source
stringlengths
1
2.05k
target
stringlengths
1
11.7k
Note that the computed errors in (BJ—VJ, for the field stars are at least 5 times larger than for all other stars.
Note that the computed errors in $(B-V)_0$ for the field stars are at least 5 times larger than for all other stars.
For NGC 6823 230 (open diamond) only a single value for V. and. D ean be found in
For NGC 6823 230 (open diamond) only a single value for $V$ and $B$ can be found in
huncdred: particles inside the softening radius (Romeo A.C. 1997).
hundred particles inside the softening radius (Romeo A. G. 1997).
We consider a Plummer softening parameter and keep it constant along the simulation.
We consider a Plummer softening parameter and keep it constant along the simulation.
Ht turns out ot be 510.7. ‘To mimic the infall of gas inside the potential well of the halo. we distribute gas particles on the top of DAL particles.
It turns out ot be $5 \times 10^{-3}$ To mimic the infall of gas inside the potential well of the halo, we distribute gas particles on the top of DM particles.
The barvonic fraction adopted is fj=0.1. and gas particles are Plummersoftened in the same way as the DM Under cooling and. because of the velocity. [field of the halo. the gas is expected to settle down in a rotating thin The results are shown in Fig.
The baryonic fraction adopted is $f_{b}~=0.1~$, and gas particles are Plummer–softened in the same way as the DM Under cooling and because of the velocity field of the halo, the gas is expected to settle down in a rotating thin The results are shown in Fig.
SN. where upper panels refer to the evolution of the dark component. whereas lower panels refer to the σας.
8, where upper panels refer to the evolution of the dark component, whereas lower panels refer to the gas.
Left. panels show the evolution in the NoZ plane. while right. panels show the evolution in the XNY plane.
Left panels show the evolution in the $X-Z$ plane, while right panels show the evolution in the $X-Y$ plane.
Due to the spin the dark halo Ilattens (upper-left panel). whereas the gas settles in a thin disk (bottom-left panel) which exhibits some spiral arms (lower-right These results are similar to those we have obtained with our scalar Tree-SPLII code (Lia et al This simulation is by no means aimed. at. producing a real disk. but only to show that cooling is) properly implemented.
Due to the spin the dark halo flattens (upper-left panel), whereas the gas settles in a thin disk (bottom-left panel) which exhibits some spiral arms (lower-right These results are similar to those we have obtained with our scalar Tree-SPH code (Lia et al This simulation is by no means aimed at producing a real disk, but only to show that cooling is properly implemented.
A similar series of simulations. but. without cooling. have been run. aiming at checking how the code
A similar series of simulations, but without cooling, have been run, aiming at checking how the code
evolution of 53 chemical species from !H to "CI using the ominal NACRE nuclear reaction rates by default and those given in Appendix A otherwise.
evolution of 53 chemical species from $^{1}$ H to $^{37}$ Cl using the nominal NACRE nuclear reaction rates by default and those given in Appendix A otherwise.
The equation of state is described 1n details in and accounts for the non-ideal effects due to Coulomb interactions and pressure. ionisation.
The equation of state is described in details in and accounts for the non-ideal effects due to Coulomb interactions and pressure ionisation.
The treatment of convection is based on the classical mixing lengthformalism with αν=1.6. and no convective overshoot is icluded.
The treatment of convection is based on the classical mixing lengthformalism with $\alpha_{\rm MLT} = 1.6$, and no convective overshoot is included.
The mass loss rates are computed with formula (with Ne = 0.5) up to the early-AGB phase. and with prescription on the TP-AGB.
The mass loss rates are computed with formula (with $\eta_R$ = 0.5) up to the early-AGB phase, and with prescription on the TP-AGB.
The thermohaline instability occurs in a stable stratification that satisfies the Ledoux criterion for convective instability: but where the molecular weight decreases with depth: with the classical notations for V=(¢@InT/dInP). @=(dInp/dInj)py and 0—(dinp/dInTipy. Vy and Vag being respectively the molecular weight gradient. and the adiabatic gradient.
The thermohaline instability occurs in a stable stratification that satisfies the Ledoux criterion for convective instability: but where the molecular weight decreases with depth: with the classical notations for $\nabla=(\partial \ln {\rm T} / \partial \ln {\rm P})$, $\varphi=(\partial \ln \rho / \partial \ln \mu)_{{\rm P,T}}$ and $\delta=-(\partial \ln \rho / \partial \ln {\rm T})_{{\rm P},\mu}$, $\nabla_\mu$ and $\nabla_{\rm ad}$ being respectively the molecular weight gradient and the adiabatic gradient.
