We examine optical V-band light curves in luminous eclipsing black hole X-ray binaries, using a supercritical accretion/outflow model that is more realistic than the formerly used ones. In order to compute the theoretical light curve in the binary system, we do not only apply the global analytic solution of the disk, but also include the effect of the optically thick outflow. We found that the depth of eclipse of the companion star by the disk changes dramatically when including the effect of the outflow. Due to the effect of outflow, we can reproduce the optical light curve for typical binary parameters in SS433. Our model with outflow velocity v~3000 km/s can fit whole shape of the averaged V-band light curve in SS433, but we found a possible parameter range consistent with observations, such as \dot{M}~5000-10000 L_E/c^2 (with L_E being the Eddington luminosity and $c$ being the speed of light) and T_C~10000K-14000 K for the accretion rate and donor star temperature, respectively. Furthermore, we briefly discuss observational implications for ultraluminous X-ray sources.

We examine hybrid thermal-nonthermal synchrotron spectra from a spherically symmetric, optically-thin wind, taking into account the relativistic effect. In the relativistic flow from the central object, due to the relativistic beaming effect, the observed spectra often shift towards high frequency and high intensity directions. In the optically thin outflows, however, we find that the intensity of the observed spectra decreases compared with that of the emitted ones, although the peak frequency shifts towards the high frequency direction. This is because in the optically thin outflows we can see the far side flows that go away from the observer. We thus carefully consider optically thin relativistic flows around a black hole such as Sgr A$^*$.

We examine the behavior of the variable Eddington factor for a relativistically moving radiative flow in the vertical direction. We adopt the "one-tau photo-oval" approximation in the comoving frame. Namely, the comoving observer sees radiation coming from a closed surface where the optical depth measured from the observer is unity; such a surface is called a one-tau photo-oval. In general, the radiative intensity emitted by the photo-oval is non-uniform and anisotropic. Furthermore, the photo-oval surface has a relative velocity with respect to the comoving observer, and therefore, the observed intensity suffers from the Doppler effect and aberration. In addition, the background intensity usually depends on the optical depth. All of these introduce the anisotropy to the radiation field observed by the comoving observer. As a result, the relativistic Eddington factor $f$ generally depends on the optical depth $\tau$, the four velocity $u$, and the velocity gradient $du/d\tau$. % In the case of a plane-parallel vertical flow, we found that the relativistic variable Eddington factor $f$ generally decreases as the velocity gradient increases, but it increases as the velocity increases for some case. When the comoving radiation field is uniform, it is well approximated by $3f \sim 1/[ 1+ ({16}/{15})(-{du}/{\gamma d\tau}) +(-{du}/{\gamma d\tau})^{1.6-2} ]$. When the radiation field in the inertial frame is uniform, on the other hand, it is expressed as $f = (1+3\beta^2)/(3+\beta^2$). These relativistic variable Eddington factors can be used in various relativistic radiatively-driven flows, such as black-hole accretion flows, relativistic astrophysical jets and outflows, and relativistic explosions like gamma-ray bursts.

Radiative transfer in a relativistic plane-parallel flow, e.g., an accretion disk wind, is examined in the fully special relativistic treatment. Under the assumption of a constant flow speed, for the relativistically moving atmosphere we analytically obtain generalized Milne-Eddington solutions of radiative moment equations; the radiation energy density, the radiative flux, and the radiation pressure. In the static limit these solutions reduce to the traditional Milne-Eddington ones for the plane-parallel static atmosphere, whereas the source function nearly becomes constant as the flow speed increases. Using the analytical solutions, we analytically integrate the relativistic transfer equation to obtain the specific intensity. This specific intensity also reduces to the Milne-Eddinton case in the static limit, while the emergent intensity is strongly enhanced toward the flow direction due to the Doppler and aberration effects as the flow speed increases (relativistic peaking).

We examine the observational appearance of an optically thick, spherically symmetric, relativistic wind (a black hole wind). In a relativistic flow, the apparent optical depth becomes small in the downstream direction, while it is large in the upstream direction due to the relativistic effect. Hence, the apparent photosphere of the relativistic flow depends on the flow velocity and direction as well as the density distribution. We calculate the temperature distribution of the apparent photosphere of the optically thick black hole wind, where the wind speed is assumed to be constant and radiation dominates matter, for various values of the wind speed and mass-outflow rate. We found that the limb-darkening effect is strongly enhanced in the relativistic regime. We also found that the observed luminosities of the black hole wind become large as the wind speed increases, but do not depend on the mass-outflow rate.

