated in the simulated scenario such that the difference between the master and slave (free-running) laser frequencies is 0.9 GHz or less. Using the simulation parameters listed in the above table but varying the linewidth enhancement factor, some further stable operation points are tabu- lated in the below table. As evident, at least in current simulation with the listed parameters, the locking range narrows down with increasing linewidth enhance- ment factor.

Obviously, the values given in the above table are to be considered only exemplary. It is envisioned that using some other set of laser characteristic param- eters, or real physical lasers with suitable characteristics, the locking range may be potentially further increased compared to the locking ranges listed in the table.

It should be noted that, in some embodiments, the first and second slave lasers may be operated at slightly different frequencies. Such selection of operating frequencies may be advantageous in view of stability in some cases.

Figure 5A illustrates one possible detailed implementation of the opti- cal memory unit of Figure 1A. The definitions provided in connection with Figure 1A apply also for this more detailed embodiment (unless otherwise explicitly stated).

Referring to Figure 5A, the optical memory unit comprises (similar to Figure 1A) a first slave laser 501 and a second slave laser 502 which are mutually coupled as indicated by the left and right facing arrows in the beam connecting the first and second slave lasers 501, 502. To ensure that the phase shift between the first and second slave laser is equal to π , a phase shifting element 504 (e.g., a phase shifter) is arranged in the optical path between the first and second slave lasers 501, 502.

Additionally, the optical memory unit comprises a reference master la- ser 503 used for injecting the first and second slave lasers 501, 502 so as to achieve injection locking (and thus steady state operation of the optical memory unit). The reference master laser 503 is isolated from the other elements of the optical memory unit using an optical isolator 505. The optical isolator 505 enables light to pass (substantially) only in one direction (that is, only away from the reference master laser 503 but not towards the reference master laser 503 as indicated by arrow in the optical isolator element 505). In some embodiments, the reference master laser 503 may not form a part of the optical memory unit (i.e., it may be external to it and potentially shared by multiple optical memory units).

In the illustrated embodiment, the output beam of the master reference laser 503 is guided to the first and second slave lasers 501, 502 for injection-lock- ing using first, second and third beam splitter/combiner elements 506, 507, 508. Each of the elements 506, 507, 508 is capable of acting as a beam splitter when a single input beam is received at the element 506, 507, 508 and as a combiner when two input beams are received at the element 506, 507, 508. Specifically, the output beam of the master reference laser 503 is first split into two beams in the third beam splitter/combiner element 508, where said two beams are intended for the first and second slave lasers 501, 502, respectively. Each of said two beams output- ted by the first beam splitter/combiner element 508 are, again, split into two more beams in the first and second beam splitter/combiner elements 506, 507. One of the beams outputted by the first beam splitter/combiner element 506 is directed to the first slave laser 501 while one of the beams outputted by the second beam splitter/combiner elements 507 is directed to the second slave laser 502.

It should be noted that only the output beams of the beam splitter/com- biner elements 506, 507, 508 which are or may be relevant for the operation of the optical memory unit are shown in Figure 5A. Similarly, the arrows show only the desired beam directions (not, e.g., beam directions not relevant for the operation of the optical memory unit or which are associated with beams to be terminated by the optical isolator 505).

The first, second and third beam splitter/combiner elements 506, 507, 508 may specifically have a 50/50 split ratio, that is, the power may be split equally between the two output beams (though other split ratio may also be employed).

The optical path length from the third beam splitter/combiner element 508 to the first slave laser 501 may be (substantially) the same as the optical path length from the third beam splitter/combiner element 508 to the second slave la- ser 502. "Substantially the same" may mean here specifically sufficiently close rel- ative to given coherence time of the reference light source.

The first and second beam splitter/combiner elements 506, 507 serve also to introduce or add the amplitude and phase perturbation component or beam to the first and/or second injection laser beam. Specifically, in Figure 5, the second beam splitter/combiner element 507 is shown to receive a first amplitude and phase perturbation laser beam (or equally a first write laser, WL1) while the first beam splitter/combiner 506 may optionally receive a second amplitude and phase perturbation laser beam (or equally a second write laser, WL2). A part of the first amplitude and phase perturbation laser beam is directed by the second beam split- ter/combiner 507 to the first slave laser 501 while a part of the second amplitude and phase perturbation laser beam (if one exists) is directed by the first beam split- ter/combiner 506 to the second slave laser 502. It is assumed that the first ampli- tude and phase perturbation laser beam is aligned such that it is effectively com- bined with the first injection laser beam in the second beam splitter/combiner el- ement 507. The same applies, mutatis mutandis, for the second amplitude and phase perturbation laser beam, the second injection laser beam and the first beam splitter/combiner element 506. The first and second amplitude and phase pertur- bation laser beams may be injected to the first and second slave lasers in a phase- conjugated forms relative to each other, as discussed in connection with above em- bodiments.

