WorksPhysics and Astronomy
Gravitational Light Deflection via Fractional Calculus and Chebyshev Polynomial Approximation
Abstract
We develop a fractional-calculus-based framework for evaluating gravitational light deflection in static, spherically symmetric spacetimes. By introducing a suitable transformation of the radial coordinate, the deflection integral is recast into an Abel-type integral equation, from which the associated kernel is reconstructed using fractional integration. The leading singular behaviour near the photon sphere is isolated analytically, and the remaining regular part is approximated by a finite Chebyshev polynomial expansion. This combination retains the correct strong-field behaviour while providing an efficient numerical representation of the deflection integral. We demonstrate the accuracy of the method using the Schwarzschild spacetime as a benchmark and compare the resulting deflection angles with the standard strong-deflection approximation and a two-term weak-deflection expansion. We further apply the method to calculate strong-field lensing observables, including the limiting angular position, angular separation, and relative magnification of relativistic images, for the supermassive black holes Sgr A*. The results show that the proposed approach provides an accurate and flexible means of evaluating gravitational deflection angles across a broad range of closest-approach radii. The framework can be extended to other static, spherically symmetric black-hole geometries, including solutions arising in modified theories of gravity.
Introduction
Gravitational lensing is one of the important predictions of general relativity and provides a powerful means of investigating the geometry of spacetime in the vicinity of compact objects. According to general relativity, the gravitational field of a massive body curves spacetime, causing the propagation of light to deviate from its otherwise straight-line trajectory. Although the deflection of light in the weak-field regime has been extensively investigated, the strong-field regime offers a richer structure in which photons can undergo substantial deflection before reaching a distant observer. In particular, when a light ray passes sufficiently close to a black hole, its trajectory can approach an unstable circular null orbit, commonly known as the photon sphere. As a result, the deflection angle increases without bound as the closest-approach radius approaches the photon sphere, giving rise to an infinite sequence of relativistic images [1, 2, 3].
The study of strong-field gravitational lensing is important for understanding the spacetime geometry around black holes and for extracting information about their physical properties from observable lensing signatures. Several analytical approaches have been developed to describe the propagation of light rays in the strong-field regime. Among them, the strong-deflection-limit formalism provides a particularly useful description of the logarithmic divergence of the deflection angle near the photon sphere [4, 5]. This approach has been widely applied to different black-hole spacetimes and has enabled the calculation of observables associated with relativistic images [3, 6, 7]. However, the standard strong-deflection expansion is an asymptotic approximation constructed for trajectories close to the photon sphere. Its accuracy may therefore deteriorate when it is applied to trajectories with larger closest-approach radii. This motivates the development of alternative methods capable of evaluating the deflection angle accurately over a broader range of impact parameters.
In this work, we develop a numerical framework for gravitational light deflection by combining fractional calculus with Chebyshev polynomial approximation. We consider a general static, spherically symmetric spacetime described by
where , , and are functions of the radial coordinate. Starting from the null geodesic equations, the deflection angle is expressed as an integral involving the metric functions and the closest-approach radius. By introducing a suitable transformation of the radial coordinate, we recast this integral into an Abel-type integral equation. The corresponding kernel can then be reconstructed using fractional integration [8, 9].
A central difficulty in evaluating the resulting integral is the singular behaviour of its kernel near the photon sphere. Rather than approximating the complete kernel directly, we isolate its leading singular contribution analytically and subtract it to obtain a regularised function. The regularised part is subsequently approximated by a finite expansion in Chebyshev polynomials. The Chebyshev–Gauss nodes provide an efficient means of determining the expansion coefficients, while the analytically extracted singular term preserves the correct be- @body haviour in the strong-field limit. This procedure combines the advantages of an analytical treatment near the photon sphere with an accurate numerical approximation over a finite radial interval. Chebyshev polynomial expansions are particularly useful in this context because of their favourable approximation properties and their widespread application in spectral numerical methods [10, 11].
To examine the accuracy and applicability of the proposed framework, we consider the Schwarzschild black hole as a benchmark. We compare the resulting deflection angles with those obtained from the standard strong-deflection-limit approximation and a two-term weak-deflection expansion. We then use the calculated deflection angle to investigate strong-field lensing observables, including the limiting angular position of relativistic images, their angular separation, and their relative magnifications. As illustrative applications, we consider the supermassive black holes Sgr A* [12].
The remainder of this paper is organised as follows. In (§ 2), we present the general formulation of gravitational light deflection in static, spherically symmetric spacetimes. In (§ 3), we develop the fractional-calculus and Chebyshev polynomial framework for evaluating the deflection integral, including the analytical treatment of the singular behaviour near the photon sphere. In (§ 6), we apply the proposed method to the Schwarzschild black hole and examine its numerical accuracy by comparing the resulting deflection angles with standard strong- and weak-field approximations. In (§ 7), we investigate the strong-field lensing observables derived from the Chebyshev deflection angle, with particular applications to Sgr A*. Finally, Sec. (VI) summarises our main results and discusses possible directions for future research.