The photon mean free path in a relativistically moving medium becomes long in the down-stream direction while short in the up-stream direction. As a result, the observed optical depth $\tau$ becomes small in the downstream direction while large in the upstream direction. Hence, if a relativistic spherical wind blows off, the optical depth depends strongly on its speed and the angle between the velocity and the line-of-sight. Abramowicz et al. (1991) examined such a relativistic wind, and found that the shape of the photosphere at $\tau=1$ appears convex in the non-relativistic case, but concave for relativistic velocities. We further calculated the temperature distribution and luminosity of the photosphere both in the comoving and inertial frames. We found that the limb-darkening effect would strongly modified in the relativistic regime. We also found that luminosities of the photosphere becomes large as the wind speed increases due to the relativistic effects. In addition, the luminosity in the inertial frame is higher than that in the comoving frame. These results suggest that the observed temperature and brightness in luminous objects may be overestimated when there are strong relativistic winds.

We examine the Eddington factor in an optically thick, relativistic flow accelerating in the vertical direction. % When the gaseous flow is radiatively accelerated and there is a velocity gradient, there also exists a density gradient. The comoving observer sees radiation coming from a closed surface where the optical depth measured from the observer is unity. Such a surface, called a {\it one-tau photo-oval}, is elongated in the flow direction. In general, the radiation intensity emitted by the photo-oval is non-uniform, and the photo-oval surface has a relative velocity with respect to the position of the comoving observer. Both effects introduce some degree of anisotropy in the radiation field observed in the comoving frame. As a result, the radiation field observed by the comoving observer becomes {\it anisotropic}, and the Eddington factor must deviate from the usual value of 1/3. Thus, the relativistic Eddington factor generally depends on the optical depth $\tau$ and the velocity gradient $du/d\tau$, $u$ being the four velocity. % In the case of a plane-parallel vertical flow, we obtain the shape of the photo-oval and calculate the Eddington factor in the optically thick regime. We found that the Eddington factor $f$ is well approximated by $f(\tau, \frac{du}{d\tau}) = {1/3} \exp (\frac{1}{u} \frac{du}{d\tau}) $. % This relativistic variable Eddington factor can be used in various relativistic radiatively-driven flows.

We analytically derive a relativistic variable Eddington factor in the relativistic radiative flow, and found that the Eddington factor depends on the {\it velocity gradient} as well as the flow velocity. When the gaseous flow is accelerated and there is a velocity gradient, there also exists a density gradient. As a result, an unobstructed viewing range by a comoving observer, where the optical depth measured from the comoving observer is unity, is not a sphere, but becomes an oval shape elongated in the direction of the flow; we call it a {\it one-tau photo-oval}. For the comoving observer, an inner wall of the photo-oval generally emits at a non-uniform intensity, and has a relative velocity. Thus, the comoving radiation fields observed by the comoving observer becomes {\it anisotropic}, and the Eddington factor must deviate from the value for the isotropic radiation fields. % In the case of a plane-parallel vertical flow, we examine the photo-oval and obtain the Eddington factor. In the sufficiently optically thick linear regime, the Eddington factor is analytically expressed as $f (\tau, \beta, \frac{d\beta}{d\tau}) = {1/3} (1 + {16/15} \frac{d\beta}{d\tau})$, where $\tau$ is the optical depth and $\beta$ ($=v/c$) is the flow speed normalized by the speed of light. %i.e., the Eddington factor depends on the velocity gradient. We also examine the linear and semi-linear regimes, and found that the Eddington factor generally depends both on the velocity and its gradient.

We solve spherically symmetric radiation flows under full special relativity with the help of a variable Eddington factor $f(\tau, \beta)$, where $\tau$ is the optical depth and $\beta$ is the flow velocity normalized by the speed of light. Relativistic radiation hydrodynamics under the moment formalism has several complex problems, such as a closure relation. Conventional moment equations closed with the traditional Eddington approximation in the comoving frame have singularity, beyond which the flow cannot be accelerated. In order to avoid such a pathological behavior inherent in the relativistic moment formalism, we propose a variable Eddington factor, which depends on the flow velocity as well as the optical depth, for the case of the sperically symmertic one-dimensional flow. We then calculate the relaticistic spherical flow with such variable Eddington factors to investigate the case that gas is accelerated by radiative force. As a result, it is shown that the gas speed reaches around the speed of light by radiation pressure.