Finally, it should be noted that the first and second beam splitter/com- biner elements 506, 507 may be arranged such that the first and second laser beams outputted by the first and second slave lasers 501, 502 are able to pass through them (though their amplitude is reduced according to the split ratio of the corresponding beam splitter/combiner element 506, 507).

In some embodiments, the optical memory unit may comprise one or more optical elements not shown in Figure 5A such as at least one focusing lens.

It is obvious to any skilled person that various configurations and ar- rangements of optical elements may be employed for achieving the same goal of guiding the output beam of the master reference laser 503 to the first and second slave lasers 501, 502 for achieving injection locking and combining an amplitude and phase perturbation laser beam to at least one of the injected laser beams with- out disrupting the mutual coupling between the first and second slave lasers 501, 502. The optical memoiy unit of Figure 5B provides an alternative to the optical memory unit of Figure 5A in regard to the means used for achieving said goal. Spe- cifically, while the optical memory unit of Figure 5A employed three beam split- ter/combiner elements 506, 507, 508, the optical memory unit of Figure 5B em- ploys three beam splitter/combiner elements 511, 512, 516 as well as four opaque reflector elements 513, 514, 515, 517 (i.e., mirrors) for the same purpose. The op- erating principle of the optical memory unit is still largely the same and thus the discussion provided in connection with Figure 5A applies, mutatis mutandis, also for Figure 5B. Only the arrangement and alignment of some of the elements is dif- ferent due to the introduction of the reflector elements 513, 514, 515, 517. The re- flector elements 513, 514, 515, 517 serve merely to connect outputs of the beam splitter/combiner elements 511, 512, 516 to other beam splitter/elements 511, 512, 516. The configuration of Figure 5B may be beneficial in some cases, e.g., in view of mechanical layout considerations.

Figure 6 illustrates another detailed implementation of the optical memory unit of Figure 1A. The definitions provided in connection with Figure 1A apply also for this more detailed embodiment (unless otherwise explicitly stated). Specifically, Figure 6 corresponds to an integrated optics analog to the non-inte- grated optics (i.e., free-space optics) implementation of Figure 5A. Specifically, ele- ments 601 to 608 of Figure 6 correspond to integrated optics elements analogous with elements 501 to 508 of Figure 5A, respectively. Said integrated optics ele- ments 501 to 508 are connected to each other using optical fibers 609, as opposed to over air. In other embodiments, other optical waveguides (e.g., planar wave- guides, channel waveguides and/or hybrid plasmonic waveguides) may be used instead of or in addition to optical fibers. The discussion provided in connection with Figure 5A in regard to the operating principle of the optical memory unit ap- plies, mutatis mutandis, also for Figure 6 and is, thus, discussed in the following only briefly.

Referring to Figure 6, the optical memory unit comprises a first and sec- ond slave lasers 601, 602 and a reference master laser driving said first and second slave lasers 601, 602, similar to above embodiments. The lasers 601, 602, 603 may comprise, as depicted in Figure 6, one partially reflecting wall (facing the direction of propagation) and one (highly) reflective wall opposite to the partially reflecting wall. The reflective wall is substantially more reflective than the partially reflective wall. The optical memory unit comprises a phase shifting element 604 which may correspond to a distributed phase shifting element (i.e., a section of the optical fiber or other optical waveguide) or a discrete (tunable) phase shifter. The optical memory unit also comprises elements 606, 607, 608 corresponding to beam split- ter/combiner element as described in connection with above free-space optics em- bodiments. Here, the elements 606, 607, 608 may be specifically integrated optics directional couplers. At least some of the integrated optics directional couplers may have different split ratios. For example, the integrated optics directional cou- pler 608 may have a 50/50 split ratio while the integrated optics directional cou- plers 606, 607 may have a non-equal split ratio such as 90/10 or 99/1. The write signal (i.e., the amplitude and phase perturbation signal) may be received and the read signal transmitted via the unconnected port of the integrated optics direc- tional coupler(s) 606, 607 (i.e., in an analogous manner with Figures 5A and 5B). Alternatively, the write signal may be received and the read signal transmitted through a partially reflecting back wall mirror of the first slave laser 601 and/or the second slave laser 602.

