@MattT I'm not sure how changing the JeansRefinementColdTemperature behavior would affect your simulations. In runs with heated regions, there was too much refinement. But I think of this parameter to avoid refinement in the IGM at temperatures T < T_CMB and when the mean density is high at early times.
Hi John, I always used it as a theoretical minimum to which the gas could cool -- to increase refinement. I think the idea was "What's the minimum temperature you would want things to cool to when you evaluate the Jeans". Maybe it was poorly named.
Oh, I see now. This would change the behavior, and I think we should keep it the same. I see two solutions: (1) If the parameter is negative, then we use it as a floor. If it's positive, then we use it as a constant temperature. (2) Just make a new parameter, JeansRefinementTemperatureFloor, to avoid refinement in very cold regions.
I'm leaning toward option 2 to avoid confusion in the parameter, but it's yet another parameter.