For the turbulent diffusivity produced by the thermohaline instablity we use the prescription advocated by 6207 based on arguments for the aspect ratio & (length/width) of the salt fingers as supported by laboratory experiments and including extended expression for the case of a non-perfect gas (including radiation pressure. degeneracy): with Κ the thermal diffusivity.
For the turbulent diffusivity produced by the thermohaline instablity we use the prescription advocated by CZ07 based on arguments for the aspect ratio $\alpha$ (length/width) of the salt fingers as supported by laboratory experiments and including extended expression for the case of a non-perfect gas (including radiation pressure, degeneracy): with K the thermal diffusivity.
and with a=5 (Ulrich 1972) this coefficient is C(2658.
and with $\alpha = 5$ (Ulrich 1972) this coefficient is ${\rm C_t}$ =658.
For consistency reasons we assume actually C,21000 as in CZ07.
For consistency reasons we assume actually ${\rm C_t}$ =1000 as in CZ07.
For the treatment of rotation-induced mixing we proceed as follows.
For the treatment of rotation-induced mixing we proceed as follows.
Solid-body rotation 1s assumed when the star arrives on the zero age main sequence (ZAMS).
Solid-body rotation is assumed when the star arrives on the zero age main sequence (ZAMS).
Typical initial (1.8. ZAMS) rotation velocities are chosen depending on the stellar mass based on observed rotation distributions in young open clusters(?).
Typical initial (i.e., ZAMS) rotation velocities are chosen depending on the stellar mass based on observed rotation distributions in young open clusters.
. Surface braking by a magnetic torque is applied for stars with an effective temperature on the ZAMS lower than 6900 K that have relatively a thick convective envelope as discussed in?) and?:: the adopted braking law follows the deseription of?..
Surface braking by a magnetic torque is applied for stars with an effective temperature on the ZAMS lower than 6900 K that have relatively a thick convective envelope as discussed in and; the adopted braking law follows the description of.
From the ZAMS on the evolution of the internal angular momentum profile is accounted for with the complete formalism developed by and that takes into account advection by meridional circulation and diffusion by shear turbulence (for a deseription of the implementation in STAREVOL. see?..?.. and ?)).
From the ZAMS on the evolution of the internal angular momentum profile is accounted for with the complete formalism developed by and that takes into account advection by meridional circulation and diffusion by shear turbulence (for a description of the implementation in STAREVOL, see, and ).
The transport of chemicals resulting from meridional circulation and both horizontal and vertical turbulence is computed as a diffusive process throughout the evolution.
The transport of chemicals resulting from meridional circulation and both horizontal and vertical turbulence is computed as a diffusive process throughout the evolution.
The complete treatment for the transport of angular momentum and chemicals is applied up to the RGB tip or up to the second dredge-up for the stars with initial masses below or above 2.0 M. respectively.
The complete treatment for the transport of angular momentum and chemicals is applied up to the RGB tip or up to the second dredge-up for the stars with initial masses below or above 2.0 $_{\odot}$ respectively.
The convective envelope is supposed to rotate as a solid body (uniform angular velocity) throughout the evolution: we discuss the implications of this assumption in 3.2.2.
The convective envelope is supposed to rotate as a solid body (uniform angular velocity) throughout the evolution; we discuss the implications of this assumption in 3.2.2.
The transport of angular momentum by internal gravity waves (which is efficient only in main sequence stars with effective temperatures on the ZAMS lower than 6500 K. see ?)). is neglected.
The transport of angular momentum by internal gravity waves (which is efficient only in main sequence stars with effective temperatures on the ZAMS lower than 6500 K, see ), is neglected.
In the present work the transport coefficients for chemicals associated to thermohaline and rotation-induced mixings are simply added in the diffusion equation and we do not consider the possible interactions between the two mechanisms. nor with magnetic diffusion.
In the present work the transport coefficients for chemicals associated to thermohaline and rotation-induced mixings are simply added in the diffusion equation and we do not consider the possible interactions between the two mechanisms, nor with magnetic diffusion.
As a matter of fact under the present assumptions the thermohaline diffusion coefficient is several orders of magnitude higher than the total diffusion coefficient üssoclatec to rotation (see $33).