Construction
We first briefly discuss the construction of the light-deflection angle for the general static, spherically symmetric spacetime. We first consider a general static, spherically symmetric spacetime described by the line element
where
and , , and are functions of the radial coordinate . Owing to spherical symmetry, the photon trajectory can be restricted, without loss of generality, to the equatorial plane, . Consequently, and , such that
The motion of a photon in a curved spacetime is governed by the null geodesic equations. A convenient way to obtain these equations is through the variational formulation of geodesic motion. For a spacetime characterized by the metric tensor , the geodesic Lagrangian is defined as
where denotes the spacetime coordinates and a dot represents differentiation with respect to an affine parameter , i.e., . For the static, spherically symmetric spacetime considered above, and restricting the motion to the equatorial plane , Eq. (4) takes the explicit form
Since the trajectory under consideration corresponds to a photon, the spacetime interval along the trajectory vanishes, . Consequently, the null condition requires
The physical trajectory is obtained by extremizing the corresponding action,
Applying the principle of stationary action, , leads to the Euler–Lagrange equations for each generalized coordinate ,
These equations are equivalent to the geodesic equations and completely determine the photon trajectory once the appropriate initial conditions are specified [6, 13]. Moreover, because the metric coefficients depend only on the radial coordinate , the coordinates and are cyclic coordinates. Their corresponding Euler–Lagrange equations therefore give rise to two conserved quantities along the photon trajectory, which may be identified with the conserved energy and angular momentum, respectively. These constants of motion will be used below to obtain the radial equation and, subsequently, the expression for the light-deflection angle.
Since the metric functions , , and depend only on the radial coordinate , the Lagrangian has no explicit dependence on and . Hence, and are cyclic coordinates. According to Noether’s theorem, the corresponding time-translation and rotational symmetries lead to the conservation of the photon energy and angular momentum , respectively. The associated conserved quantities are
or, equivalently,
Substituting Eq. (10) into the null geodesic condition given by Eq. (6), we obtain the radial equation of motion for the photon as
Using and , the variation of the azimuthal angle with respect to the radial coordinate is given by
Introducing the impact parameter
and factoring from the term inside the square root, the above expression reduces to
At the point of closest approach, , the radial velocity of the photon vanishes, i.e., . Applying this condition to the radial equation of motion, we obtain the relation between the impact parameter and the closest-approach distance as
As the photon approaches the compact object from infinity, the radial coordinate decreases from to the point of closest approach , corresponding to the incoming branch of the trajectory. After reaching , the photon moves outward from to , which represents the outgoing branch. Owing to the symmetry of the trajectory about the point of closest approach, the magnitudes of the angular changes along the incoming and outgoing branches are equal. Therefore, the total change in the azimuthal angle is
Therefore, the deflection angle of the photon is obtained by subtracting the corresponding angular change for an undeflected trajectory from the total change in the azimuthal angle. Thus,
where denotes the deflection angle, while the total change in the azimuthal angle is given by
Figure 1. The photon approaches the compact object from infinity, reaches its closest-approach distance , and subsequently escapes to infinity.
In principle, once the integral is evaluated, the corresponding deflection angle can be determined directly [2, 5, 6]. However, an analytical evaluation of this integral is, in general, nontrivial. In particular, the integrand exhibits a singular behavior at the lower limit , while the upper integration limit extends to infinity. Various approximation schemes have been developed to address these difficulties, including the weak-deflection expansion and the formalism introduced by Bozza [5, 7]. Although these approaches provide powerful tools for studying gravitational deflection, the accuracy of an approximate evaluation of the integral and the associated truncation error are not always controlled explicitly within the approximation itself. Motivated by this issue, in the following section we develop an alternative approximation scheme based on fractional calculus together with a Chebyshev–minimax treatment. This construction is intended to provide a systematic approximation of the deflection integral while allowing the corresponding approximation error to be quantified and controlled.
Fractional–Chebyshev Treatment of the Deflection Integral
In this section, we develop an approximation scheme for evaluating the deflection angle given by Eq. (18), employing fractional calculus together with a Chebyshev–minimax treatment. Since the deflection angle is expressed in terms of , the problem reduces to obtaining an appropriate approximation for the integral .
Theorem 1. The integral governing the deflection angle admits a representation in terms of a Weyl fractional integral.
Proof. We begin by introducing the function
At the point of closest approach, , we define
Using the relation between the impact parameter and the closest-approach distance, we immediately obtain
We now employ the transformation to recast the deflection integral into a form suitable for a Weyl fractional-integral representation.[6] We now change the integration variable from to by using . Accordingly,
The transformation is well defined in the region where , i.e., away from the photon sphere. Under this transformation, corresponds to . The integral can therefore be recast as
Introducing
the above expression takes the compact form
To identify the fractional structure of Eq. (26), we recall the right-sided Weyl fractional integral.[8, 14] Let and let be such that the following integral exists.
exists. The right-sided Weyl fractional integral of order is defined by
where is the Euler Gamma function. In particular, for , the kernel possesses an integrable singularity at the lower limit , provided that is sufficiently regular in a neighbourhood of . Convergence at the upper limit further requires an appropriate decay of as . Setting
and using
the Weyl fractional integral becomes
Comparison with Eq. (26) therefore yields
Hence, the deflection integral is represented by a right-sided Weyl fractional integral of order , completing the proof.
Corollary 1. The Weyl fractional representation of the deflection integral admits an inversion formula for reconstructing the metric-dependent function from the deflection angle.
