Spherical Bessel functions of the first kind #
The spherical Bessel function of the first kind of order l is given by the Rayleigh formula
j_l(x) = x ^ l * (-(1 / x) d/dx) ^ l (sin x / x).
The function sin x / x is Real.sinc. The iterates of -(1 / x) d/dx are taken with the
operator f ↦ -dslope (deriv f) 0, which maps analytic functions to analytic functions
(AnalyticAt.dslope), so every iterate, and with it every sphericalBesselJ l, is
real analytic on the whole line. Every iterate is moreover even, so its derivative vanishes at the
origin. For such f the operator agrees with -f' x / x for x ≠ 0, and at x = 0 it takes the
value -f''(0), the limit of -f' x / x as x → 0.
Mathlib's Complex.besselJ is the Bessel function of the first kind of complex order and
argument. The spherical functions are its half-integer-order special case: j_l(x) is
√(π / (2 * x)) * Complex.besselJ (l + 1 / 2) x for x > 0 (DLMF §10.47); on the negative
half-line the principal branches of the two half-integer powers make that product -j_l(x). That
development is not the basis here: its analyticity is available on the slit plane
Complex.slitPlane (Complex.analyticAt_besselJ), with the whole-plane form
(Complex.analyticAt_besselJ_int) restricted to integer order, and it is complex-valued, so it
provides neither the whole-line real analyticity of j_l nor its value at the removable
singularity x = 0. The Rayleigh formula above supplies both directly.
Main definitions #
EpsilonEridani.sphericalBesselJ l x: the spherical Bessel functionj_l(x); the Rayleigh formula defining it issphericalBesselJ_deftogether withsphericalBesselJDivPow_def.EpsilonEridani.sphericalBesselJDivPow l x: the even analytic factor(-(1 / x) d/dx) ^ l (sin x / x), withj_l(x) = x ^ l * sphericalBesselJDivPow l x; it isj_l(x) / x ^ lforx ≠ 0(sphericalBesselJDivPow_eq_div_of_ne_zero) and its limit1 / (2 l + 1)‼at the origin.
Main statements #
sphericalBesselJ_zero:j_0(x) = sinc x, andsphericalBesselJ_onegives the closed formj_1(x) = sin x / x ^ 2 - cos x / x.analyticAt_sphericalBesselJ: eachj_lis real analytic on the whole line.sphericalBesselJ_apply_zero:j_0(0) = 1andj_l(0) = 0forl ≠ 0.tendsto_sphericalBesselJ_div_pow_nhdsNE_zero: the leading behaviourj_l(x) / x ^ l → 1 / (2 l + 1)‼at the origin.sphericalBesselJ_neg: the parityj_l(-x) = (-1) ^ l j_l(x).sphericalBesselJDivPow_zeroandsphericalBesselJDivPow_succ:sphericalBesselJDivPow 0issinc, andsphericalBesselJDivPow (l + 1)is-(1 / x) d/dxofsphericalBesselJDivPow l; forx ≠ 0these factors equalj_l(x) / x ^ l.sphericalBesselJDivPow_apply_zero: the value ofsphericalBesselJDivPow lat the origin, the limiting value1 / (2 l + 1)‼ofj_l(x) / x ^ lasx → 0, andhasDerivAt_sphericalBesselJDivPowwithderiv_sphericalBesselJDivPow: its Rayleigh step, the derivative ofsphericalBesselJDivPow lis-xtimessphericalBesselJDivPow (l + 1).sq_mul_sphericalBesselJDivPow_add_two: the three-term recurrence of the factorssphericalBesselJDivPow l, which equalj_l(x) / x ^ lforx ≠ 0, on the whole line; it is the even factor ofmul_sphericalBesselJ_add_twoand stays informative at the origin.mul_sphericalBesselJ_add_two: the three-term recurrence in the order on the whole line, andsphericalBesselJ_add_two_of_ne_zero: its form solved forj_{l + 2}(x), forx ≠ 0.mul_deriv_sphericalBesselJandmul_deriv_sphericalBesselJ_succ': the two derivative relations, which raise and lower the order. They form a ladder pair. They hold on the whole line but degenerate to0 = 0at the origin;hasDerivAt_sphericalBesselJ_of_ne_zeroandhasDerivAt_sphericalBesselJ_succ'_of_ne_zeroare their forms forx ≠ 0, and the derivative at the origin ishasDerivAt_sphericalBesselJ_apply_zero.hasDerivAt_sphericalBesselJwithderiv_sphericalBesselJ: the derivative ofj_lon the whole line, the origin included, by the product rule forj_l(x) = x ^ l * sphericalBesselJDivPow l x. At order zero it is the ladder relationj_0' = -j_1on the whole line,hasDerivAt_sphericalBesselJ_zero.hasDerivAt_sphericalBesselJ_apply_zero: the derivative at the origin,j_1'(0) = 1 / 3andj_l'(0) = 0forl ≠ 1;deriv_deriv_sphericalBesselJ_apply_zero: the second derivative there,j_0''(0) = -1 / 3,j_2''(0) = 2 / 15andj_l''(0) = 0otherwise.hasDerivAt_deriv_sphericalBesselJwithderiv_deriv_sphericalBesselJ: the second derivative ofj_lon the whole line, the origin included, by the product rule; forx ≠ 0,hasDerivAt_deriv_sphericalBesselJ_of_ne_zeroandderiv_deriv_sphericalBesselJ_of_ne_zerogive it asj_l'' = -(2 / x) j_l' + (l (l + 1) / x ^ 2 - 1) j_l.sphericalBesselJ_equation:j_lsolves the spherical Bessel equationx ^ 2 y'' + 2 x y' + (x ^ 2 - l (l + 1)) y = 0on the whole line; at the origin it degenerates to0 = 0.