Radiative transfer in a relativistic accretion disk wind is examined under the plane-parallel approximation in the fully special relativistic treatment. For an equilibrium flow, where the flow speed and the source function are constant, the emergent intensity is analytically obtained. In such an equilibrium flow the usual limb-darkening effect does not appear, since the source function is constant. Due to the Doppler and aberration effects associated with the relativistic motion of winds, however, the emergent intensity is strongly enhanced toward the flow direction. This is the {\it relativistic peaking effect}. We thus carefully treat and estimate the appearance of relativistic winds and jets, when we observe them in an arbitrary direction.

Accretion disk winds browing off perpendicular to a luminous disk are examined in the framework of fully special relativistic radiation hydrodynamics. The wind is assumed to be steady, vertical, and isothermal. %and the gravitational fields is approximated by a pseudo-Newtonian potential. Using a velocity-dependent variable Eddington factor, we can solve the rigorous equations of relativistic radiative hydrodynamics, and can obtain radiatively driven winds accelerated up to the {\it relativistic} speed. For less luminous cases, disk winds are transonic types passing through saddle type critical points, and the final speed of winds increases as the disk flux and/or the isothermal sound speed increase. For luminous cases, on the other hand, disk winds are always supersonic, since critical points disappear due to the characteristic nature of the disk gravitational fields. The boundary between the transonic and supersonic types is located at around $\hat{F}_{\rm c} \sim 0.1 (\epsilon+p)/(\rho c^2)/\gamma_{\rm c}$, where $\hat{F}_{\rm c}$ is the radiative flux at the critical point normalized by the local Eddington luminosity, $(\epsilon+p)/(\rho c^2)$ is the enthalpy of the gas divided by the rest mass energy, and $\gamma_{\rm c}$ is the Lorentz factor of the wind velocity at the critical point. In the transonic winds, the final speed becomes 0.4--0.8$c$ for typical parameters, while it can reach $\sim c$ in the supersonic winds.

Radiative transfer equation in an accretion disk wind is examined analytically and numerically under the plane-parallel approximation in the subrelativistic regime of $(v/c)^1$, where $v$ is the wind vertical velocity. Emergent intensity is analytically obtained for the case of a large optical depth, where the flow speed and the source function are almost constant. The usual limb-darkening effect, which depends on the direction cosine at the zero-optical depth surface, does not appear, since the source function is constant. Because of the vertical motion of winds, however, the emergent intensity exhibits the {\it velocity-dependent} limb-darkening effect, which comes from the Doppler and aberration effects. Radiative moments and emergent intensity are also numerically obtained. When the flow speed is small ($v \leq 0.1c$), the radiative structure resembles to that of the static atmosphere, where the source function is proportional to the optical depth, and the usual limb-darkening effect exists. When the flow speed becomes large, on the other hand, the flow speed attains the constant terminal one, and the velocity-dependent limb-darkening effect appears. We thus carefully treat and estimate the wind luminosity and limb-darkening effect, when we observe an accretion disk wind.

Transfer equation in a geometrically thin accretion disk is reexamined under the plane-parallel approximation with finite optical depth. Emergent intensity is analytically obtained in the cases with or without internal heating. For large or infinite optical depth, the emergent intensity exhibits a usual limb-darkening effect, where the intensity linearly changes as a function of the direction cosine. For small optical depth, on the other hand, the angle-dependence of the emergent intensity drastically changes. In the case without heating but with uniform incident radiation at the disk equator, the emergent intensity becomes isotropic for small optical depth. In the case with uniform internal heating, the limb brightening takes place for small optical depth. We also emphasize and discuss the limb-darkening effect in an accretion disk for several cases.

Radiatively-driven transfer flow perpendicular to a luminous disk is examined in the relativistic regime of $(v/c)^2$, taking into account the gravity of the central object. The flow is assumed to be vertical, and the gas pressure as well as the magnetic field are ignored. Using a velocity-dependent variable Eddington factor, we can solve the rigorous equations of the relativistic radiative flow accelerated up to the {\it relativistic} speed. For sufficiently luminous cases, the flow resembles the case without gravity. For less-luminous or small initial radius cases, however, the flow velocity decreases due to gravity. Application to a supercritical accretion disk with mass loss is briefly discussed.

We have proposed in previous papers that the high-frequency pair QPOs observed in black-hole binaries with frequency ratio 3:2 are inertial-acoustic oscillations (nearly horizontal oscillations with no node in the vertical direction) or g-mode oscillations, which are resonantly excited on warped relativistic disks. The resonance occurs through horizontal motions. In this model the dimensionless spin parameter $a_*$ of the central sources can be estimated when their masses are known from other observations. This estimate is done for three sources (GRO J1665-40, XTE J1550-564, GRS 1915+105). For all of them we have $a_*\leq 0.45$.