In general, phases at the two outputs of the beam splitter 608 may not be the same. To mitigate this difference/asymmetry in the phase shifts and also to mitigate any possible asymmetries between the two sections of optical fiber and/or waveguide between the elements 608 and 606 and between the elements 608 and 607, two (tunable) phase shifting elements may be integrated to said two sections of optical fiber and/or waveguide, respectively. This way, it may be ensured that phases of the signals injected to the first and second slave lasers 601, 602 are suit- able. Similar adjustment may be applied in other embodiments, and especially, mu- tatis mutandis, in connection with either of the embodiments of Figures 5A and 5B.

In some embodiments, the directional coupler 608 may be replaced with an optical arrangement for signal modulation or specifically for generating the amplitude and phase perturbation component

Figure 7 illustrates another detailed implementation of the optical memory unit of Figure 1A using integrated optics. The definitions provided in con- nection with Figure 1A apply also for this more detailed embodiment (unless oth- erwise explicitly stated).

Referring to Figure 7, the optical memory unit comprises, similar to above embodiments, a first slave laser 701, a second slave laser 702 and a reference master laser 703 driving said first and second slave lasers 701, 702

The reference master laser 703 may be specifically a local reference la- ser which is configured to receive an external global reference signal or beam, that is., it is configured to be seeded by the global reference. The local reference master laser 703 comprises an amplifying medium 707 (e.g., a semiconductor material) and a (waveguide) loop mirror 708 which are connected together via an optical waveguide (and a directional coupler 704). The amplifying medium comprises a partially reflecting back mirror for partially receiving the global reference signal and partially reflecting any signals received from the integrated optics directional coupler 704 and/or the waveguide loop mirror 708 (and no front mirror).

The waveguide loop mirror 708 comprises an integrated optics direc- tional coupler 709 and a loop 710 formed with the optical waveguide (e.g., an opti- cal fiber, a planar waveguide, a channel waveguide or a hybrid plasmonic wave- guide). One of the ports of the integrated optics directional coupler of the wave- guide loop mirror 708 is connected to the integrated optics directional coupler 704 (and thus to the amplifying medium 707 and the first and second slave laser 701, 702) while two other ports of the integrated optics directional coupler of the wave- guide loop mirror 708 are connected to the two ends of the waveguide loop 710. Consequently, any signal coupled into the loop 710 will couple partially to the clockwise propagating arm and partially to the anticlockwise propagating arm. Both of said propagating beams are coupled again in the directional coupler 709 where they interfere destructively. Consequently, the directional coupler 709 of the loop mirror 708 does not transmit optical power and the light effectively re- flects back, similar to a reflection from a mirror. One 711 of the ports of the inte- grated optics directional coupler of the waveguide loop mirror 708 is not used by the optical memory unit, at least not if the directional coupler 709 associated with the loop mirror 708 splits the power evenly, i.e., it provides a 50/50 split ratio. If the directional coupler 709 has some other split ratio, such as ratio of 51/49, the reflection becomes partial in the loop mirror 708 and one 711 of the ports of the integrated optics directional coupler 709 of the waveguide loop mirror 708 can be used to monitor the operation of the local reference laser 703 of the optical memory unit

An extended local reference laser cavity is formed between the wave- guide loop mirror 708 and the end mirror of the amplifying medium 707. This ex- tended cavity leaks light via the integrated optics directional coupler 704 and the (tunable) phase shifting elements 705, 706 into the first and second slave lasers 701, 702, respectively, and thus injection-locking by the local reference master la- ser is enabled. The (tunable) phase shifting elements 705, 706 are used, similar to above embodiments, for ensuring that a phase shift of π exists between the first and second slave lasers 701, 702.

While the write signal (i.e., the amplitude and phase perturbation sig- nal) was received and the read signal transmitted via the unconnected port of the integrated optics directional coupler(s) 606, 607 in the embodiment of Figure 6 (i.e., in an analogous manner with Figures 5A and 5B), in the embodiment of Figure 7, the write signal is received and the read signal is transmitted directly via the first and second slave lasers 701, 702. To enable this functionality, the first and second slave lasers 701, 702 comprise partially reflecting front and back walls (or mir- rors).

Figure 8 illustrates another detailed implementation of the optical memory unit of Figure 1A using integrated optics. The definitions provided in con- nection with Figure 1A apply also for this more detailed embodiment (unless oth- erwise explicitly stated). Specifically, Figure 8 corresponds to an optical memory unit implemented using waveguide lasers as the first and second slave lasers. As in the above embodiments, optical waveguides used in connection with this embodi- ment may be optical fibers, planar waveguides, channel waveguides and/or hybrid plasmonic waveguides.