Proof. The right-sided Weyl fractional integrals satisfy the semigroup property [9, 14]
provided that the corresponding operators and integrals are well defined. In particular,
where reduces to ordinary right-sided integration. From the fractional representation obtained above, we define
so that
Introducing the right-sided Weyl fractional derivative , which acts as a left inverse of under the appropriate regularity and decay conditions, we apply it to Eq. (35) to obtain
For sufficiently regular and decaying , the right-sided Weyl fractional derivative of order is given by [9, 14]
Since
the function can be determined once the deflection angle is known. Equation (37) therefore provides an inversion of the deflection integral, yielding
Thus, knowledge of the deflection angle as a function of the closest-approach parameter determines the corresponding metric-dependent combination through fractional inversion, without requiring a weak- or strong-deflection expansion.
The inverse problem formulated above was previously developed and solved in our earlier work by Ghosh, Debnath, et al. [15].
We now examine the singularity structure of the function , which plays a central role in the subsequent analysis. In particular, the location and nature of its singularities determine the analytic behaviour of the transformed deflection integral and are therefore essential for constructing a reliable approximation scheme.
We first define the photon-sphere radius . For a static, spherically symmetric spacetime, the photon-sphere radius is determined by the condition [5, 6, 16]
where a prime denotes differentiation with respect to the radial coordinate . Differentiating with respect to , we obtain
Using the photon-sphere condition in Eq. (40), the quantity in square brackets vanishes at . Consequently,
We define the corresponding value of the transformed radial coordinate as
Since , defined in Eq. (39), contains in the denominator, the vanishing of at the photon sphere implies that becomes singular as .
An important distinction must be made between the radius of closest approach and the photon-sphere radius . For the scattering trajectories considered here, namely photons arriving from infinity, reaching a radial turning point, and subsequently escaping back to infinity, the physically relevant branch satisfies
Thus, represents the limiting value of the closest approach for the scattering branch and is approached only from above,
The different possibilities are illustrated by the photon trajectories shown in Figure 2. When , represented by the blue trajectory in Figure 2, the photon remains far from the photon sphere and undergoes only a small change in its propagation direction. This corresponds to the usual weak-deflection regime.
Figure 2. Schematic illustration of photon trajectories in the vicinity of a black hole and its photon sphere. The blue trajectory represents weak deflection, for which the closest-approach radius satisfies . The orange trajectory illustrates strong deflection as , where the photon winds around the black hole before escaping to infinity. The purple trajectory represents a captured photon that crosses the critical photon-sphere orbit and does not return to infinity.
As the closest-approach radius decreases towards the photon-sphere radius, , the behaviour changes significantly. This case is represented by the orange trajectory in Figure 2. The photon approaches the unstable circular null orbit and spends an increasingly long interval in its vicinity before eventually escaping to infinity. Consequently, it can wind around the black hole one or more times, and the accumulated azimuthal angle increases without bound. For a nondegenerate unstable photon sphere, this behaviour gives rise to the characteristic logarithmic divergence of the deflection angle [2, 5, 7].
At the critical value
the radial turning point becomes degenerate, and the photon follows the unstable circular null orbit at . This limiting orbit separates the scattering and capture branches. On the capture side of this critical orbit, represented by the purple trajectory in Figure 2, the photon has no outer turning point corresponding to the scattering integral considered here and instead propagates towards the black-hole interior. Therefore, , as defined for a photon that returns to infinity, does not describe this captured trajectory.
The above behaviour also provides a physical interpretation of the singularity of at . As , the photon spends progressively more time near the unstable circular null orbit, producing the multiple-winding behaviour illustrated by the orange trajectory in Figure 2. The divergence of the deflection angle is therefore directly associated with the photon-sphere structure rather than being merely an artefact of the change of variables.
It is important, however, to distinguish this singularity from the turning-point singularity appearing at the lower limit of the deflection integral. For every scattering trajectory with , the transformed integral contains the kernel . The singularity at is an integrable square-root singularity associated with the ordinary radial turning point and yields a finite value of for fixed . In contrast, the singularity of at originates from the geometric condition and is therefore a property of the spacetime geometry itself. As the lower limit approaches this geometric singularity, , the combined behaviour of and the Abel kernel produces the divergent deflection characteristic of the photon-sphere limit.
To further characterize the photon-sphere singularity, we now examine the local behaviour of in the vicinity of . The leading-order behaviour of near this critical point is established by the following lemma.
Lemma 1 (Local behaviour of ). For a nondegenerate photon sphere at , satisfying and , the function exhibits the following leading-order behaviour in the vicinity of :
where
Proof. Since is the photon-sphere radius, it satisfies
We expand about . Since the first derivative vanishes at the photon sphere, the leading nonvanishing contribution is quadratic:
Using and , we obtain
Therefore, on the physically relevant outer branch , we obtain
We next expand in the neighbourhood of the photon sphere. Since , its leading-order behaviour is
Substituting Eq. (52) into Eq. (53), we obtain
Substituting Eq. (54) into the definition of given in Eq. (39), and evaluating the remaining metric functions at , we obtain
Introducing
we therefore arrive at
To express the coefficient explicitly in terms of the metric functions, we differentiate
The first derivative is given by
Differentiating once more, we obtain
At , the photon-sphere condition implies
Consequently, the second term in Eq. (60) vanishes, yielding
This completes the proof.
We now introduce a fixed radius , with . Motivated by Lemma 1, we separate the leading singular contribution from by defining
From Lemma 1, it follows that remains bounded as . Moreover, assuming sufficient regularity of the metric functions, admits a continuous extension to . It is important to emphasise that the decomposition
is an exact algebraic identity rather than an approximation. Consequently, the leading singular contribution is explicitly isolated without discarding any information about the original function .