We calculated the bolometric images of relativistic slim disks during radiation-pressure-driven thermal instability. When the mass-accretion rate exceeds the critical one, an inner region of the standard accretion disk bursts to change to a slim disk state having a large scale height. That is, the inner region of the disk becomes high temperature, and the thickness of the disk increases due to the increase of the radiation pressure. As a result, we found that the observed image of the disk during the burst strongly depends on the inclination angle. That is, radiation from the innermost disk would be occulted by the disk outer rim for high inclination angles($i \gtrsim 70^\circ$). We also calculated the spectral energy distribution during thermal instability. The Wien peak of the spectrum of high inclination angles becomes softer than that of low inclination angles due to the geometrical thickness. From these facts, even if the burst happens in black hole candidates, we may not observe the burst when the inclination angle is large. We may suggest that numerous luminous black hole candidates are still hidden in our Galaxy.

We examined the effect of toroidal magnetic fields on a viscous gaseous disk around a central object under an advection dominated stage. We found self-similar solutions for radial infall velocity, rotation velocity, sound speed, with additional parameter $\beta$ [$=c_{\rm A}^2/(2c_{\rm s}^2)$], where $c_{\rm A}$ is the Alfv\'en speed and $c_{\rm s}$ is the isothermal sound speed. Compared with the non-magnetic case, in general the disk becomes thick due to the magnetic pressure, and the radial infall velocity and rotation velocity become fast. In a particular case, where the magnetic field is dominant, on the other hand, the disk becomes to be magnetically supported, and the nature of the disk is significantly different from that of the weakly magnetized case.

We propose a variable Eddington factor, depending on the {\it flow velocity} $v$, for the relativistic radiative flow, whose velocity becomes of the order of the speed of light. When the gaseous flow is radiatively accelerated up to the relativistic regime, the velocity gradient becomes very large in the direction of the flow. As a result, the radiative diffusion may become {\it anisotropic} in the comoving frame of the gas. Hence, in a flow that is accelerated from subrelativistic to relativistic regimes, the Eddington factor should be different from 1/3 even in the diffusion limit. As a simple form, the velocity-dependent Eddington factor may be written as $f(\beta) = 1/3+(2/3)\beta$, where $\beta=v/c$. Using the velocity-dependent Eddington factor, we can solve the rigorous equations of the relativistic radiative flow accelerated up to the relativistic speed. We also propose a generalized form for a variable Eddington factor as a function of the optical depth $\tau$ as well as the flow velocity: %$f(\tau, \beta) = {1/3} + {2/3} % \frac{1+(\tau+1)\beta}{1+\tau+\beta}$ $f(\tau, \beta) = 1/3 + (2/3) [{1+(\tau+1)\beta}]/({1+\tau+\beta})$ for a spherically symmetric case. The velocity-dependent Eddington factor can be used in various relativistic radiatively-driven flows, such as black-hole accretion flows, relativistic astrophysical jets and outflows, and relativistic explosions like gamma-ray bursts.

Radiatively driven transfer flow perpendicular to a luminous disk was examined under a fully special relativistic treatment, taking into account radiation transfer. The flow was assumed to be vertical, and the gravity, the gas pressure, and the viscous heating were ignored. In order to construct the boundary condition at the flow top, the magic speed above the flat source was re-examined, and it was found that the magic speed above a moving source can exceed that above a static source ($\sim 0.45~c$). Then, the radiatively driven flow in a luminous disk was numerically solved, from the flow base (disk ``inside''), where the flow speed is zero, to the flow top (disk ``surface''), where the optical depth is zero. For a given optical depth and appropriate initial conditions at the flow base, where the flow starts, a loaded mass in the flow was obtained as an eigenvalue of the boundary condition at the flow top. Furthermore, a loaded mass and the flow final speed at the flow top were obtained as a function of the radiation pressure at the flow base; the flow final speed increases as the loaded mass decreases. Moreover, the flow velocity and radiation fields along the flow were obtained as a function of the optical depth. Within the present treatment, the flow three velocity $v$ is restricted to be within the range of $v < c/\sqrt{3}$, which is the relativistic sound speed, due to the relativistic effect.