As a matter of fact under the present assumptions the thermohaline diffusion coefficient is several orders of magnitude higher than the total diffusion coefficient associated to rotation (see 3).
This is confirmed by who also show that magnetic diffusion in RGB stars is much less efficient than thermohaline mixing.
This is confirmed by who also show that magnetic diffusion in RGB stars is much less efficient than thermohaline mixing.
However. we should keep an eye on future hydrodynamic calculations that are required to evaluate with confidence the possible interactions of thermohaline fingers with differential rotation and magnetic fields in red giants.
However, we should keep an eye on future hydrodynamic calculations that are required to evaluate with confidence the possible interactions of thermohaline fingers with differential rotation and magnetic fields in red giants.
We first consider the case of low-mass stars that ignite burning by a flash at the tip of the RGB well above (in terms of luminosity) the bump.
We first consider the case of low-mass stars that ignite helium-burning by a flash at the tip of the RGB well above (in terms of luminosity) the bump.
With the considered metallicity and input physics this corresponds to stars with initial masses below ~2.2M..
With the considered metallicity and input physics this corresponds to stars with initial masses below $\sim$ 2.2 $_{\odot}$ .
We present detailed predictions fora 1.22 M. model computed without and with rotation (but with thermohaline
We present detailed predictions for a 1.25 $_{\odot}$ model computed without and with rotation (but with thermohaline
Spifzer/GLIAIPSL.
/GLIMPSE.
The cloud is seen in absorption against the bright background at δ qun ancl 70 pum. and in emission at longer wavelengths (Eig. 10)).
The cloud is seen in absorption against the bright background at 8 $\micron$ and 70 $\micron$, and in emission at longer wavelengths (Fig. \ref{fig:2918obs}) ).
The SED of the cloud is shown in Fig. 12..
The SED of the cloud is shown in Fig. \ref{fig:2918sed}.
Guided. by the image of the cloud at 500. pum. we model this cloud using a 2D density profile (eq. 2))
Guided by the image of the cloud at 500 $\micron$, we model this cloud using a 2D density profile (Eq. \ref{eq:flatdens}) )
with a ratio of optical depths T,τμ=1.55.
with a ratio of optical depths $\tau_{\theta=90\degr}/\tau_{\theta=0\degr}=1.55 $.
The inpu parameters of the model ancl the derived physical properties of the cloud are listed in Table 3..
The input parameters of the model and the derived physical properties of the cloud are listed in Table \ref{tab:2819}.
Phe mass of the clou is constrained by the longer. wavelength. data. which are approximately column clensitw profiles. as the οσοι of the temperature gradient is relatively weak.
The mass of the cloud is constrained by the longer wavelength data, which are approximately column density profiles, as the effect of the temperature gradient is relatively weak.
The temperature of the cloud is mainly constrained by the shorter wavelength data.
The temperature of the cloud is mainly constrained by the shorter wavelength data.
To mateh these short wavelength. data. the externa raciation field is enhanced by a factor of μη=2.5. when compared with the SRE at the solar neighbourhood. which is consistent with the higher ambient radiation field in the Galactic plane.
To match these short wavelength data, the external radiation field is enhanced by a factor of $f_{\rm ISRF}=2.5$, when compared with the ISRF at the solar neighbourhood, which is consistent with the higher ambient radiation field in the Galactic plane.
The model reproduces well the images of the clou at different wavelengths (Eig. 11)).
The model reproduces well the images of the cloud at different wavelengths (Fig. \ref{fig:2918model}) ).
At shorter wavelengths (8. TO pun) the cloud. is seen in absorption (note tha the background. is not. modelled. hence there is no actua correspondence with the observed. background). ancl in emission at longer wavelengths 170 tum).
At shorter wavelengths (8, 70 $\micron$ ) the cloud is seen in absorption (note that the background is not modelled, hence there is no actual correspondence with the observed background), and in emission at longer wavelengths $\ga170~\micron$ ).
The flux anc the shape of the eloud are well matched (assuming a viewing angle of Gn.= 407).
The flux and the shape of the cloud are well matched (assuming a viewing angle of $\theta_{\rm obs}=40\degr$ ).
At 170 pum there is secondary peak tha is not. reproduced. by the model.
At 170 $\micron$ there is secondary peak that is not reproduced by the model.
