• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1[section:bessel_first Bessel Functions of the First and Second Kinds]
2
3[h4 Synopsis]
4
5`#include <boost/math/special_functions/bessel.hpp>`
6
7   template <class T1, class T2>
8   ``__sf_result`` cyl_bessel_j(T1 v, T2 x);
9
10   template <class T1, class T2, class ``__Policy``>
11   ``__sf_result`` cyl_bessel_j(T1 v, T2 x, const ``__Policy``&);
12
13   template <class T1, class T2>
14   ``__sf_result`` cyl_neumann(T1 v, T2 x);
15
16   template <class T1, class T2, class ``__Policy``>
17   ``__sf_result`` cyl_neumann(T1 v, T2 x, const ``__Policy``&);
18
19
20[h4 Description]
21
22The functions __cyl_bessel_j and __cyl_neumann return the result of the
23Bessel functions of the first and second kinds respectively:
24
25[expression cyl_bessel_j(v, x) = J[sub v](x)]
26
27[expression cyl_neumann(v, x) = Y[sub v](x) = N[sub v](x)]
28
29where:
30
31[equation bessel2]
32
33[equation bessel3]
34
35The return type of these functions is computed using the __arg_promotion_rules
36when T1 and T2 are different types.  The functions are also optimised for the
37relatively common case that T1 is an integer.
38
39[optional_policy]
40
41The functions return the result of __domain_error whenever the result is
42undefined or complex.  For __cyl_bessel_j this occurs when `x < 0` and v is not
43an integer, or when `x == 0` and `v != 0`.  For __cyl_neumann this occurs
44when `x <= 0`.
45
46The following graph illustrates the cyclic nature of J[sub v]:
47
48[graph cyl_bessel_j]
49
50The following graph shows the behaviour of Y[sub v]: this is also
51cyclic for large /x/, but tends to -[infin] for small /x/:
52
53[graph cyl_neumann]
54
55[h4 Testing]
56
57There are two sets of test values: spot values calculated using
58[@http://functions.wolfram.com functions.wolfram.com],
59and a much larger set of tests computed using
60a simplified version of this implementation
61(with all the special case handling removed).
62
63[h4 Accuracy]
64
65The following tables show how the accuracy of these functions
66varies on various platforms, along with comparisons to other
67libraries.  Note that the cyclic nature of these
68functions means that they have an infinite number of irrational
69roots: in general these functions have arbitrarily large /relative/
70errors when the arguments are sufficiently close to a root.  Of
71course the absolute error in such cases is always small.
72Note that only results for the widest floating-point type on the
73system are given as narrower types have __zero_error.  All values
74are relative errors in units of epsilon.  Most of the gross errors
75exhibited by other libraries occur for very large arguments - you will
76need to drill down into the actual program output if you need more
77information on this.
78
79[table_cyl_bessel_j_integer_orders_]
80
81[table_cyl_bessel_j]
82
83[table_cyl_neumann_integer_orders_]
84
85[table_cyl_neumann]
86
87Note that for large /x/ these functions are largely dependent on
88the accuracy of the `std::sin` and `std::cos` functions.
89
90Comparison to GSL and __cephes is interesting: both __cephes and this library optimise
91the integer order case - leading to identical results - simply using the general
92case is for the most part slightly more accurate though, as noted by the
93better accuracy of GSL in the integer argument cases.  This implementation tends to
94perform much better when the arguments become large, __cephes in particular produces
95some remarkably inaccurate results with some of the test data (no significant figures
96correct), and even GSL performs badly with some inputs to J[sub v].  Note that
97by way of double-checking these results, the worst performing __cephes and GSL cases
98were recomputed using [@http://functions.wolfram.com functions.wolfram.com],
99and the result checked against our test data: no errors in the test data were found.
100
101The following error plot are based on an exhaustive search of the functions domain for J0 and Y0,
102MSVC-15.5 at `double` precision, other compilers and precisions are very similar - the plots simply
103illustrate the relatively large errors as you approach a zero, and the very low errors elsewhere.