Radiatively-driven flow in a luminous disk is examined in the subrelativistic regime of $(v/c)^1$, taking account of radiation transfer. The flow is assumed to be vertical, and the gravity and gas pressure are ignored. When internal heating is dropped, for a given optical depth and radiation pressure at the flow base (disk ``inside''), where the flow speed is zero, the flow is analytically solved under the appropriate boundary condition at the flow top (disk ``surface''), where the optical depth is zero. The loaded mass and terminal speed of the flow are both determined by the initial conditions; the mass-loss rate increases as the initial radiation pressure increases, while the flow terminal speed increases as the initial radiation pressure and the loaded mass decrease. In particular, when heating is ignored, the radiative flux $F$ is constant, and the radiation pressure $P_0$ at the flow base with optical depth $\tau_0$ is bound in the range of $2/3 < cP_0/F < 2/3 + \tau_0$. In this case, in the limit of $cP_0/F = 2/3 + \tau_0$, the loaded mass diverges and the flow terminal speed becomes zero, while, in the limit of $cP_0/F = 2/3$, the loaded mass becomes zero and the terminal speed approaches $(3/8)c$, which is the terminal speed above the luminous flat disk under an approximation of the order of $(v/c)^1$. We also examine the case where heating exists, and find that the flow properties are qualitatively similar to the case without heating.

Radiatively-driven transfer flow perpendicular to a luminous disk is examined in the subrelativistic regime of $(v/c)^1$, taking into account the gravity of the central object. The flow is assumed to be vertical, and the gas pressure is ignored, while internal heating is assumed to be proportional to the gas density. The basic equations were numerically solved as a function of the optical depth, and the flow velocity, the height, the radiative flux, and the radiation pressure were obtained for a given radius, an initial optical depth, and initial conditions at the flow base (disk ``inside''), whereas the mass-loss rate was determined as an eigenvalue of the boundary condition at the flow top (disk ``surface''). For sufficiently luminous cases, the flow resembles the case without gravity. For less-luminous cases, however, the flow velocity decreases, and the flow would be impossible due to the existence of gravity in the case that the radiative flux is sufficiently small. Application to a supercritical accretion disk with mass loss is briefly discussed.

We propose an eclipsing light-curve diagnosis for black-hole accretion flows. When emission from an inner accretion disk around a black hole is occulted by a companion star, the observed light curve becomes asymmetric at ingress and egress on a time scale of 0.1-1 seconds. This light-curve analysis provides a means of verifying the relativistic properties of the accretion flow, based on the special/general relativistic effects of black holes. The ``skewness'' for the eclipsing light curve of a thin disk is $\sim 0.08$, whereas that of a slim disk is $\sim 0$, since the innermost part is self-occulted by the disk's outer rim.

We investigate the dependence of the viewing angle in supercritical accretion flows and discuss the observational implications of galactic black-hole candidates and ultraluminous X-ray sources. When the mass accretion rate exceeds the critical rate, then the shape of the disk is geometrically thick due to the enhanced radiation pressure. The model spectra of supercritical accretion flows strongly depend on the inclination angle. Because the outer disk blocks the emission from the disk inner region for high inclination angle. We also find that the spectral properties of low-inclination angle and low accretion-rate disks are very similar to those of high-inclination and high accretion rate disks. That is, if an object has a high inclination and high accretion rate, such a system suffers from self-occultation and the spectrum will be extremely soft. Therefore, we cannot discriminate these differences from spectrum shapes only. Conversely, if we use the self-occultation properties, we could constrain the inclination angle of the system. We suggest that some observed high temperature ultraluminous X-ray sources have near face-on geometry, i < 40, and Galactic black hole candidate, XTE J1550-564, possesses relatively high-inclination angles, i > 60.