Referring to Figure 8, the first slave laser comprises a first optical wave- guide section 801 with integrated Bragg grating mirrors, a first amplifying medium 803, a second optical waveguide section 805 (possibly without Bragg grating mir- rors) and a third optical waveguide section 807 with integrated Bragg grating mir- rors. Similarly, the second slave laser also comprises a first optical waveguide sec- tion 802 with integrated Bragg grating mirrors, a first amplifying medium 804, a second optical waveguide section 806 (possibly without Bragg grating mirrors) and a third optical waveguide section 808 with integrated Bragg grating mirrors.

Bragg grating mirrors (integrated into the first and third optical wave- guide sections 801, 802, 807, 808) may be equally called fiber Bragg grating in con- nection with optical fibers though the same idea may be applied also to other opti- cal waveguides. Fiber Bragg grating is a type of distributed Bragg reflector con- structed in a segment of optical waveguide (e.g., optical fiber) that reflects particu- lar wavelengths of light and transmits all others. This is achieved by creating a pe- riodic variation in the refractive index of the fiber core. Thus, a wavelength-specific dielectric mirror is generated. A fiber Bragg grating can therefore be used as an inline optical filter to block certain wavelengths, or as a wavelength-specific reflec- tor.

In Figure 8, laser cavities of the first and second slave lasers are created specifically between the pairs of Bragg grating mirrors written or integrated into the first and second optical waveguide section 801, 807 & 802, 808 of the respec- tive slave lasers. The second optical waveguide sections 805, 806 of the first and second slave laser constitute a part of the respective laser cavities. As depicted in Figure 8, the two optical waveguide sections 805, 806 comprise two respective curving subsections and in-between them two respective (substantially straight) subsections which are in close proximity of each other. For example, the distance may be at least smaller than the operational wavelength. The distance should such that it enables coupling from one laser cavity to another via evanescent fields (i.e., mutually coupling between the first and second slave lasers 801, 802). Evanescent field is defined, in general, as an oscillating electric and/or magnetic field that does not propagate as an electromagnetic wave but whose energy is spatially concen- trated in the vicinity of the source (oscillating charges and currents). The amplitude of the evanescent wave decays exponentially with distance when moving away from the source (e.g., a waveguide segment) which is why the close spacing be- tween the (substantially straight) subsections of the two optical waveguide sec- tions 805, 806 is important. By controlling the properties of the close-proximity optical waveguide sections (e.g., at least the lengths of the close-proximity optical waveguide sections), the phase shift and the coupling strength (or coupling effi- ciency) between the first and second slave laser may be controlled. Specifically, said properties may be chosen so as to achieve a phase shift of π for a certain de- sired frequency range, similar to as described in connection with other embodi- ments.

A (reference) injection signal (originating, e.g., from an external refer- ence master laser not shown in Figure 8) is connected to a 50/50 directional cou- pler 811 (acting in an analogous manner, e.g., with the beam splitter/combiner el- ement 508 of Figure 5A). Thus, the 50/50 directional coupler 811 splits the injec- tion signal to into two injection signals which are fed, respectively, via first and second (tunable) phase shifting elements 809, 810 to laser cavities of the first and second slave lasers for achieving injection locking. The first and second (tunable) phase shifting elements 809, 810 may be configured to cause such phase shifts that the (relative) phase shift between a signal injected into the second slave laser and a signal component coupled from the first slave laser to the second slave laser is substantially equal to π and also that the (relative) phase shift between a signal injected into the first slave laser and a signal component coupled from the second slave laser to the first slave laser is substantially equal to π.

Figure 9 illustrates another detailed implementation of the optical memory unit of Figure 1A using integrated optics. The definitions provided in con- nection with Figure 1A apply also for this more detailed embodiment (unless oth- erwise explicitly stated). Specifically, Figure 9 corresponds to an optical memory unit implemented using ring lasers as the first and second slave lasers.

Ring lasers are, in general, lasers the resonator of which has the form of a ring. Consequently, such a ring resonator allows for two different propagation directions of the intracavity light Ring lasers may be configured to output two beams of light of the same polarization traveling in opposite directions ("counter- rotating") in a closed loop though in some cases unidirectional operation may be enforced with an optical isolator or with some other optical means.