104
105[graph j0__double]
106
107[graph y0__double]
108
109
110[h4 Implementation]
111
112The implementation is mostly about filtering off various special cases:
113
114When /x/ is negative, then the order /v/ must be an integer or the
115result is a domain error.  If the order is an integer then the function
116is odd for odd orders and even for even orders, so we reflect to /x > 0/.
117
118When the order /v/ is negative then the reflection formulae can be used to
119move to /v > 0/:
120
121[equation bessel9]
122
123[equation bessel10]
124
125Note that if the order is an integer, then these formulae reduce to:
126
127[expression J[sub -n] = (-1)[super n]J[sub n]]
128
129[expression Y[sub -n] = (-1)[super n]Y[sub n]]
130
131However, in general, a negative order implies that we will need to compute
132both J and Y.
133
134When /x/ is large compared to the order /v/ then the asymptotic expansions
135for large /x/ in M. Abramowitz and I.A. Stegun,
136['Handbook of Mathematical Functions] 9.2.19 are used
137(these were found to be more reliable
138than those in A&S 9.2.5).
139
140When the order /v/ is an integer the method first relates the result
141to J[sub 0], J[sub 1], Y[sub 0] and Y[sub 1] using either
142forwards or backwards recurrence (Miller's algorithm) depending upon which is stable.
143The values for J[sub 0], J[sub 1], Y[sub 0] and Y[sub 1] are
144calculated using the rational minimax approximations on
145root-bracketing intervals for small ['|x|] and Hankel asymptotic
146expansion for large ['|x|]. The coefficients are from:
147
148[:W.J. Cody, ['ALGORITHM 715: SPECFUN - A Portable FORTRAN Package of
149Special Function Routines and Test Drivers], ACM Transactions on Mathematical
150Software, vol 19, 22 (1993).]
151
152and
153
154[:J.F. Hart et al, ['Computer Approximations], John Wiley & Sons, New York, 1968.]
155
156These approximations are accurate to around 19 decimal digits: therefore
157these methods are not used when type T has more than 64 binary digits.
158
159When /x/ is smaller than machine epsilon then the following approximations for
160Y[sub 0](x), Y[sub 1](x), Y[sub 2](x) and Y[sub n](x) can be used
161(see: [@http://functions.wolfram.com/03.03.06.0037.01 http://functions.wolfram.com/03.03.06.0037.01],
162[@http://functions.wolfram.com/03.03.06.0038.01 http://functions.wolfram.com/03.03.06.0038.01],
163[@http://functions.wolfram.com/03.03.06.0039.01 http://functions.wolfram.com/03.03.06.0039.01]
164and [@http://functions.wolfram.com/03.03.06.0040.01 http://functions.wolfram.com/03.03.06.0040.01]):
165
166[equation bessel_y0_small_z]
167
168[equation bessel_y1_small_z]
169
170[equation bessel_y2_small_z]
171
172[equation bessel_yn_small_z]
173
174When /x/ is small compared to /v/ and /v/ is not an integer, then the following
175series approximation can be used for Y[sub v](x), this is also an area where other
176approximations are often too slow to converge to be used
177(see [@http://functions.wolfram.com/03.03.06.0034.01 http://functions.wolfram.com/03.03.06.0034.01]):
178
179[equation bessel_yv_small_z]
180
181When /x/ is small compared to /v/, J[sub v]x is best computed directly from the series:
182
183[equation bessel2]
184
185In the general case we compute J[sub v] and
186Y[sub v] simultaneously.
187
188To get the initial values, let
189[mu] = [nu] - floor([nu] + 1/2), then [mu] is the fractional part
190of [nu] such that
191|[mu]| <= 1/2 (we need this for convergence later). The idea is to
192calculate J[sub [mu]](x), J[sub [mu]+1](x), Y[sub [mu]](x), Y[sub [mu]+1](x)
193and use them to obtain J[sub [nu]](x), Y[sub [nu]](x).