References #
- M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions, §10.1.
- NIST Digital Library of Mathematical Functions, §10.47 (the equation), §10.49(ii) (Rayleigh's formula), §10.51 (recurrence and derivatives) and §10.52 (limiting forms).
- The Hermite function ladder of the TauCeti library (
TauCeti.mul_hermiteFunction,TauCeti.hasDerivAt_hermiteFunctionandTauCeti.deriv_hermiteFunctioninTauCeti/Analysis/SpecialFunctions/Hermite/Function/Ladder.lean), the template the statements of the ladder pair follow.
The function (-(1 / x) d/dx) ^ l (sin x / x), the even analytic factor in
j_l(x) = x ^ l * sphericalBesselJDivPow l x (sphericalBesselJ_def): it equals j_l(x) / x ^ l
for x ≠ 0 (sphericalBesselJDivPow_eq_div_of_ne_zero), and at the origin it takes the limiting
value 1 / (2 l + 1)‼ (sphericalBesselJDivPow_apply_zero). The operator -(1 / x) d/dx is
f ↦ -dslope (deriv f) 0 (sphericalBesselJDivPow_def), which agrees with
-f' x / x for x ≠ 0 on the even functions it is applied to and takes the limiting value at
x = 0.
Equations
Instances For
sphericalBesselJDivPow l, which equals j_l(x) / x ^ l for x ≠ 0, is the l-th
iterate of -(1 / x) d/dx on sin x / x, with -(1 / x) d/dx taken as
f ↦ -dslope (deriv f) 0.
sphericalBesselJDivPow of order zero is sin x / x.
The Rayleigh step: sphericalBesselJDivPow (l + 1) is -(1 / x) d/dx applied to
sphericalBesselJDivPow l, with -(1 / x) d/dx taken as f ↦ -dslope (deriv f) 0; for
x ≠ 0 these equal j_{l + 1}(x) / x ^ (l + 1) and j_l(x) / x ^ l.
The spherical Bessel function of the first kind of order l, given by the Rayleigh formula
j_l(x) = x ^ l * (-(1 / x) d/dx) ^ l (sin x / x) (sphericalBesselJ_def and
sphericalBesselJDivPow_def). The resulting function is analytic on the whole line
(analyticAt_sphericalBesselJ).
Equations
Instances For
The spherical Bessel function j_l(x) is x ^ l times sphericalBesselJDivPow l x.
For x ≠ 0, the even analytic factor sphericalBesselJDivPow l x is j_l(x) / x ^ l.
Every sphericalBesselJDivPow l is real analytic on the whole line.
Every sphericalBesselJDivPow l is smooth.
Every sphericalBesselJDivPow l is differentiable.
Every sphericalBesselJDivPow l is continuous.
Every sphericalBesselJDivPow l is even.
The Rayleigh step on the whole line: the derivative of sphericalBesselJDivPow l is -x
times sphericalBesselJDivPow (l + 1).
The derivative of sphericalBesselJDivPow l is -x times sphericalBesselJDivPow (l + 1),
on the whole line.
The three-term recurrence of the functions sphericalBesselJDivPow l, which equal
j_l(x) / x ^ l for x ≠ 0, on the whole line:
x ^ 2 * sphericalBesselJDivPow (l + 2) x = (2 * l + 3) * sphericalBesselJDivPow (l + 1) x - sphericalBesselJDivPow l x. Unlike
mul_sphericalBesselJ_add_two, which multiplies it by x ^ (l + 1), it does not degenerate to
0 = 0 at the origin.
The value of sphericalBesselJDivPow l at the origin, the limiting value 1 / (2 l + 1)‼
of j_l(x) / x ^ l as x → 0.
The spherical Bessel functions of the first kind #
The order-zero spherical Bessel function is sin x / x.