Referring to Figure 9, the optical memory unit comprises a first slave ring laser 901, a second slave ring laser 902, a first set of first, second, third and fourth optical waveguides 905 to 908 and a second set of fifth and sixth optical waveguides 909, 910. The first, second, third and fourth optical waveguides 905 to 908 in the first set may be parallel with each other. Similarly, the fifth and sixth optical waveguides 909, 910 in the second set may be parallel with each other and also oriented preferably orthogonally to the first set. The orthogonal orientation serves to minimize unwanted coupling between the first and second sets of optical waveguides.

The first slave laser 901 is arranged and configured such that it is capa- ble of coupling to and from the first, second, fifth and sixth optical waveguides 905, 906, 909, 910 surrounding it. Similarly, the second slave laser 902 is arranged and configured such that it is capable of coupling to and from the third, fourth, fifth and sixth optical waveguides 907, 908, 909, 910 surrounding it.

Two phase shifting elements 903, 904 are arranged, respectively, to the fifth and sixth optical waveguides 909, 910 and between the first and second slave ring lasers 901, 902 for causing the phase shift of π between the first and second slave ring lasers 901, 902 at the first frequency outputted by the first and second slave ring lasers 901, 902 during injection locking. Said two phase shifting elements 903, 904 may be identical to each other.

A reference signal may be injected to the first slave ring laser 901 using the first optical waveguide 905 and to the second slave ring laser 902 using the fourth optical waveguide 908. The reference signal may be fed to the first and fourth optical waveguides 905, 908 from opposite directions, as illustrated in Fig- ure 9. The direction of rotation of light in the first and second ring lasers 901, 902 (illustrated in Figure 9 with a curving arrow) is determined by the direction of propagation of the reference signal (the rotation being along the direction of prop- agation of the reference signal).

Due to the coupling from the first and second slave ring lasers 901, 902 to the fifth and sixth optical waveguides 909, 910, the injection of the reference signal to the first and/or fourth optical waveguides 905, 908 causes also propaga- tion of optical signals along the fifth and sixth optical waveguides 909, 910. These signals correspond to the first and second (output) laser beams outputted by the first and second slave lasers in Figure 1A (or correspondingly also in, e.g., Figures 5A and SB) during injection locking. Phase and frequency of said signals are injec- tion locked to corresponding reference signals.

A write signal (i.e., an amplitude and phase perturbation signal) may be combined with the reference signal, in the first slave ring laser 901, by feeding the write signal to the second optical waveguide 906. The other end of the second op- tical waveguide 906 may be used for reading data (i.e., determining the current state of the memory). Similar to above embodiments, a second write signal may be combined with the reference signal fed to the second slave ring laser 902 by feed- ing it via the third optical waveguide 907 to the second slave ring laser 902 (from the top to the bottom in Figure 9).

While in above embodiments, the two steady states of the optical memory unit were characterized (or identified) through a pair of values of phases of the first and second laser beams outputted by the first and second slave lasers (and having the same polarization), in other embodiments, the steady states of the optical memory unit may, instead, be characterized (or identified) by the polariza- tion of said first and second laser beams.

To facilitate the following discussion involving polarization, some defi- nitions are provided in the following (with reference to Figure 1A). Polarization as used in the following may be defined (especially when two polarizations are com- pared to each other, e.g., in terms of orthogonality) in reference to a specific phys- ical location and for the electromagnetic wave that propagates toward a specific direction of observation. Thus, expressions of "first polarization" and "second po- larization" may, e.g., be understood as states of polarization of light emitted by the first slave laser 101 and the second slave laser 102, which polarizations are ob- served at the same location and for electromagnetic fields (or beam of light) that propagates toward an agreed direction. This definition may apply equally to the above embodiments.

For example, in the case of a linear cavity with at least partially reflect- ing mirrors at the both ends of the cavity, the 'first' polarization of the first slave laser 101 may be the state of polarization we observe at the location of observation immediately after the intracavity beam has been reflected from the surface that transmits light to the second slave laser 102, such that the observer is facing inside the cavity of the first slave laser 101. The 'second' polarization of the second slave laser 102 may be the state of polarization we observe at the same location of ob- servation inside the cavity of the first slave laser 101 once the light from the second slave laser 102 has been injected into the first slave laser 101, immediately after it has entered the first slave laser 101 (i.e., traversed the mirror surface of the first slave laser 101).