We investigate Hoyle-Lyttleton accretion for the case where the central source is a luminous accretion disk. %In classical Hoyle-Lyttleton accretion onto a ``spherical'' source, accretion takes place in an axially symmetric manner around a so-called accretion axis. The accretion rate of the classical Hoyle-Lyttleton accretion onto a non-luminous object and $\Gamma$ the luminosity of the central object normalized by the Eddington luminosity. %If the central object is a compact star with a luminous accretion disk, the radiation field becomes ``non-spherical''. %Although the gravitional field remains spherical. In such a case the axial symmetry around the accretion axis breaks down; the accretion radius $R_{acc}$ generally depends on an inclination angle $i$ between the accretion axis and the symmetry axis of the disk and the azimuthal angle $\phi$ around the accretion axis. %That is, the cross section of accretion changes its shape. Hence, the accretion rate $\dot{M}$, which is obtained by integrating $R_{acc}$ around $\phi$, depends on $i$. % as well as $M$, $\Gamma$, and $v_\infty$. %In the case of an edge-on accretion ($i=90^{\circ}$), The accretion rate is larger than that of the spherical case and approximately expressed as $\dot{M} \sim \dot{M}_{HL} (1-\Gamma)$ for $\Gamma \leq 0.65$ and $\dot{M} \sim \dot{M}_{HL} (2-\Gamma)^2/5$ for $\Gamma \geq 0.65$. %Once the accretion disk forms and the anisotropic radiation fields are produced around the central object,the accretion plane will be maintained automatically (the direction of jets associated with the disk is also maintained). %Thus, the anisotropic radiation field of accretion disks drastically changes the accretion nature, that gives a clue to the formation of accretion disks around an isolated black hole.

The optical-ultraviolet continuum from active galactic nuclei (AGN) seem to originate from optically thick and/or thin disks, and occasionally from associated circumnuclear starburst regions. These different possible origins can, in principle, be discriminated by observations of gravitational microlensing event. We performed numerical simulations of microlensing of an AGN disk by a single lensing star passing in front of the AGN. Calculated spectral variations and light curves show distinct behavior, depending on the nature of emitting region; the time variation of a few months with strong wavelength dependence is expected in the case of an optically thick disk ({\it standard disk}), while an optically thin disk ({\it advection-dominated disk}) will produce shorter, nearly wavelength-independent variation. In the case of a starburst region much slower variations (over a year) will be superposed on the shorter variations.

Hoyle-Lyttleton type accretion is investigated, by taking account of not only the effect of radiation pressure but the effect of radiation drag. We calculate the trajectories of particles for three cases: only the effect of gravity is considered (case A); the effect of radiation pressure is taken into account (case B); the effect of radiation drag as well as radiation pressure is taken into account (case C). The accretion radii for former two cases are $2GM/v_{\infty}^2$ for case A and $2GM(1-\Gamma)/v_{\infty}^2$ for case B, where M is the mass of the accreted object, $v_{\infty}$ the relative velocity, and Gamma the normalized luminosity of the accreted object. We found that the accretion radius for case C is in between those of cases A and B under the present approximation; i.e., the accretion radius decreases due to radiation pressure while it increases due to radiation drag. In addition, the accretion radius for case C becomes larger as the incident velocity becomes fast. The effect of radiation drag becomes more and more important when the velocity of the incident particle is comparable to the light speed.

We examine the radiatively driven mass accretion onto galactic nuclei due to the intensive radiation from circumnuclear starbursts. The radiation from a starburst not only contracts an inner gas disk by the radiation flux force, but also extracts angular momenta due to the relativistic radiation drag, thereby inducing an avalanche of the surface layer of disk. The present radiatively driven mass accretion may provide a physical mechanism which enables mass accretion from 100 pc scales down to pc, and it can eventually link to an advection-dominated viscous accretion onto a massive black hole. The radiation-hydrodynamical and self-gravitational instabilities of the disk are briefly discussed. In particular, the radiative acceleration possibly builds up a dusty wall, which provides another version for the formation of an obscuring torus.

The innermost regions of quasars can be resolved by a gravitational-lens {\lq}telescope{\rq} on scales down to a few AU. For the purpose, X-ray observations are most preferable, because X-rays originating from the innermost regions, can be selectively amplified by microlensing due to the so-called `caustic crossing'. If detected, X-ray variations will constrain the size of the X-ray emitting region down to a few AU. The maximum attainable resolution depends mainly on the monitoring intervals of lens events, which should be much shorter than the crossing time. On the basis of this idea, we performe numerical simulations of microlensing of an optically-thick, standard-type disk as well as an optically-thin, advection-dominated accretion flow (ADAF). Calculated spectral variations and light curves show distinct behaviors, depending on the photon energy. X-ray radiation which is produced in optically thin region, exhibits intensity variation over a few tens of days. In contrast, optical-UV fluxes, which are likely to come from optically thick region, exhibit more gradual light changes, which is consistent with the microlensing events so far observed in Q2237+0305. Currently, Q2237+0305 is being monitored in the optical range at Apache Point Observatory. Simultaneous multi-wavelength observations by X-ray sattelites (e.g., ASCA, AXAF, XMM) as well as HST at the moment of a microlens event enable us to reveal an AU scale structure of the central accretion disk around black hole.