Google I/O is a wrap! Catch up on TensorFlow sessions

# tfp.math.bessel_iv_ratio

Computes `I_{v} (z) / I_{v - 1} (z)` in a numerically stable way.

Let I(v, z) be the modified bessel function of the first kind. This computes the ratio of I(v, z) / I(v - 1, z). This can be more numerically stable and faster than computing the ratio directly.

This uses a continued fraction approximation attributed to Gauss for computing this quantity in the limit where z <= v, and a continued fraction approximation attributed to Perron for z > v.

`v` value for which `I_{v}(z) / I_{v - 1}(z)` should be computed. Expect v > 0.
`z` value for which `I_{v}(z) / I_{v - 1}(z)` should be computed. Expect z > 0.
`name` A name for the operation (optional). Default value: `None` (i.e., 'bessel_iv_ratio').

I(v, z) / I(v - 1, z).

#### References

[1]: Walter Gautschi and Josef Slavik. On the Computation of Modified Bessel Function Ratios. http://www.jstor.com/stable/2006491

[{ "type": "thumb-down", "id": "missingTheInformationINeed", "label":"Missing the information I need" },{ "type": "thumb-down", "id": "tooComplicatedTooManySteps", "label":"Too complicated / too many steps" },{ "type": "thumb-down", "id": "outOfDate", "label":"Out of date" },{ "type": "thumb-down", "id": "samplesCodeIssue", "label":"Samples / code issue" },{ "type": "thumb-down", "id": "otherDown", "label":"Other" }]
[{ "type": "thumb-up", "id": "easyToUnderstand", "label":"Easy to understand" },{ "type": "thumb-up", "id": "solvedMyProblem", "label":"Solved my problem" },{ "type": "thumb-up", "id": "otherUp", "label":"Other" }]