To give another example, in the case of a ring cavity, the 'first' polariza- tion of the first slave laser 101 is the state of polarization we observe at the location of observation where light transmits into a waveguide that carries it toward the second slave laser 102, such that the observer is facing toward the direction of propagation inside the cavity of the first slave laser 101. The 'second' polarization of the second slave laser 102 is the state of polarization we observe at the same location of observation inside the cavity of the first slave laser 101, such that the light from the second slave laser 102 has been injected into the first slave laser 101 and has propagated to the location of observation.

Moreover, in the following, we assume a pair of isotropic (slave) lasers, that is, the cavity medium has no significant linear or circular birefringence and that two orthogonal states of polarization, be they linear or circular or elliptical, have effectively uniform gain.

The optical memory unit 100 of Figure 1A may be employed also in po- larization-based embodiments. The discussion provided in connection with Figure 1A applies, mutatis mutandis, also in this case (but not the discussion of the oper- ation of the optical memory unit in connection with Figure 1B where the steady states are defined as a pair of phases and an amplitude and phase perturbation is used for switching steady states). Namely, also in this case, the optical memory comprises a first slave laser 101 configured to output a first laser beam having a first frequency when injection-locked with a first injection laser beam 103 having the first frequency and a second slave laser 102 configured to output a second laser beam having the first frequency when injection-locked with a second injection la- ser beam 104 having the first frequency. Here, the first slave laser 101 and the first injection beam 103 may not always have the same polarizations. The same is true for the second slave laser 102 and the second injection beam 104.

Moreover, the first and second slave lasers 101, 102 are also here con- figured to be mutually coupled and the optical memory unit 100 is configured so that an optical path length between the first and second slave lasers 101, 102 cor- responds to a phase shift substantially equal to π at the first frequency. Said injec- tion-locking of the first and second slave lasers 101, 102 with the first and second injection laser beams 103, 104 enables maintaining, at a time, one of two steady states of the optical memory unit when the first and second injection beams 101, 102 have, at the first and second slave laser 101, 102, the same amplitude and their phases for at least one state of polarization are locked.

In contrast to the phase-based optical memory unit, said two steady states may comprise, here specifically, a first steady state associated with a first polarization of the first laser beam outputted by the first slave laser 101 and a sec- ond polarization (different from the first polarization) of the second laser beam outputted by the second slave laser 102 and a second steady state associated with the second polarization of the first laser beam outputted by the first slave laser 101 and the first polarization of the second laser beam outputted by the second slave laser 102. During steady state operation, the state of polarization of the first and/or second laser beams outputted by the first and/or second slave laser 101, 102 stay constant and may differ from the state of polarization of the first and/or second injection laser beams 103, 104. The first polarization during steady state operation is a different polarization compared to the second polarization (at least at one com- mon point of observation at the first and/or second slave laser 101, 102). Prefera- bly, the first and second polarizations during steady state operation are orthogonal or substantially orthogonal (at least at one common point of observation located at the first and/or second slave laser 101, 102).

In practice, each of the first and second polarizations may be divisible into a primary polarization component corresponding to the intended polarization state and a minor or secondary polarization component corresponding to an un- wanted polarization state arising due to practical considerations and being orthog- onal to the primary polarization component. Here, the primary polarization com- ponents of the first and second polarizations are preferably orthogonal to each other. Further, the minor polarization components of the first and second polariza- tions are also preferably orthogonal to each other. Obviously, the minor polariza- tion components are preferably much smaller than the corresponding primary po- larization components.

In some embodiments, the first and second polarization may specifi- cally correspond to orthogonal elliptical polarizations. In such embodiments, the first polarization may correspond to a first elliptical state of polarization with a primary linear polarization state (e.g., vertical polarization) and a minor linear po- larization state (e.g., horizontal polarization). Similarly, the second polarization may correspond to a second elliptical state of polarization with a second primary linear polarization state (e.g., horizontal polarization) and a second minor linear polarization state (e.g., vertical polarization).

In other embodiments, the first and second polarizations may corre- spond to orthogonal circular polarizations. In such embodiments, the first polari- zation may, for example, correspond to a first elliptical state of polarization with a primary circular polarization state (e.g., right-handed circular polarization) and a minor polarization state (e.g., left-handed circular polarization). Similarly, the sec- ond polarization may correspond to a second elliptical state of polarization with a second primary circular polarization state (e.g., left-handed circular polarization) and a second minor circular polarization state (e.g., right-handed circular polariza- tion).

