Flavio

This is a bug fix release due to a problem with the interpolation of NNLO corrections in b->sll at very low q² (see issue 61). Note that this release does not yet contain the modifications in the master branch since v0.29.1.

Due to a mistake, the release notes for v0.29.1, the previous bugfix release, had not been published on Github. Instead they are given below.

Bugs fixed

- Overall sign error in the new physics contribution to ε'/ε
- Sign error in one of the K→ππ matrix elements
- Normalization of K→ππ lattice matrix elements changed (numerical impact small)

This was used in v2 of [arXiv:1807.02520](https://arxiv.org/abs/1807.02520).

Other changes

The default values for the kaon bag parameters of BSM operators have been changed from ETM 2015 to [RBC/UKQCD 2016](https://arxiv.org/abs/1609.03334).

New physics in ε'/ε

The highlight of this release is the addition of the observable ε'/ε, measuring the ratio of direct over indirect CP violation in K→ππ decays. This implementation is based on the results presented in [arXiv:1807.02520](https://arxiv.org/abs/1807.02520) by J. Aebischer, C. Bobeth, A. Buras, J.-M. Gérard, and D. Straub. This is the first open source implementation of new physics contributions to this important observable taking into account all 114 ΔS=1 operators below the electroweak scale (or arbitrary SMEFT operators through [wilson](https://wilson-eft.github.io/)).

Bugs fixed

- Sign error in new physics contributions to τ→Vl decays fixed
- Font issue when plotting on certain systems fixed (thanks to peterstangl)
- Regression introduced in v0.28 affecting new physics contributions to meson mixing fixed

Other improvements

- Various speed optimizations

This release brings a large number of new observables as well as changes under the hood.

Treatment of new physics

The treatment of new physics Wilson coefficients was completely overhauled and is now entirely based on the brand new [wilson package](https://wilson-eft.github.io). Since wilson is capable of RG running and matching the *complete set* of dimension-6 operators not only in the weak effective theory but also in SMEFT, this opens up many new possibilities for flavio. As a start, it is now possible to use instances of wilson's `Wilson` class as an argument to functions like `np_prediction`. The fit classes in `flavio.statistics.fits` support the new optional arguments `fit_wc_eft` and `fit_wc_basis` that allow to directly make fits in a different EFT (e.g. SMEFT) or basis.

New observables

- Various observables in the neutral *D* meson system have been added. Note that the SM predictions are not really predictions, as the long distance contributions are poorly known, but merely parametrizations accounting for the ballpark of the experimental measurements. For testing CP violating new physics in D mixing, the observable `x12Im_D` can be used, that is negligibly small in the SM.
- The neutron EDM `d_n`
- Lepton flavour violating muon and tau decays
- `BR(mu->egamma)`
- `BR(mu->eee)`
- `BR(tau->egamma)`
- `BR(tau->mugamma)`
- `BR(tau->eee)`
- `BR(tau->mumumu)`
- `BR(tau->muee)`
- `BR(tau->rhoe)`
- `BR(tau->rhomu)`
- Hadronic tau decays
- `BR(tau->pinu)`
- `BR(tau->Knu)`
- Leptonic tau decays
- `BR(tau->mununu)`
- `BR(tau->enunu)`

This minor release fixes a bug and brings a few minor improvements.

- A bug in the new physics contributions to B→lν decays was fixed.
- The methods `get_central_all` and `get_random_all` of `ParameterConstraints` have been accelerated.
- The `mpmath` package was dropped as a dependency since the complex dilogarithm, for which it was needed in the past, has been implemented in scipy and is 20 times faster. The minimum scipy version required is now 0.18.
- A new probability distribution `LogNormalDistribution` has been added. This is particularly useful for poorly known multiplicative factors in theory calculations. In YAML files, it can be specified with the notation `1 */ 3`, corresponding to the exponential of a random variable normally distributed around ln(1) with mean and standard deviation ln(3) (the notation reminds of the fact that 68% probability is contained within 1*3 and 1/3).

This release fixes a bug affecting B→D(*)τν in the presence of new physics in tensor operators. Thanks to Olcyr Sumensari for pointing it out.

In addition, the most recent LHCb measurement of R_D* was added.