The Bessel function calculator above evaluates Jν(x) and Yν(x) for any real order and any argument from zero up to twenty, and reports the neighbouring order and both derivatives alongside. It uses the ascending power series rather than a lookup table, so fractional orders work as readily as whole ones.
Arb Digital publishes it because Bessel functions turn up the moment a wave, a heat flow or a vibration meets a circular boundary, and the values are awkward to get hold of outside a full numerical library. The page also states where its own accuracy runs out, which a calculator quoting ten digits regardless of argument is quietly failing to do.
What This Bessel Function Calculator Does
It computes the two standard solutions of Bessel's differential equation. The first kind, Jν, is the solution that stays finite at the origin. The second kind, Yν, is the independent solution that diverges there. Together they span every solution of the equation, which is why both are needed even though only one of them appears in many physical problems.
Alongside the two functions it gives Jν⁺₁(x) and both derivatives, computed from the standard recurrence relations rather than by numerical differencing. That matters for accuracy: the derivative of Jν equals half the difference of the neighbouring orders exactly, so no step size has to be chosen and no truncation error enters.
The order may be fractional or negative. Half-integer orders are the useful special case: J at order one half reduces to a sine, and at order minus one half to a cosine, both scaled by the square root of two over pi x. Those are worth trying first, because you can check them against our trigonometric functions calculator without trusting this page at all.
How to Use It
- Enter the order ν. Whole numbers are the common case; fractional and negative orders are supported.
- Enter the argument x, from zero up to twenty. Accuracy warnings appear automatically at larger values.
- Choose which function fills the headline panel. Both are computed either way.
- Read the grid for the second kind, the next order up, and the two derivatives.
- Check the working panel for the series terms used and any note about precision at that argument.
The Formulas and How They Are Calculated
The first kind is computed directly from its ascending power series. The NIST Digital Library of Mathematical Functions gives it in §10.2 Definitions, in the Bessel and Hankel Functions chapter: Jν(z) is the sum over k of (−1)k (z/2)ν+2k divided by k! times the gamma function of ν+k+1. This calculator evaluates the reciprocal gamma factor through a Lanczos approximation to the log-gamma function, with the reflection formula handling negative arguments, and sums terms until they stop contributing.
The second kind is obtained from the standard combination of first-kind functions of order ν and −ν, divided by the sine of νπ. That expression is indeterminate at whole-number orders, where the true value is the limit as ν approaches the integer. The tool evaluates that limit numerically by averaging the expression a small distance either side of the integer, which cancels the leading error and recovers roughly ten correct digits at small arguments.
The derivatives use the recurrence relations from DLMF §10.6 Recurrence Relations and Derivatives. For any cylinder function, the derivative equals half the difference between the order below and the order above, and the same relation holds for J and Y alike.
The values have been checked against published tables. At order zero and argument one the tool returns J = 0.76519769 and Y = 0.08825696, matching the standard table values of 0.7651976866 and 0.0882569642. At order one and argument one it gives 0.44005059 and −0.78121282, against published 0.4400505857 and −0.7812128213. Order one half at argument one returns 0.67139671, which equals the square root of two over pi times the sine of one to every digit shown.
Where the Accuracy Runs Out, and Why
The ascending series converges for every argument, but converging is not the same as being computable. As x grows the individual terms first become very large before the alternating signs cancel them down to a small answer. At x = 20 the largest term is several million times the size of the final value, so several million-fold cancellation happens inside double-precision arithmetic, and about seven significant digits are destroyed in the process.
That is why the argument is capped at twenty here and why the tool warns above ten. Professional libraries switch method rather than pushing the series further: for large arguments they use the asymptotic expansion, in which J behaves like a decaying cosine and Y like a decaying sine, both falling off as one over the square root of x. If you need values well beyond twenty, use a numerical library or published tables rather than any browser calculator.
The second kind carries slightly more error than the first at whole-number orders, because it is reached through a limit rather than computed directly. At small arguments both are good to ten digits; by x = 10 the second kind is reliable to about seven. The working panel states which regime you are in rather than leaving you to guess from the digit count.
What Bessel Functions Are Actually For
They are what the trigonometric functions become when a problem is circular rather than rectangular. Separating the wave equation in Cartesian coordinates produces sines and cosines; separating the same equation in cylindrical coordinates produces Bessel functions, and the reason is the extra first-derivative term that the radial coordinate introduces.