194
195The algorithm is called Steed's method, which needs two
196continued fractions as well as the Wronskian:
197
198[equation bessel8]
199
200[equation bessel11]
201
202[equation bessel12]
203
204See: F.S. Acton, ['Numerical Methods that Work],
205    The Mathematical Association of America, Washington, 1997.
206
207The continued fractions are computed using the modified Lentz's method
208(W.J. Lentz, ['Generating Bessel functions in Mie scattering calculations
209using continued fractions], Applied Optics, vol 15, 668 (1976)).
210Their convergence rates depend on ['x], therefore we need
211different strategies for large ['x] and small ['x]:
212
213[:['x > v], CF1 needs O(['x]) iterations to converge, CF2 converges rapidly]
214
215[:['x <= v], CF1 converges rapidly, CF2 fails to converge when ['x] [^->] 0]
216
217When ['x] is large (['x] > 2), both continued fractions converge (CF1
218may be slow for really large ['x]). J[sub [mu]], J[sub [mu]+1],
219Y[sub [mu]], Y[sub [mu]+1] can be calculated by
220
221[equation bessel13]
222
223where
224
225[equation bessel14]
226
227J[sub [nu]] and Y[sub [mu]] are then calculated using backward
228(Miller's algorithm) and forward recurrence respectively.
229
230When ['x] is small (['x] <= 2), CF2 convergence may fail (but CF1
231works very well). The solution here is Temme's series:
232
233[equation bessel15]
234
235where
236
237[equation bessel16]
238
239g[sub k] and h[sub k]
240are also computed by recursions (involving gamma functions), but the
241formulas are a little complicated, readers are referred to
242N.M. Temme, ['On the numerical evaluation of the ordinary Bessel function
243of the second kind], Journal of Computational Physics, vol 21, 343 (1976).
244Note Temme's series converge only for |[mu]| <= 1/2.
245
246As the previous case, Y[sub [nu]] is calculated from the forward
247recurrence, so is Y[sub [nu]+1]. With these two
248values and f[sub [nu]], the Wronskian yields J[sub [nu]](x) directly
249without backward recurrence.
250
251[endsect] [/section:bessel_first Bessel Functions of the First and Second Kinds]
252
253[section:bessel_root Finding Zeros of Bessel Functions of the First and Second Kinds]
254
255[h4 Synopsis]
256
257`#include <boost/math/special_functions/bessel.hpp>`
258
259Functions for obtaining both a single zero or root of the Bessel function,
260and placing multiple zeros into a container like `std::vector`
261by providing an output iterator.
262
263The signature of the single value functions are:
264
265  template <class T>
266  T cyl_bessel_j_zero(
267           T v,            // Floating-point value for Jv.
268           int m);         // 1-based index of zero.
269
270  template <class T>
271  T cyl_neumann_zero(
272           T v,            // Floating-point value for Jv.
273           int m);         // 1-based index of zero.
274
275and for multiple zeros:
276
277 template <class T, class OutputIterator>
278 OutputIterator cyl_bessel_j_zero(
279                      T v,                       // Floating-point value for Jv.
280                      int start_index,           // 1-based index of first zero.
281                      unsigned number_of_zeros,  // How many zeros to generate.
282                      OutputIterator out_it);    // Destination for zeros.
283
284 template <class T, class OutputIterator>
285 OutputIterator cyl_neumann_zero(
286                      T v,                       // Floating-point value for Jv.
287                      int start_index,           // 1-based index of zero.
288                      unsigned number_of_zeros,  // How many zeros to generate
289                      OutputIterator out_it);    // Destination for zeros.
290
291There are also versions which allow control of the __policy_section for error handling and precision.
292
293  template <class T>
294  T cyl_bessel_j_zero(
295           T v,            // Floating-point value for Jv.
296           int m,          // 1-based index of zero.
297           const Policy&); // Policy to use.
298
299  template <class T>
300  T cyl_neumann_zero(
301           T v,            // Floating-point value for Jv.
302           int m,          // 1-based index of zero.
303           const Policy&); // Policy to use.
304
305
306 template <class T, class OutputIterator>
307 OutputIterator cyl_bessel_j_zero(
308                      T v,                       // Floating-point value for Jv.
309                      int start_index,           // 1-based index of first zero.