Every spherical Bessel function j_l is real analytic on the whole line.
Every spherical Bessel function j_l is smooth.
Every spherical Bessel function j_l is differentiable.
Every spherical Bessel function j_l is continuous.
The value at the origin: j_0(0) = 1, and j_l(0) = 0 for every l ≠ 0.
The leading behaviour at the origin: j_l(x) / x ^ l → 1 / (2 l + 1)‼ as x → 0.
The parity of the spherical Bessel functions: j_l(-x) = (-1) ^ l j_l(x).
The three-term recurrence in the order on the whole line:
x j_{l + 2}(x) = (2 l + 3) j_{l + 1}(x) - x j_l(x).
The three-term recurrence in the order:
j_{l + 2}(x) = (2 l + 3) / x * j_{l + 1}(x) - j_l(x) for x ≠ 0.
The derivative of sphericalBesselJ l on the whole line, by the product rule for
j_l(x) = x ^ l * sphericalBesselJDivPow l x (sphericalBesselJ_def):
j_l'(x) = l * x ^ (l - 1) * sphericalBesselJDivPow l x - x ^ (l + 1) * sphericalBesselJDivPow (l + 1) x, with l - 1 truncated ℕ-subtraction as in Mathlib's
hasDerivAt_pow, so at l = 0 the first term vanishes and this reads j_0' = -j_1
(hasDerivAt_sphericalBesselJ_zero). Unlike hasDerivAt_sphericalBesselJ_of_ne_zero, which
holds for x ≠ 0 only, this holds at every point, the origin included.
The derivative of sphericalBesselJ l on the whole line, by the product rule for
j_l(x) = x ^ l * sphericalBesselJDivPow l x: the origin included.
The ladder relation at order zero, j_0' = -j_1, on the whole line, the origin included: the
order-zero case of hasDerivAt_sphericalBesselJ (DLMF 10.51.2 at l = 0).
The ladder relation at order zero, j_0' = -j_1, on the whole line, the origin included.
The derivative relation raising the order on the whole line,
x j_l'(x) = l j_l(x) - x j_{l + 1}(x). At the origin it degenerates to 0 = 0 and carries no
information about j_l'(0); that is hasDerivAt_sphericalBesselJ_apply_zero.
The derivative relation lowering the order on the whole line,
x j_{l + 1}'(x) = x j_l(x) - (l + 2) j_{l + 1}(x). At the origin it degenerates to 0 = 0 and
carries no information about j_{l + 1}'(0); that is hasDerivAt_sphericalBesselJ_apply_zero.
The derivative relation raising the order, j_l' = l / x * j_l - j_{l + 1} for x ≠ 0.
The derivative relation lowering the order,
j_{l + 1}' = j_l - (l + 2) / x * j_{l + 1} for x ≠ 0.
The derivative of j_l for x ≠ 0, j_l' = l / x * j_l - j_{l + 1}.
The derivative of j_{l + 1} for x ≠ 0, j_{l + 1}' = j_l - (l + 2) / x * j_{l + 1}.
The derivative at the origin: j_1'(0) = 1 / 3, and j_l'(0) = 0 for every l ≠ 1.
The second derivative of sphericalBesselJ l on the whole line, by the product rule applied to
deriv_sphericalBesselJ: j_l''(x) = l (l - 1) x ^ (l - 2) * sphericalBesselJDivPow l x - (2 l + 1) x ^ l * sphericalBesselJDivPow (l + 1) x + x ^ (l + 2) * sphericalBesselJDivPow (l + 2) x, with l - 1 and l - 2 truncated
ℕ-subtraction as in Mathlib's hasDerivAt_pow. Unlike
hasDerivAt_deriv_sphericalBesselJ_of_ne_zero, which holds for x ≠ 0 only, this holds at every
point, the origin included.
The second derivative of sphericalBesselJ l on the whole line, the origin included.
The second derivative of j_l for x ≠ 0, as a HasDerivAt statement:
j_l'' = -(2 / x) j_l' + (l (l + 1) / x ^ 2 - 1) j_l.
The second derivative of j_l for x ≠ 0:
j_l'' = -(2 / x) j_l' + (l (l + 1) / x ^ 2 - 1) j_l.
The second derivative at the origin, as a HasDerivAt statement: j_0''(0) = -1 / 3,
j_2''(0) = 2 / 15, and j_l''(0) = 0 for every other l.
j_l solves the spherical Bessel equation of order l on the whole line:
x ^ 2 j_l'' + 2 x j_l' + (x ^ 2 - l (l + 1)) j_l = 0. At the origin it degenerates to 0 = 0
and carries no information about j_l''(0) or j_l'(0); those are
deriv_deriv_sphericalBesselJ_apply_zero and deriv_sphericalBesselJ_apply_zero.