The concrete consequences are everywhere. A circular drumhead vibrates in modes whose radial shape is J₀, J₁ and so on, and the allowed frequencies are set by the zeros of those functions — which is why a drum, unlike a string, has overtones that are not whole-number multiples of the fundamental. Heat flowing radially in a cylinder, the field pattern inside a circular waveguide, the diffraction pattern of a circular aperture, and the modes of an optical fibre all reduce to the same functions.
The zeros carry most of the physical content. The first zero of J₀ is at 2.404825558, the second at 5.520078110, and the gaps between successive zeros settle towards pi as the argument grows. Load the first zero with the button above and the tool returns a value indistinguishable from zero, which is a useful check that the series is behaving. To move a point into the coordinates these problems are stated in, our 3D coordinate converter handles cylindrical and spherical forms.
The Two Kinds, and When Each Belongs
Both J and Y solve the same differential equation, so the general solution is a combination of the two. Which one survives is decided by the physical boundary, not by the mathematics.
If the region includes the axis — a solid drumhead, a filled cylinder, a circular waveguide with no inner conductor — then Y is excluded, because it diverges at the centre and no physical quantity does. The solution is a pure J, and the boundary condition at the rim then picks out the allowed values, which is where the zeros enter. If the region is an annulus, with an inner boundary as well as an outer one, both kinds are needed and the coefficients are fixed by conditions at both edges.
Two further families are worth knowing about because their names cause confusion. The modified Bessel functions I and K solve a closely related equation with a sign changed, and they grow or decay rather than oscillating; they arise in steady heat conduction and in skin-effect problems. The Hankel functions are combinations of J and Y that represent travelling rather than standing waves. This calculator covers J and Y only. For the general oscillation and decay behaviour, our wave equation calculator, wavelength calculator and hyperbolic functions calculator are the neighbouring tools.
Arb Digital builds technical tools that report their limits instead of printing digits they cannot justify.
Browse All Free Tools Talk To Our TeamCommon Mistakes to Avoid
- Using Y where the region includes the axis — it diverges at the origin, so it cannot appear in any solution covering the centre.
- Trusting every printed digit at large arguments — the ascending series loses accuracy to cancellation, and roughly seven digits are gone by x = 20.
- Confusing J and Y with the modified functions I and K — those solve a different equation and grow or decay instead of oscillating.
- Expecting drum overtones to be whole-number multiples — the frequencies come from the zeros of Bessel functions, which are not evenly spaced.
- Differentiating numerically — the recurrence relations give the derivative exactly, with no step size to choose and no truncation error.
Related Free Tools From Arb Digital
Check half-integer orders against the trigonometric functions calculator, handle growth and decay with the hyperbolic functions calculator, work with waves using the wave equation calculator and wavelength calculator, or move between coordinate systems with the 3D coordinate converter. The full free online tools hub lists every mathematics tool we publish.
Frequently Asked Questions
Both solve Bessel's equation, but the first kind J is finite at the origin while the second kind Y diverges there. Any problem whose region includes the centre uses J alone.
Yes. The series works for any real order. Half-integer orders are especially useful because they reduce to sines and cosines and can be checked by hand.
About ten significant digits at small arguments. Accuracy falls as the argument grows because the alternating series cancels heavily, so roughly seven digits are lost by an argument of twenty.
Because beyond that the ascending series cannot be evaluated reliably in double precision. Numerical libraries switch to an asymptotic expansion for large arguments instead of pushing the series further.
They appear when the wave equation, the heat equation or Laplace's equation is separated in cylindrical coordinates. They are the circular counterpart of the sines and cosines that arise in rectangular coordinates.
They set the allowed frequencies of a circular vibrating membrane and the cutoff frequencies of a circular waveguide. The first zero of the order-zero function is at approximately 2.404825558.
No. They solve a related equation with a sign change and grow or decay rather than oscillating. This page covers the first and second kind only.
From the standard recurrence relations, where the derivative equals half the difference between the order below and the order above. That is exact, so no numerical differencing is involved.
This page evaluates published mathematical functions for educational purposes. For high-precision work or arguments beyond its stated range, use a numerical library or published tables.