“Phis is probably due to asvmmetric heating as this secondary. peak does not appear at longer wavelengths (cf.
This is probably due to asymmetric heating as this secondary peak does not appear at longer wavelengths (cf.
Nutter et al.
Nutter et al.
2009): such heating is not included in the current model.
2009); such heating is not included in the current model.
The observed SED is well fitted by the model (Eis. 12..
The observed SED is well fitted by the model (Fig. \ref{fig:2918sed},
dotted line).
dotted line).
The dust temperature inside the cloud. drops from to 21 Ix at its edge to 10 Ix at its centre (Fig. 13:
The dust temperature inside the cloud drops from to 21 K at its edge to 10 K at its centre (Fig. \ref{fig:2918temp};
cf.
cf.
Peretto et al.
Peretto et al.
2010).
2010).
Acditionallv. there is a variance of the temperature with the polar angle 8. with the more dense region at the “equator” of the cloud being colder by up to ~3 dx. às compared with the corresponding ecqual-racdius region at the “pole” of the cloud.
Additionally, there is a variance of the temperature with the polar angle $\theta$ , with the more dense region at the "equator" of the cloud being colder by up to $\sim 3$ K, as compared with the corresponding equal-radius region at the "pole" of the cloud.
The mass of the cloud is calculated to be 530 Mi.
The mass of the cloud is calculated to be 530 $_{\sun}$.
We also compare the radiative transfer model with a simple grev-body. sinele-temperature fitting of the SED.
We also compare the radiative transfer model with a simple grey-body single-temperature fitting of the SED.
The SED is well reproduced using a temperature of 16 Ix (Fig. 12..
The SED is well reproduced using a temperature of 16 K (Fig. \ref{fig:2918sed}, ,
dashed line)
dashed line).
The mass of cloud. calculated using the above temperature and the Dux at 500. qun. is 520 AL... which is slightly lower but consistent with the mass obtained. from the radiative transfer model. despite using an average. and not the actual. dust temperature.
The mass of cloud calculated using the above temperature and the flux at 500 $\micron$, is 520 $_{\sun}$, which is slightly lower but consistent with the mass obtained from the radiative transfer model, despite using an average, and not the actual, dust temperature.
Llowever. he inner dense regions of the cloud. where star formation will occur are colder (bv ~5 Ix) than the estimated average empoerature.
However, the inner dense regions of the cloud where star formation will occur are colder (by $\sim$ 5 K) than the estimated average temperature.
This example. demonstrates the need. for. radiative ransfer modelling for determining the temperature structure of HUDC's.
This example demonstrates the need for radiative transfer modelling for determining the temperature structure of IRDCs.
Despite the [act that a simple grev-body sinele-temperature model provides a &ood fit to the SED. his model cannot accurately determine the temperature structure in the cloud. and overestimates the temperature of the interesting. in terms of potential for star formation. inner dense regions of the cloud.
Despite the fact that a simple grey-body single-temperature model provides a good fit to the SED, this model cannot accurately determine the temperature structure in the cloud and overestimates the temperature of the interesting, in terms of potential for star formation, inner dense regions of the cloud.
We have demonstrated the use of the 3D. multi-wavelength Monte Carlo code.PUAETHON. to model the transfer. of radiation in IRDCs. that are externally illuminated by the interstellar raciation Ποια.
We have demonstrated the use of the 3D, multi-wavelength Monte Carlo code, to model the transfer of radiation in IRDCs, that are externally illuminated by the interstellar radiation field.
The cores of these clouds: are believed to be where high-mass stars form. ie. they are the high-mass equivalent of prestellar clouds.
The cores of these clouds are believed to be where high-mass stars form, i.e. they are the high-mass equivalent of prestellar clouds.
We have presented three widely cillerent models. in which we varied the mass. density. racius. morphology and internal velocity field of the cloud.
We have presented three widely different models, in which we varied the mass, density, radius, morphology and internal velocity field of the cloud.
We have shown the predicted. output of the model ab the wavebands of{Herschel ancSpizer.
We have shown the predicted output of the model at the wavebands of and.
We also passed the model output through the SPIRE simulator to produce simulated: observations of these ItDCSs.
We also passed the model output through the SPIRE simulator to produce simulated observations of these IRDCs.
These were then analysed as ifthey were real observations.
These were then analysed as if they were real observations.
Subsequently. the results of this analysis were compared with the results of the radiative transfer mocelling.