Having established the local singular structure of , we now proceed to reformulate the deflection integral in a form that separates the dominant photon-sphere contribution from the remaining regular and asymptotic contributions. To this end, we introduce the matching point , where , and partition the integration domain at . Employing the exact decomposition of derived above, the deflection integral, for , can be expressed as
where
and
This representation constitutes an exact partition of the original integral, with no approximation introduced at this stage. The first contribution explicitly isolates the leading photon-sphere singularity and admits an analytical evaluation. The second contains the bounded remainder and is therefore suitable for a systematic Chebyshev–minimax treatment, whereas the third accounts for the asymptotic region beyond the matching point. Such a decomposition enables the distinct analytical features of the deflection integral to be treated independently, providing a natural basis for constructing a controlled approximation scheme.
We first consider the singular contribution , which encodes the dominant behavior of the deflection integral in the vicinity of the photon sphere. By isolating the leading algebraic singularity of , this contribution admits an exact analytical evaluation without invoking any approximation. Explicitly,
The integral is elementary and can be expressed in logarithmic form through the identity
Consequently, the singular contribution reduces exactly to
The logarithmic structure explicitly characterizes the asymptotic behavior of the deflection integral in the critical limit . In particular, the divergent contribution is governed by , whereas the remaining terms approach finite values. Thus, the photon-sphere divergence is recovered analytically, with its leading coefficient determined entirely by the local geometric quantities entering [5, 7].
In the present approximation scheme, we employ a Chebyshev–minimax approach to construct a systematic approximation of the regular contribution to the deflection integral. Before developing this construction, we briefly review the relevant mathematical definitions and fundamental properties of Chebyshev polynomials and minimax approximation theory [10, 11].
Chebyshev Polynomials: Definitions and Properties
The Chebyshev polynomials of the first kind constitute a sequence of orthogonal polynomials on the interval , defined through the trigonometric representation [10, 11]
Introducing the transformation , with , and employing the trigonometric identity
we obtain the three-term recurrence relation
subject to the initial conditions and . This recurrence establishes inductively that is a polynomial of degree . In particular, the first few members of the sequence are
a. Property 1: Uniform boundedness. From the trigonometric representation of the Chebyshev polynomials and the bound , it follows immediately that
for every nonnegative integer . Thus, the Chebyshev polynomials are uniformly bounded in absolute value, independently of their degree. As illustrated in Figure 3, each polynomial oscillates entirely within the interval .
Figure 3. Plots of the first six Chebyshev polynomials , with , over the interval .
b. Property 2: Equioscillation. A fundamental characteristic of Chebyshev polynomials is their equioscillatory structure. Setting , the extremal values of occur when . Consequently, for , the extremal points are given by
at which
Hence, attains the values and alternately at distinct points of , as illustrated in Figure 3. This alternating extremal structure underlies the Chebyshev alternation theorem, which characterizes the polynomial of best uniform approximation. Specifically, for a continuous function on , a polynomial of degree at most is the minimax approximant if and only if the error attains its maximum absolute magnitude with alternating signs at at least ordered points, provided the error is nonzero. The polynomial provides a canonical example of such an alternating structure [11, 17].
c. Property 3: Roots and Chebyshev nodes. The roots of the Chebyshev polynomials follow directly from their trigonometric representation. Setting , with , yields . Consequently, for , the polynomial possesses distinct roots in the interval , given by
These roots define the Chebyshev–Gauss nodes, which provide a nonuniform distribution of interpolation points over . Unlike equally spaced nodes, they become increasingly concentrated near the endpoints , while remaining relatively sparse near the center. This distribution follows from the projection of uniformly spaced angular points onto the real axis.
In the present approximation scheme, these nodes will be employed to sample the regularized function , after mapping the physical interval onto .
Their endpoint clustering is particularly advantageous for resolving variations near the boundaries and constructing stable high-order polynomial approximations [10, 11].
d. Property 4: Orthogonality. The Chebyshev polynomials of the first kind form an orthogonal sequence on with respect to the weight function . More precisely, they satisfy the orthogonality relation
This property follows directly from the substitution , which transforms the integral into
The orthogonality relation provides a natural basis for expanding sufficiently regular functions in Chebyshev series. In particular, a function can be represented, under suitable convergence conditions, as
where the coefficients are determined by
with denoting the Kronecker delta. This orthogonal structure facilitates the systematic construction of polynomial approximations to the regularized function .
e. Property 5: Extremal minimax property. A fundamental feature of Chebyshev polynomials is their extremal property with respect to the uniform norm. Among all monic polynomials of degree , the rescaled Chebyshev polynomial uniquely minimizes the maximum absolute deviation on . More precisely,
where denotes the space of polynomials of degree at most , and
Equivalently, every monic polynomial of degree satisfies
with equality if and only if .
This extremal behaviour is a direct consequence of the equioscillation property. The alternating extrema of distribute the maximum deviation uniformly across distinct points, preventing any further reduction of the uniform norm while preserving the leading coefficient. This principle forms the mathematical foundation of the Chebyshev alternation theorem, which characterizes the best uniform polynomial approximation through the alternating extrema of its error function. Consequently, Chebyshev polynomials provide a natural basis for the minimax treatment of the regularized function in the present formulation [11, 17].