310                      unsigned number_of_zeros,  // How many zeros to generate.
311                      OutputIterator out_it,     // Destination for zeros.
312                      const Policy& pol);        // Policy to use.
313
314 template <class T, class OutputIterator>
315 OutputIterator cyl_neumann_zero(
316                      T v,                       // Floating-point value for Jv.
317                      int start_index,           // 1-based index of zero.
318                      unsigned number_of_zeros,  // How many zeros to generate.
319                      OutputIterator out_it,     // Destination for zeros.
320                      const Policy& pol);        // Policy to use.
321
322[h4 Description]
323
324Every real order [nu] cylindrical Bessel and Neumann functions have an infinite
325number of zeros on the positive real axis. The real zeros on the positive real
326axis can be found by solving for the roots of
327
328[:['J[sub [nu]](j[sub [nu], m]) = 0]]
329
330[:['Y[sub [nu]](y[sub [nu], m]) = 0]]
331
332Here, ['j[sub [nu], m]] represents the ['m[super th]]
333root of the cylindrical Bessel function of order ['[nu]],
334and ['y[sub [nu], m]] represents the ['m[super th]]
335root of the cylindrical Neumann function of order ['[nu]].
336
337The zeros or roots (values of `x` where the function crosses the horizontal `y = 0` axis)
338of the Bessel and Neumann functions are computed by two functions,
339`cyl_bessel_j_zero` and `cyl_neumann_zero`.
340
341In each case the index or rank of the zero
342returned is 1-based, which is to say:
343
344[:cyl_bessel_j_zero(v, 1);]
345
346returns the first zero of Bessel J.
347
348Passing an `start_index <= 0` results in a `std::domain_error` being raised.
349
350For certain parameters, however, the zero'th root is defined and
351it has a value of zero. For example, the zero'th root
352of `J[sub v](x)` is defined and it has a value of zero for all
353values of `v > 0` and for negative integer values of `v = -n`.
354Similar cases are described in the implementation details below.
355
356The order `v` of `J` can be positive, negative and zero for the `cyl_bessel_j`
357and `cyl_neumann` functions, but not infinite nor NaN.
358
359[graph bessel_j_zeros]
360
361[graph neumann_y_zeros]
362
363[h4 Examples of finding Bessel and Neumann zeros]
364
365[import ../../example/bessel_zeros_example_1.cpp]
366
367[bessel_zeros_example_1]
368[bessel_zeros_example_2]
369
370[import ../../example/bessel_zeros_interator_example.cpp]
371
372[bessel_zeros_iterator_example_1]
373[bessel_zeros_iterator_example_2]
374
375[import ../../example/neumann_zeros_example_1.cpp]
376
377[neumann_zeros_example_1]
378[neumann_zeros_example_2]
379
380[import ../../example/bessel_errors_example.cpp]
381
382[bessel_errors_example_1]
383[bessel_errors_example_2]
384
385The full code (and output) for these examples is at
386[@../../example/bessel_zeros_example_1.cpp Bessel zeros],
387[@../../example/bessel_zeros_interator_example.cpp Bessel zeros iterator],
388[@../../example/neumann_zeros_example_1.cpp Neumann zeros],
389[@../../example/bessel_errors_example.cpp  Bessel error messages].
390
391[h3 Implementation]
392
393Various methods are used to compute initial estimates
394for ['j[sub [nu], m]] and ['y[sub [nu], m]] ; these are described in detail below.
395
396After finding the initial estimate of a given root,
397its precision is subsequently refined to the desired level
398using Newton-Raphson iteration from Boost.Math's __root_finding_with_derivatives
399utilities combined with the functions __cyl_bessel_j  and __cyl_neumann.
400
401Newton iteration requires both ['J[sub [nu]](x)] or ['Y[sub [nu]](x)]
402as well as its derivative. The derivatives of ['J[sub [nu]](x)] and ['Y[sub [nu]](x)]
403with respect to  ['x] are given by __Abramowitz_Stegun.