In yet other embodiments, some other orthogonal bases of polariza- tions may be employed for the first and second polarizations. In the polarization-based memory unit, the switching between the two steady states may be achieved by introducing an amplitude and polarization per- turbation component (or an amplitude, phase and polarization perturbation com- ponent), as opposed to an amplitude and phase perturbation component, to at least one of the first and second injection laser beams 103, 104. If the amplitude and polarization perturbation component is introduced to both of the firstand second injection laser beams 103, 104, it may be introduced to them in a polarization-ro- tated form relative to each other. Preferably, the amplitude and polarization per- turbation component is introduced to the first and second injection laser beams 103, 104 in a polarization-orthogonal form relative to each other. For example, the amplitude and polarization perturbation component when introduced to the first injection laser beam 103 may have a (substantially) vertical linear polarization while the amplitude and polarization perturbation component when introduced to the second injection laser beam 105 may have a (substantially) horizontal linear polarization.

In addition to the general optical memory unit 100 of Figure 1A, any of the detailed implementations of the optical memory unit 100 of Figure 1A may be equally used as a polarization-based optical memory unit. However, some of the embodiments such as the free-space-optics-based embodiments discussed in con- nection with Figures 5A and 5B may be more readily suitable for use as a polariza- tion-based optical memory units compared to other embodiments such as wave- guide-based embodiments of Figures 6 to 9 where, for example, birefringence of waveguides and optical fibers affecting the state of polarization of propagating light must be taken in account in the design of the polarization-based optical memory unit.

Figure 10A illustrates, as a flowchart, the general operation (i.e., switch- ing between steady states) of the polarization-based optical memory unit while Figure 10B illustrates corresponding exemplary simulation results. Specifically, Figure 10B illustrates exemplary simulation results for an execution of the process of Figure 10A carried out for an optical memory unit of Figure 1A. Figure 10A illus- trates input signal in terms of power separated to two orthogonal linear polariza- tion components (x and y polarization components illustrated with solid and dashed lines) over time for a first injection laser beam (top subfigure), for a first laser beam (i.e., output) of the first slave laser L1 (middle subfigure) and for a sec- ond laser beam (i.e., an output) of the second slave laser L2 (bottom sub-figure). In Figure 10B, the amplitude and polarization perturbation is injected into the first slave laser (L1) and with opposite sign (i.e., in a phase-conjugated form) to the sec- ond slave laser (L2), similar to Figure 4 though power for only the first injection laser beam is shown in Figure 10B. The power traces are shown normalized to a pre-defined value (exact power values being irrelevant for the following discus- sion). In Figure 10B, the steady states are defined through two orthogonal elliptical polarization states with primary linear components (i.e., the first and second pri- mary polarizations mentioned above) though the discussion applies equally for other orthogonal polarizations as discussed above.

Referring to Figure 10A, the optical memory unit is initially operated, in block 1001, in a first steady state. As described above, the first steady state may be associated with a first primary polarization of the first laser beam outputted by the first slave laser and a second primary polarization of the second laser beam out- putted by the second slave laser (preferably orthogonal to the first primary polar- ization). The first and second primary polarization may be defined as described above. In the first steady state, the first and/or second injection laser beam may have a third polarization corresponding to a linear combination (preferably, equal linear combination) of the first and second primary polarizations.

In some embodiments, one or more polarization rotators (or other po- larization modifying means) may be arranged in an optical path between the first and second slave lasers. Specifically in such a case, the first polarization and the second polarization associated with the two steady states as described above may be orthogonal only when observed at an input/output of the first slave laser and/or at an input/output of the second slave laser (i.e., they are orthogonal after polari- zation rotation has been applied to one of the outputted first and second laser beams). Said one or more polarization rotators may comprise one or more recip- rocal polarization rotators and/or one or more non-reciprocal polarization rota- tors (e.g., Faraday rotators).

In Figure 10B, the first steady state may be interpreted to correspond to the state of the system between roughly 60 ns and 90 ns. Thus, based on middle and bottom subfigures, the first polarization corresponds in this example to a lin- ear primary polarization aligned with the x direction and a minor polarization aligned with they direction (corresponding to close-to-zero power) and the second polarization corresponds to a linear primary polarization aligned with they direc- tion and a minor polarization aligned with the x direction (corresponding to close- to-zero power). In other words, at the first steady state, the electric field of the first laser beam outputted by the first slave laser has a minory-component (as indicated by the power being close to zero) and a non-zero primary x-component, as shown in the middle subfigure. Conversely, the electric field of the second laser beam out- putted by the second slave laser has a minor x-component and a non-zero primary /-component, as shown in the bottom subfigure. The first (and second) injection laser beam has, at the first steady state, a "diagonal" linear polarization, that is, a linear polarization forming a 45° angle with the x and y directions, as shown in the top subfigure by the two lines corresponding to x and y polarizations being fully overlapping. In other words, it has the same amplitude (and power) for horizontal and vertical polarization states and the phase shift between the electromagnetic waves of these two orthogonal polarization states is zero.