Having established the definition and fundamental properties of Chebyshev polynomials, we now apply these results to approximate the regularized function arising in the lensing problem. As a first step, we map its domain onto the canonical Chebyshev interval .
Chebyshev–Minimax Approximation of the Regularized Lensing Function
a. Compactified variable. To implement the Chebyshev approximation, we first introduce the inverse radial coordinate . For a fixed matching radius , the physical interval is mapped onto , where
We then map this interval affinely onto the canonical Chebyshev domain through
Accordingly,
Since reverses the radial ordering, the endpoint corresponds to the photon sphere, whereas corresponds to the outer matching radius. The subsequent transformation is an order-preserving affine rescaling, placing the physical domain precisely on the standard interval required for the Chebyshev construction.
b. Regularized function in the Chebyshev variable. The regularized function was previously introduced as
To express this function on the standard Chebyshev interval, we define
The corresponding sequence of transformations is therefore
No new physical function is introduced through this transformation. The quantities and represent the same regularized function expressed in different variables. The former notation is introduced solely to formulate the approximation on the canonical Chebyshev interval .
c. Chebyshev expansion. Since is regular on the interval under consideration, it may be expanded in the Chebyshev basis as
where
For practical calculations, the infinite expansion is truncated at a finite degree , yielding
Thus, the approximation of reduces to the determination of the coefficients .
d. Determination of the Chebyshev coefficients. The coefficients follow from the orthogonality relation
Multiplying the Chebyshev expansion by and integrating over , we obtain
By orthogonality, all terms with vanish, and hence
where denotes the Kronecker delta. Explicitly,
whereas, for ,
e. Fourier–cosine representation. Introducing
together with
and using
the coefficient formula becomes
Thus, the Chebyshev expansion can be interpreted as a Fourier–cosine expansion under the transformation .
f. Chebyshev–Gauss sampling. For numerical implementation, we evaluate at the Chebyshev–Gauss nodes
Equivalently,
For each node , the corresponding physical quantities are obtained successively through
so that
Since , neither endpoint is sampled directly. In particular, the photon-sphere endpoint is avoided.
g. Discrete determination of the coefficients. The -point Gauss–Chebyshev quadrature rule is
Applying this relation to gives
Since
the coefficients may equivalently be written as
This expression is precisely a discrete cosine transform of the sampled values .
Equivalently, the coefficients may be regarded as the solution of the system
Thus, the degree- Chebyshev approximation is completely determined once the regularized lensing function has been evaluated at the Chebyshev–Gauss nodes.
h. Approximation Accuracy and Near-Minimax Property. The accuracy and near-optimal character of the present Chebyshev approximation are supported by two classical results of approximation theory. Jackson’s theorem guarantees that the best polynomial approximation converges systematically as the polynomial degree increases, with the convergence rate controlled by the smoothness of the approximated function. On the other hand, the Chebyshev equioscillation theorem states that the minimax polynomial is characterized by an error that attains equal maximum magnitude with alternating signs at at least points of . Since the regularization removes the leading photon-sphere singularity from and the Chebyshev basis naturally possesses this extremal oscillatory structure, the resulting Chebyshev approximation provides a stable and near-minimax representation of the regularized lensing function.
i. Reconstruction in the original variable. Once has been constructed, we return to the original variable and define
Hence,
The regular contribution to the deflection integral therefore takes the approximate form
Combining this result with the analytically isolated photon-sphere contribution and the far-field contribution, the degree- representation of the complete deflection integral becomes
The three terms have distinct roles: the first contains the photon-sphere singularity and is known analytically [Eq. (70)]; the second contains the regularized near-field contribution and is treated by the Chebyshev approximation; and the third is a smooth far-field integral that may be evaluated by standard quadrature. In this manner, the singular and regular structures of the lensing integral are separated before any numerical approximation is introduced.
Application to the Schwarzschild Black Hole
In this section, we calculate the gravitational deflection angle of light in Schwarzschild spacetime using the fractional-calculus and Chebyshev approximation method developed in the preceding section. We derive the relevant expressions, evaluate the deflection angle, and examine the accuracy of the proposed method by comparing its results with direct numerical integration.
Step 1: Determination of the Photon-Sphere Radius
We consider the Schwarzschild spacetime, setting and measuring all lengths in units of the black hole mass . Accordingly, we use the dimensionless radial coordinate , denoted simply by , so that corresponds to in physical units. The metric functions are
The photon-sphere radius is determined by the condition , where
Differentiating with respect to , we obtain
Setting gives
Since the physical domain is , the only admissible solution is
Thus, the Schwarzschild photon sphere is located at in dimensionless units, corresponding to when physical units are restored.
Step 2: Determination of
The coefficient (), which characterises the leading-order behaviour of near the photon sphere, is given by
In dimensionless units, we set . The photon-sphere radius is therefore . From the Schwarzschild metric, we obtain
Substituting the above values, we find
Hence,
Step 3: Verification of Lemma 1
We now verify Lemma 1, which states that near the photon sphere,
For the Schwarzschild spacetime, using dimensionless units with , we have
The function can be expressed in terms of the radial coordinate as
To examine the behaviour near the photon sphere, we expand about . Since and , the leading-order expansion is
Consequently,
Taking the limit of as , we obtain
Using the result from Step 2, this limit is precisely . Therefore,
Thus, Lemma 1 is verified for the Schwarzschild spacetime in dimensionless units, with and .