404In particular,
405
406[expression d/[sub dx] ['J[sub [nu]](x)] = ['J[sub [nu]-1](x)] - [nu] J[sub [nu]](x) / x]
407
408[expression d/[sub dx] ['Y[sub [nu]](x)] = ['Y[sub [nu]-1](x)] - [nu] Y[sub [nu]](x) / x]
409
410Enumeration of the rank of a root (in other words the index of a root)
411begins with one and counts up, in other words
412['m,=1,2,3,[ellipsis]] The value of the first root is always greater than zero.
413
414For certain special parameters, cylindrical Bessel functions
415and cylindrical Neumann functions have a root at the origin. For example,
416['J[sub [nu]](x)] has a root at the origin for every positive order
417['[nu] > 0], and for every negative integer order
418['[nu] = -n] with ['n [isin] [negative] [super +]] and ['n [ne] 0].
419
420In addition, ['Y[sub [nu]](x)] has a root at the origin
421for every negative half-integer order ['[nu] = -n/2], with ['n [isin] [negative] [super +]] and
422and ['n [ne] 0].
423
424For these special parameter values, the origin with
425a value of ['x = 0] is provided as the ['0[super th]]
426root generated by `cyl_bessel_j_zero()`
427and `cyl_neumann_zero()`.
428
429When calculating initial estimates for the roots
430of Bessel functions, a distinction is made between
431positive order and negative order, and different
432methods are used for these. In addition, different algorithms
433are used for the first root ['m = 1] and
434for subsequent roots with higher rank ['m [ge] 2].
435Furthermore, estimates of the roots for Bessel functions
436with order above and below a cutoff at ['[nu] = 2.2]
437are calculated with different methods.
438
439Calculations of the estimates of ['j[sub [nu],1]] and ['y[sub [nu],1]]
440with ['0 [le] [nu] < 2.2] use empirically tabulated values.
441The coefficients for these have been generated by a
442computer algebra system.
443
444Calculations of the estimates of ['j[sub [nu],1]] and ['y[sub [nu],1]]
445with ['[nu][ge] 2.2] use Eqs.9.5.14 and 9.5.15 in __Abramowitz_Stegun.
446
447In particular,
448[expression j[sub [nu],1] [cong] [nu] + 1.85575 [nu][super [frac13]] + 1.033150 [nu][super -[frac13]] - 0.00397 [nu][super -1] - 0.0908 [nu][super -5/3] + 0.043 [nu][super -7/3] + [ellipsis]]
449
450and
451
452[expression y[sub [nu],1] [cong] [nu] + 0.93157 [nu][super [frac13]] + 0.26035 [nu][super -[frac13]] + 0.01198 [nu][super -1] - 0.0060 [nu][super -5/3] - 0.001 [nu][super -7/3] + [ellipsis]]
453
454Calculations of the estimates of ['j[sub [nu], m]]  and  ['y[sub [nu], m]]
455with rank ['m > 2] and ['0 [le] [nu] < 2.2]  use
456McMahon's approximation, as described in M. Abramowitz and I. A. Stegan, Section 9.5 and 9.5.12.
457In particular,
458
459[:['j[sub [nu],m], y[sub [nu],m] [cong]]]
460[:[:[beta] - ([mu]-1) / 8[beta]]]
461[:[:['- 4([mu]-1)(7[mu] - 31) / 3(8[beta])[super 3]]]]
462[:[:['-32([mu]-1)(83[mu][sup2] - 982[mu] + 3779) / 15(8[beta])[super 5]]]]
463[:[:['-64([mu]-1)(6949[mu][super 3] - 153855[mu][sup2] + 1585743[mu]- 6277237) / 105(8a)[super 7]]]]
464[:[:['- [ellipsis]] [emquad] (5)]]
465
466where ['[mu] = 4[nu][super 2]] and ['[beta] = (m + [frac12][nu] - [frac14])[pi]]
467for ['j[sub [nu],m]] and
468['[beta] = (m + [frac12][nu] -[frac34])[pi] for ['y[sub [nu],m]]].
