rotation and RTI do not respect no_mix from mlt.

[Debraheem]

Tagging people for a second opinion.

I have noticied: when mlt sets a zone to radiative, the diffusive mixing coefficient from mlt is set to 0, which prevents mixing. This is fine, but it is not clear how this should behave with rotation. Should the rotational mixing coefficient also be set to 0?

This can be problematic when controls like max_abs_du_div_cs_for_convection are used to limit convection over shocks. While convection across the shock is prohibited, rotation and potentially RTI can still mix zones across the shock interface, which can result in all sorts of issues.

see:

mesa/star/private/turb_info.f90

Lines 291 to 332 in 735d8fe

	subroutine set_no_mixing(str)
	character (len=*) :: str
	include 'formats'
	s% mlt_mixing_type(k) = no_mixing
	s% grada_face_ad(k) = grada_face_ad
	s% grada_face(k) = grada_face_ad%val
	gradT_ad = gradr_ad
	s% gradT_ad(k) = gradT_ad
	s% gradT(k) = s% gradT_ad(k)%val
	Y_face_ad = gradT_ad - grada_face_ad
	s% Y_face_ad(k) = Y_face_ad
	s% Y_face(k) = s% Y_face_ad(k)%val
	s% mlt_vc_ad(k) = 0d0
	if (s% okay_to_set_mlt_vc) s% mlt_vc(k) = 0d0
	s% mlt_D_ad(k) = 0d0
	s% mlt_D(k) = 0d0
	s% mlt_cdc(k) = 0d0
	s% mlt_Gamma_ad(k) = 0d0
	s% mlt_Gamma(k) = 0d0
	s% gradr_ad(k) = gradr_ad
	s% gradr(k) = s% gradr_ad(k)%val
	s% gradL_ad(k) = 0d0
	s% gradL(k) = 0d0
	s% scale_height_ad(k) = scale_height_ad
	s% scale_height(k) = scale_height_ad%val
	s% Lambda_ad(k) = mixing_length_alpha*scale_height_ad
	s% mlt_mixing_length(k) = s% Lambda_ad(k)%val
	s% L_conv(k) = 0d0
	end subroutine set_no_mixing

What does this affect? : Mainly rotating ppisn or cc models or really any model that is rotating and also has a shock interface.

What is the solution? : We need some check in mix_info.f90 that sets the diffusion coefficient to 0 when max_abs_du_div_cs_for_convection.

Perhaps somewhere inside

mesa/star/private/mix_info.f90

Line 48 in 735d8fe

subroutine set_mixing_info(s, skip_set_cz_bdy_mass, ierr)