Step 4: Chebyshev Approximation of the Regularised Function
In this step, we construct a Chebyshev polynomial approximation to the regularised function . We choose the matching radius , while the photon-sphere radius is . To map the interval onto the standard Chebyshev interval , we introduce the transformation
Substituting and , we obtain
The regularised function is defined by subtracting the leading singular contribution from :
where and . Using the transformation above, we express the regularised function in terms of the Chebyshev coordinate . For the Schwarzschild spacetime, this gives
We then approximate on by a truncated Chebyshev series of degree :
where denotes the Chebyshev polynomial of the first kind of degree , and are the corresponding expansion coefficients. Returning to the variable , the approximation is written as
This approximation will be used to evaluate the regular contribution to the deflection integral. To determine the expansion coefficients , we employ the Chebyshev–Gauss nodes,
The coefficients are calculated using the discrete Chebyshev expansion,
while the zeroth coefficient is given by
Choosing the truncation order , we obtain the following coefficients: Thus, the degree-12 Chebyshev approximation to the regularised function is
This polynomial approximation is subsequently used to evaluate the regular part of the deflection integral. From Table I, we observe a rapid decay in the magnitudes of the Chebyshev coefficients, with for , apart from small numerical variations. This behaviour suggests that the truncated expansion captures the regularised function reasonably well at relatively low polynomial order. We now transform the Chebyshev approximation back to the radial coordinate using the
| 0 | |
| 1 | |
| 2 | |
| 3 | |
| 4 | |
| 5 | |
| 6 | |
| 7 | |
| 8 | |
| 9 | |
| 10 | |
| 11 | |
| 12 |
Table I. Chebyshev expansion coefficients for .
inverse mapping and compare it with the exact regularised function . The comparison is presented in Figure 4, where the two curves overlap closely, indicating good agreement between the Chebyshev approximation and the exact function. The corresponding absolute error, defined as
Figure 4. Comparison of the exact regularised function and its Chebyshev polynomial approximation of degree as functions of the radial coordinate .
is shown in Figure 5 and (Table II) to assess the accuracy of the approximation across the chosen radial interval.
| 3.01 | |
| 4.01 | |
| 5.01 | |
| 6.01 | |
| 7.01 | |
| 8.01 | |
| 9.01 | |
| 10.01 | |
| 11.01 | |
| 12.01 | |
| 13.01 | |
| 14.01 | |
| 15.01 | |
| 16.01 | |
| 17.01 | |
| 18.01 | |
| 19.01 | |
| 20.01 | |
| 21.01 | |
| 22.01 | |
| 23.01 | |
| 24.01 |
Table II. Absolute error between the exact and Chebyshev-approximated regularised functions.
Figure 5. Absolute error between the exact regularised function and its Chebyshev polynomial approximation
As shown in Figure 5 and (Table II), the absolute error is of the order of throughout the sampled radial interval. This indicates excellent numerical agreement between the exact regularised function and its degree-12 Chebyshev approximation, demonstrating the high accuracy of the approximation over the interval considered.
Step 5: Calculation of the Deflection Angle
In this step, we calculate the deflection angle for the Schwarzschild black hole using the results obtained in the preceding steps. Substituting the expressions for the singular, regular, and far-field contributions into Eq. (117), we evaluate the total deflection integral and subtract to obtain the deflection angle. The resulting deflection angle is presented in Figure 6. For comparison, we also plot the weak-deflection approximation and the result obtained using Bozza’s method. As shown in Figure 6, our method provides a consistent description of the deflection angle over the radial interval considered. In contrast, the Bozza approximation may yield unphysical negative values sufficiently far from the photon sphere because it is based on a strong-deflection expansion designed for trajectories whose closest-approach radius approaches the photon-sphere radius, . Such behaviour outside its intended regime of validity should not be interpreted as a physical prediction. The weak-deflection approximation is also included to compare the results in the regime of relatively large closest-approach radii. A quantitative comparison of the deflection angles obtained using our method, the weak-deflection approximation, and Bozza’s method is provided in (Table III). The table shows that the Chebyshev-based results agree closely with Bozza’s strong-deflection approximation near the photon sphere. As increases, the two results gradually diverge, reflecting the limited validity of the strong-deflection expansion away from the photon sphere. For , 15, and 24, Bozza’s approximation yields negative deflection angles, whereas the Chebyshev results remain positive and approach zero as increases. The two-term weak-deflection approximation also approaches the Chebyshev result at large , as expected in the weak-field regime.
| (Chebyshev) | (Bozza) | (weak, 2 terms) | ||
| 3.001 | 5.19615 | 15.2079 | 15.2079 | 1.20613 |
| 3.01 | 5.19624 | 10.6108 | 10.6107 | 1.20611 |
| 3.05 | 5.19823 | 7.42779 | 7.42644 | 1.20548 |
| 3.1 | 5.20411 | 6.08632 | 6.08173 | 1.20362 |
| 3.2 | 5.22558 | 4.78849 | 4.77358 | 1.19690 |
| 3.5 | 5.34634 | 3.20612 | 3.14357 | 1.16034 |
| 4.0 | 5.65685 | 2.18410 | 2.02269 | 1.07526 |
| 5.0 | 6.45497 | 1.37674 | 1.01751 | 0.902421 |
| 8.0 | 9.23760 | 0.669155 | −0.148916 | 0.571071 |
| 15.0 | 16.1126 | 0.307264 | −1.14258 | 0.293632 |
| 24.0 | 25.0672 | 0.181547 | −1.74157 | 0.178320 |
Table III. Comparison of the deflection angle obtained using the Chebyshev approximation, Bozza’s strong-deflection approximation, and the two-term weak-deflection approximation for the Schwarzschild black hole. All quantities are expressed in dimensionless units with .