469
470Calculations of the estimates of ['j[sub [nu], m]]  and  ['y[sub [nu], m]]
471with ['[nu] [ge] 2.2] use
472one term in the asymptotic expansion given in
473Eq.9.5.22 and top line of Eq.9.5.26 combined with Eq. 9.3.39,
474all in __Abramowitz_Stegun explicit and easy-to-understand treatment
475for asymptotic expansion of zeros.
476The latter two equations are expressed for argument ['(x)] greater than one.
477(Olver also gives the series form of the equations in
478[@http://dlmf.nist.gov/10.21#vi [sect]10.21(vi) McMahon's Asymptotic Expansions for Large Zeros] - using slightly different variable names).
479
480In summary,
481
482[expression j[sub [nu], m] [sim] [nu]x(-[zeta]) + f[sub 1](-[zeta]/[nu])]
483
484where ['-[zeta] = [nu][super -2/3]a[sub m]] and ['a[sub m]] is
485the absolute value of the ['m[super th]] root of ['Ai(x)] on the negative real axis.
486
487Here ['x = x(-[zeta])] is the inverse of the function
488
489[expression [frac23](-[zeta])[super 3/2] = [radic](x[sup2] - 1) - cos[supminus][sup1](1/x)]     (7)
490
491Furthermore,
492
493[expression f[sub 1](-[zeta]) = [frac12]x(-[zeta]) {h(-[zeta])}[sup2] [sdot] b[sub 0](-[zeta])]
494
495where
496
497[expression h(-[zeta]) = {4(-[zeta]) / (x[sup2] - 1)}[super 4]]
498
499and
500
501[expression b[sub 0](-[zeta]) = -5/(48[zeta][sup2]) + 1/(-[zeta])[super [frac12]] [sdot] { 5/(24(x[super 2]-1)[super 3/2]) + 1/(8(x[super 2]-1)[super [frac12])]}]
502
503When solving for ['x(-[zeta])] in Eq. 7 above,
504the right-hand-side is expanded to order 2 in
505a Taylor series for large ['x]. This results in
506
507[expression [frac23](-[zeta])[super 3/2] [approx] x + 1/2x - [pi]/2]
508
509The positive root of the resulting quadratic equation
510is used to find an initial estimate ['x(-[zeta])].
511This initial estimate is subsequently refined with
512several steps of Newton-Raphson iteration
513in Eq. 7.
514
515Estimates of the roots of cylindrical Bessel functions
516of negative order on the positive real axis are found
517using interlacing relations. For example, the ['m[super th]]
518root of the cylindrical Bessel function ['j[sub -[nu],m]]
519is bracketed by the ['m[super th]] root and the
520['(m+1)[super th]] root of the Bessel function of
521corresponding positive integer order. In other words,
522
523[expression j[sub n[nu], m] < j[sub -[nu], m] < j[sub n[nu], m+1]]
524
525where ['m > 1] and ['n[sub [nu]]] represents the integral
526floor of the absolute value of ['|-[nu]|].
527
528Similar bracketing relations are used to find estimates
529of the roots of Neumann functions of negative order,
530whereby a discontinuity at every negative half-integer
531order needs to be handled.
532
533Bracketing relations do not hold for the first root
534of cylindrical Bessel functions and cylindrical Neumann
535functions with negative order. Therefore, iterative algorithms
536combined with root-finding via bisection are used
537to localize ['j[sub -[nu],1]] and ['y[sub -[nu],1]].
538
539[h3 Testing]
540
541The precision of evaluation of zeros was tested at 50 decimal digits using `cpp_dec_float_50`
542and found identical with spot values computed by __WolframAlpha.
543
544[endsect]  [/section:bessel Finding Zeros of Bessel Functions of the First and Second Kinds]
545
546[/
547  Copyright 2006, 2013 John Maddock, Paul A. Bristow, Xiaogang Zhang and Christopher Kormanyos.
548
549  Distributed under the Boost Software License, Version 1.0.
550  (See accompanying file LICENSE_1_0.txt or copy at
551  http://www.boost.org/LICENSE_1_0.txt).
552]
553