Referring to Figure 10A, switching of the optical memory unit from the first steady state to the second steady state is caused, in block 1002, by introducing an ampli- tude and polarization perturbation component having a pre-defined duration to at least one of the first and second injection laser beams. The discussion regarding the peak power and duration of the amplitude and phase perturbation component pro- vided in connection with Figure 1A applies, mutatis mutandis, for the peak power and duration of the amplitude and polarization perturbation component.

In Figure 10B, an amplitude and polarization perturbation component is introduced to the first and second injection laser beams between 90 ns and 100 ns. This is visible in the top subfigure corresponding to the first injection laser beam as a sudden increase in the power associated with the linear polarization aligned with the y direction at 90 ns while no change is observed in the power of the linear polarization aligned with the x direction. While here a square pulse is used as the amplitude and polarization perturbation component, other pulse shapes may also be used. In regard to the middle subfigure corresponding to the first laser beam outputted by the first slave laser, the introduction of they-directed perturbation to the first injection laser beam causes the power of the x-directed polarization component to fall to zero and the power of they-directed polarization component to jump to a high value (close to a steady state value). Conversely, in regard to the bottom subfigure corresponding to the second laser beam outputted by the second slave laser, the introduction of the y-directed perturbation causes the power ofy-directed polarization component to fall to zero and the power of the x-directed polarization component to jump to a high value (close to a steady state value).

After the duration of the amplitude and polarization perturbation com- ponent, the optical memory unit settles quickly to the second steady state (100 ns - 140 ns). The second steady state corresponds to the first steady state with the difference that the primary polarizations of the first and second laser beams out- putted by the first and second slave lasers have been rotated by 90°, i.e., the align- ment of the linear primary polarizations have switched from x to y and from y to x, respectively.

It should be noted before entering the first steady state, the optical memory unit of Figure 10B is initially (time being between 40 to 50 ns) in an initial start-up state which does not correspond to either of the first and second steady states. The optical memory unit is brought to the first steady state by introducing an amplitude and polarization perturbation (at time 50 ns to 60 ns). Notably, this amplitude and polarization perturbation is orthogonal in terms of polarization to the amplitude and polarization perturbation introduced at time 90-100 ns.

Reading of data from the polarization-based optical memory unit may be performed simply by measuring the polarization of at least one of the first and second laser beams outputted by the first and second slave lasers, as opposed to measuring the phases of the first and second laser beams. As described also above, one or more optical elements (e.g., a beam splitter/combiner) may be, in some em- bodiments, configured to output the first and/or second laser beams (or a particu- lar polarization component thereof) from the optical memory unit for enabling reading of data of the optical memory unit.

The following parameters were used for the simulations of Figure 10B. The same equations as cited in connection with Figure 4 were used also in these simulations.

In some embodiments, the concepts of the phase-based optical memory unit and the polarization-based optical memory unit may be combined. In such em- bodiments, the optical memory unit may be defined as described above (e.g., in connection with Figure 1A), but the two steady states and the switching between them may be defined differently. Namely, the two steady states of the phase & po- larization -based optical memory unit comprise a first steady state associated with a first pair of values of phases of the first and second laser beams outputted by the first and second slave lasers (preferably, +π/3 & -π/3) and with a first polarization and a second steady state associated with a second pair of values of the phases of the first and second laser beams outputted by the first and second slave lasers (preferably, -π/3 & +π/3) and a second polarization (preferably, orthogonal with the first polarization). The first and second polarization may be defined as de- scribed in connection with the polarization-based optical memory unit. Moreover, in such embodiments, switching between the two steady states would be enabled by introducing an amplitude, phase and polarization perturbation component (i.e., a combination of the amplitude and perturbation component and the amplitude and polarization perturbation component as discussed above), to at least one of the first and second slave lasers.

Even though the invention has been described above with reference to examples according to the accompanying drawings, it is clear that the invention is not restricted thereto but can be modified in several ways within the scope of the appended claims. Therefore, all words and expressions should be interpreted broadly, and they are intended to illustrate, not to restrict, the embodiment. It will be obvious to a person skilled in the art that, as technology advances, the inventive concept can be implemented in various ways. Further, it is clear to a person skilled in the art that the described embodiments may, but are not required to, be com- bined with other embodiments in various ways.