Figure 6. Caption
Strong-field lensing observables from the Chebyshev deflection angle
Lens equation and relativistic images
We consider a gravitational lensing configuration in which a source, a black hole acting as the lens, and an observer are nearly aligned. The observer–lens, lens–source, and observer–source distances are denoted by , , and , respectively, with . We assume that these distances are much larger than the characteristic length scale of the black hole, allowing us to work in the small-angle approximation. For a photon whose trajectory has a closest-approach radius , the impact parameter in the Schwarzschild spacetime is
The corresponding deflection angle, , is evaluated using the Chebyshev-based regularisation method developed in (§ 2).
The image position and source position are related by the lens equation [5, 18]
where denotes the number of complete revolutions made by the photon around the black hole. Since the deflection angle diverges logarithmically as the impact parameter approaches the critical value,
the lens equation admits an infinite sequence of relativistic images. These images accumulate at the limiting angular position
Image positions and magnifications
For a source located exactly on the optical axis, , the image positions are determined by
For a small but nonzero source angle, we expand the deflection angle about the corresponding aligned-image solution. Defining
the first-order expansion of the lens equation gives
The magnification of an image is determined by the Jacobian of the lens mapping. In the small-angle approximation, its signed value is
The absolute value gives the observable flux magnification. Using the linearised lens equation, we obtain
The standard strong-field lensing observables are the limiting angular position , the angular separation between the outermost relativistic image and the limiting position, and the magnitude difference between the first relativistic image and the remaining images [5, 19]:
Evaluation using the Chebyshev deflection angle
In the conventional strong-deflection approach, the deflection angle is approximated by its leading logarithmic form,[5, 19]
where, for the Schwarzschild spacetime,
This approximation yields the standard expressions
In the present approach, we instead evaluate the deflection angle directly from the Chebyshev approximation, without imposing a logarithmic functional form over the entire domain. The relativistic-image positions and magnifications are obtained through the following procedure:
For each winding number , solve
numerically for . To improve numerical stability near the photon sphere, we introduce the auxiliary variable and use an initial estimate motivated by the strong-deflection expansion.
Calculate the corresponding impact parameter,
Because approaches rapidly as increases, direct subtraction may cause a loss of numerical precision. We therefore evaluate their difference using the identity
Evaluate the derivative numerically from the Chebyshev deflection function. The chain rule gives
Substitute the resulting image positions and derivatives into Eqs. (147)–(149) to calculate the angular positions, magnifications, and strong-field observables.
This procedure allows the lensing observables to be evaluated directly from the numerically reconstructed deflection angle. It also provides a means of assessing the accuracy of the standard logarithmic approximation in the regime where that approximation is expected to be valid.
Results
We illustrate the method using the supermassive black hole at the Galactic centre, Sgr A*[20, 21], with
For this configuration, the limiting angular position is . The first four relativistic images calculated using the degree-12 Chebyshev approximation with matching radius are listed in Table IV.
| () | |||||
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 |
Table IV. First four relativistic images of Sgr A* calculated using the Chebyshev deflection angle with and .
The first four relativistic images approach the limiting angular position as the winding number increases, while their magnifications decrease rapidly. This trend is consistent with the accumulation of relativistic images near the critical impact parameter[2, 3, 19].
Table IV compares the strong-field observables obtained from the Chebyshev deflection angle with those predicted by the standard logarithmic approximation.
| Observable | Chebyshev | Bozza | Difference |
| (Sgr A*) | 0 | ||
| (Sgr A*) | |||
| (M87*) | 0 | ||
| (M87*) | |||
Table V. Comparison of the strong-field lensing observables obtained using the Chebyshev method and the standard strong-deflection approximation.
The comparison indicates close agreement between the Chebyshev-based results and the standard strong-deflection predictions for the Schwarzschild spacetime. The relative difference in the angular separation is approximately , while the difference in the magnitude observable is approximately mag. These results demonstrate that the Chebyshev procedure reproduces the standard strong-field observables with high accuracy for this benchmark spacetime, while allowing the deflection angle to be evaluated without assuming a logarithmic form throughout the calculation.
The angular separation of the first relativistic image from the limiting position is approximately , and the image magnification is extremely small. Resolving these individual relativistic images therefore remains challenging with current very-long-baseline interferometric observations[19, 20, 21]. By contrast, the limiting angular position provides the characteristic angular scale of the black-hole shadow and is more directly connected to current horizon-scale imaging observations[20, 21].
Conclusion
In this work, we have developed a fractional-calculus-based framework for evaluating the gravitational deflection of light in static, spherically symmetric spacetimes. By introducing a suitable transformation of the radial coordinate, the deflection integral was recast into an Abel-type integral equation, allowing the underlying kernel to be reconstructed through fractional integration. A central feature of the proposed approach is the analytical isolation of the leading singular contribution near the photon sphere. Subtracting this contribution yields a regularised function that can be efficiently approximated using Chebyshev polynomials.
