# Even after reading the FAQ, still having trouble with a "Bad initial energy" error

**URL:** <https://discourse.pymc.io/t/even-after-reading-the-faq-still-having-trouble-with-a-bad-initial-energy-error/5479>\
**Category:** Questions\
**Created:** [July 20, 2020, 1:17pm UTC](https://discourse.pymc.io/t/even-after-reading-the-faq-still-having-trouble-with-a-bad-initial-energy-error/5479 "2020-07-20T13:17:11Z")\
**Posts on this page:** 1\
**Showing post:** 4

<div class="post-metadata">

**Author:** ![Jesse\_DuMond](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/jesse_dumond/32/2415_2.png) [@Jesse\_DuMond](https://discourse.pymc.io/u/Jesse_DuMond)\
**Post date:** [July 22, 2020, 2:06am UTC](https://discourse.pymc.io/t/even-after-reading-the-faq-still-having-trouble-with-a-bad-initial-energy-error/5479/4 "2020-07-22T02:06:36Z")

</div>

Yep. [This](https://discourse.pymc.io/t/negative-binomial-returning-inf-logp/5409/8) did the trick. Setting `theano.config.floatX` to “float64” allowed the sampler to run with out the “Bad initial energy” error and without having to scale. Thanks for pointing me to this!

---

_[View the full topic](https://discourse.pymc.io/t/even-after-reading-the-faq-still-having-trouble-with-a-bad-initial-energy-error/5479)._