Subsequently, the results of this analysis were compared with the results of the radiative transfer modelling.
Our study. highlights the need for. detailed radiative transfer modelling when using multi-wavelength observations [romHerschel to accurately determine the properties of LRDCs.
Our study highlights the need for detailed radiative transfer modelling when using multi-wavelength observations from to accurately determine the properties of IRDCs.
This method was applied for the study of 29.55|00.18. an Επ from the Li-GAL survey.
This method was applied for the study of G29.55+00.18, an IRDC from the Hi-GAL survey.
Aodelling of a larger sample of LUDCSs found in the Lli-GAL survey will appear in a future publication (Wilcock οἱ αἱ.
Modelling of a larger sample of IRDCs found in the Hi-GAL survey will appear in a future publication (Wilcock et al.,
in prep).
in prep).
We would like to thank the referee. for his report. that helped improving the original manuscript.
We would like to thank the referee for his report that helped improving the original manuscript.
Simulations were performed. using the Cardiff! LPC ClusterMERLIN.
Simulations were performed using the Cardiff HPC Cluster.
. The colour plots of Fig.
The colour plots of Fig.
5 were produced. using (Price 2007).
5 were produced using (Price 2007).
DS and JAIN acknowledge post-doctoral support from the Science TechnologyFacilities Council (SPEC)uncler the auspices of the Cardiff Astronomy. Rolling Cirant.
DS and JMK acknowledge post-doctoral support from the Science TechnologyFacilities Council (STFC)under the auspices of the Cardiff Astronomy Rolling Grant.
Since verv compact and luminous stellar systems. were discovered. in the central region of the LFornax cluster of galaxies (e.g. Lilker et al.
Since very compact and luminous stellar systems were discovered in the central region of the Fornax cluster of galaxies (e.g, Hilker et al.
1999: Drinkwater et al.
1999; Drinkwater et al.
2000a. b). physical properties of these systems now referred to as. “ultra-compact dwarl (UCD) galaxies have been extensively investigated by observational stucdies (o.g.. Phillipps et al.
2000a, b), physical properties of these systems – now referred to as “ultra-compact dwarf” (UCD) galaxies – have been extensively investigated by observational studies (e.g., Phillipps et al.
2001: Mieske et al.
2001; Mieske et al.
2002. 2004. 2006: Drinkwater et al.
2002, 2004, 2006; Drinkwater et al.
2003: Llaseegan ct al.
2003; Haşeegan et al.
2005: Ixarick οἱ al.
2005; Karick et al.
2006: Firth et al.
2006; Firth et al.
2006).
2006).
These observations have reported very unique properties of LCDs. such as very compact sizes («LOO pe). possibly higher mass-to-light-ratio (AL/L) indicative of the presence of cark matter. anc scaling relations of dynamical properties (e.g.. internal velocity dispersions) dillerent from those of CC's (e.g. Drinkwater et al.
These observations have reported very unique properties of UCDs, such as very compact sizes $<100$ pc), possibly higher mass-to-light-ratio $M/L$ ) indicative of the presence of dark matter, and scaling relations of dynamical properties (e.g., internal velocity dispersions) different from those of GCs (e.g., Drinkwater et al.
2003: Llagecgan et al.
2003; Haşeegan et al.
2005).
2005).
Physical properties of UCDs in dillerent clusters and. groups of galaxies are now being investigated (e.g. Jones ct al.
Physical properties of UCDs in different clusters and groups of galaxies are now being investigated (e.g., Jones et al.
2006: Ixilborn et al.
2006; Kilborn et al.
2005)
2005).
In spite of these observational progresses. it remains unclear how LCDs formec and. evolved. in the. central regions of clusters of galaxies (e.g. Bekki et al.
In spite of these observational progresses, it remains unclear how UCDs formed and evolved in the central regions of clusters of galaxies (e.g., Bekki et al.
2001. 2003a: Fellhauer Ixroupa 2002).
2001, 2003a; Fellhauer Kroupa 2002).
Bekki et al. (
Bekki et al. (
2001. 2003a) proposed the "galaxy threshing" scenario in which UCDs originate from. nuclei of nucleated clwarl galaxies that had been destroved by strong tidal fields of clusters of galaxies.
2001, 2003a) proposed the “galaxy threshing” scenario in which UCDs originate from nuclei of nucleated dwarf galaxies that had been destroyed by strong tidal fields of clusters of galaxies.