(i) We have employed Chebyshev–Gauss nodes to construct a truncated polynomial approximation of the regularised kernel. This procedure provides an accurate numerical representation over a finite radial interval while retaining the analytically known singular behaviour. The Schwarzschild spacetime was considered as a benchmark to examine the accuracy and reliability of the method. The numerical comparison between the exact regularised function and its Chebyshev approximation demonstrates excellent agreement Figure 4Figure 5, with the absolute error reaching approximately for the degree-12 approximation considered (II).
(ii) Using the reconstructed kernel, we evaluated the deflection angle by combining the singular, regular, and far-field contributions. The resulting deflection angle was compared with the standard strong-deflection approximation of Bozza and the two-term weak-deflection approximation Figure 6. The Chebyshev-based results reproduce the expected behaviour near the photon sphere and remain well behaved at larger closest-approach radii. In particular, the comparison illustrates the limitations of extrapolating the logarithmic strong-deflection approximation beyond its intended regime of validity.
(iii) Furthermore, we applied the resulting deflection angle to investigate strong-field gravitational lensing observables, including the limiting angular position, the angular separation of relativistic images, and their relative magnifications. As an illustrative application, we considered the supermassive black holes Sgr A*. For the Schwarzschild benchmark, the observed quantities using the Chebyshev procedure show close agreement with the standard strong-deflection predictions, with a relative difference of approximately in the angular separation and a difference of approximately mag in the magnitude observable.
(iv) The present framework therefore provides an accurate and flexible alternative for evaluating gravitational deflection angles without imposing a logarithmic approximation over the entire domain of integration. Although the Schwarzschild spacetime serves as the benchmark in this work, the formulation is applicable to a broader class of static, spherically symmetric geometries, provided that the relevant photon-sphere structure and asymptotic behaviour are appropriately incorporated. Its extension to black holes in modified theories of gravity and other non-Schwarzschild geometries offers a promising direction for future investigations.
Acknowledgements
AG is thankful to IIEST, Shibpur, India, for providing an Institute Fellowship (SRF).
References
- [1]Einstein, A. Lens-like action of a star by the deviation of light in the gravitational field. Science 84, 506–507 (1936).pubmed.ncbi.nlm.nih.gov/17769014
- [2]Darwin, C. The gravity field of a particle. Proceedings of the Royal Society of London. Series A 249, 180–194 (1959).
- [3]Virbhadra, K. S. & Ellis, G. F. R. Schwarzschild black hole lensing. Physical Review D 62, 084003 (2000).DOI
- [4]Bozza, V., Capozziello, S., Iovane, G. & Scarpetta, G. Strong field limit of black hole gravitational lensing. General Relativity and Gravitation 33, 1535–1548 (2001).DOI
- [5]Bozza, V. Gravitational lensing in the strong field limit. Physical Review D 66, 103001 (2002).DOI
- [6]Perlick, V. Exact gravitational lens equation in spherically symmetric and static spacetimes. Physical Review D 69, 064017 (2004).DOI
- [7]Iyer, S. V. & Petters, A. O. Light’s bending angle due to black holes: From the photon sphere to infinity. General Relativity and Gravitation 39, 1563–1582 (2007).DOI
- [8]Gorenflo, R. & Vessella, S. Abel Integral Equations: Analysis and Applications (Springer, Berlin, 1991).
- [9]Podlubny, I. Fractional Differential Equations (Academic Press, San Diego, 1999).
- [10]Boyd, J. P. Chebyshev and Fourier Spectral Methods (Dover Publications, New York, 2001), 2 edn.
- [11]Trefethen, L. N. Approximation Theory and Approximation Practice (SIAM, Philadelphia, 2013).
- [12]Melia, F. & Falcke, H. The supermassive black hole at the galactic center. Annual Review of Astronomy and Astrophysics 39, 309–352 (2001).
- [13]Chandrasekhar, S. The Mathematical Theory of Black Holes (Oxford University Press, Oxford, 1983).
- [14]Samko, S. G., Kilbas, A. A. & Marichev, O. I. Fractional Integrals and Derivatives: Theory and Applications (Gordon and Breach Science Publishers, Yverdon, 1993).
- [15]Ghosh, A. & Debnath, U. An abel-inversion formalism for spacetime metric reconstruction from light deflection. arXiv preprint arXiv:2609.16844 (2026).arxiv.org/abs/2609.16844
- [16]Claudel, C.-M., Virbhadra, K. S. & Ellis, G. F. R. The geometry of photon surfaces. Journal of Mathematical Physics 42, 818–838 (2001).DOI
- [17]Rivlin, T. J. Chebyshev Polynomials: From Approximation Theory to Algebra and Number Theory (John Wiley & Sons, New York, 1990), 2 edn.
- [18]Virbhadra, K. S. & Ellis, G. F. R. Schwarzschild black hole lensing. Physical Review D 62, 084003 (2000).DOI
- [19]Bozza, V. Gravitational lensing by black holes. General Relativity and Gravitation 42, 2269–2300 (2010).DOI
- [20]Event Horizon Telescope Collaboration. First m87 event horizon telescope results. i. the shadow of the supermassive black hole. The Astrophysical Journal Letters 875, L1 (2019).DOI
- [21]Event Horizon Telescope Collaboration. First sagittarius a* event horizon telescope results. i. the shadow of the supermassive black hole in the center of the milky way. The Astrophysical Journal Letters 930, L12 (2022).DOI





