Home

Torus User Manual - University of Exeter

1. Parameter Default value Default Units Unit Type Description outputfile none N A The output grid filename inputfile none N A The input grid filename readgrid false N A Read in an input grid required writegrid false N A Write an output grid required justdump false N A Dump a vtk vtu file for a given grid amrid false N A The AMR grid is 1 dimensional amr2d false N A The AMR grid is 2 dimensional amr3d false N A The AMR grid is 3 dimensional dorefine true N A Allow the AMR grid to refine adaptively dounrefine true N A Allow the AMR grid to coarsen adaptively mindepthamr 5 N A Minimum refinement depth of AMR grid maxdepthamr 31 N A Maximum refinement depth of AMR grid maxdepthamr 31 N A Maximum refinement depth of AMR grid amrgridsize 1000 10 10cm distance The linear size of the AMR grid amrgridcentrex 0 10 10cm distance AMR grid centre x coordinate amrgridcentrey 0 10 10cm distance AMR grid centre y coordinate amrgridcentrez 0 10 10cm distance AMR grid centre z coordinate griddistancescale 1 d10 cm Distance scale of grid required cylindrical false N A Grid uses 3D cylindrical coordinates geometry sphere N A Name of model to run splitovermpi false N A Split grid across threads binaryxml true N A Use binary XML VTK files refineonmass false N A Refine AMR grid using cell mass refineontemperature false N A Refine AMR gri
2. Dimensionality Nthreads Description 1D 3 grid split into 2 plus 1 control thread 2D 5 grid split into 4 plus 1 control thread 2D 17 grid split into 16 plus 1 control thread 3D 9 grid split into 8 plus 1 control thread 3D 65 grid split into 64 plus 1 control thread There is an automatic check in place so if you do not use one of the above then the model will not run The initial generation of photon packets is performed by the governor rank This stores photon packets that are ready in a stack until there is a predefined number ready to send to the appropriate domain thread The domain thread then receives the stack and starts processing photons as outlined above In the event that a photon packet crosses the boundary between two computational domains it is added to a to send stack that is owned by the domain rank Similarly to before once this to send stack reaches a predefined size it will be sent to the neighbouring thread Photon packets are communicated in these stacks to reduce the number of times that threads have to communicate with one another as this can significantly improve the computation time To run domain decomposed photoionization models simply include splitovermpi T photoionphysics T photoioneq T in the parameters file You will also have to ensure that you are actually trying to run the job in parallel If you are using zen see the zen wiki 5 1 2 The Stack Limi
3. radiationHydrodynamics false Perform a radiation hydrodynamics calculation TABLE tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 4666666 databg FAFOD4 F3DFA8 headerrows 2 image false Calculate an image nphotons 10000 Number of photons in image distance 100 Grid distance from observer nimage 1 Number of images to calculate imageaxisunits AU Axis units for image AU pc cm arcsec imagesize whole grid imageaxisunits Size of image imageaspect 1 0 Aspect ratio of image polimage false Write polarization images imagefile none Output image filename lambdaimage 6562 8 Angstroms Wavelength for monochromatic image imagetype none Type of output image npixels 200 Number of pixels per side in image inclination 0 degrees Inclination angle positionangle 0 degrees position angle fitsbitpix 32 BITPIX value for FITS images and cubes SED parameters TABLE tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 666666 databg FAFOD4 F3DFA8 headerrows 2 96 38 CHAPTER 10 TORUS INPUT PARAMETERS 10 1 UNIT DIRECTORY spectrum false Calculate a spectrum nphotons 100900 Number of photons in SED inclinations none Inclination angles posa
4. 2 37 CHAPTER 10 TORUS INPUT PARAMETERS photoionphysics false N A Include photoionization physics in calculation quickthermal false N A compute photoionization equilibrium usemetals true N A Use metals Z gt 2 in photoionization calculation hOnly false N A Hydrogen only model nodiffuse false N A Turn off the diffuse radiation field monochromatic false N A Use a monochromatic 13 6 10 5eV radiation field bufferCap 5000 N A Number of photon packet stacks accommodated in to send buffer periodicX false Periodic photon boundaries in the x direction periodicY false Periodic photon boundaries in the y direction periodicZ false Periodic photon boundaries in the z direction biasToLyman false Increase sampling of ionizing photons biasMagnitude 100 d0 Magnitude of bias towards ionizing photons binPhotons false Dump a spectrum formed from processed photon packets h_abund 1 X H X H Hydrogen abundance he_abund 0 1 X He X H Helium abundance c abund 22 e 5 X C X H Carbon abundance n_abund 4 e 5 X N X H Nitrogen abundance o_abund 33 e 5 X O X H Oxygen abundance ne_abund 5 e 5 X Ne X H Neon abundance s_abund 0 9e 5 X S X H Sulphur abundance TABLE tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 666666 databg 4FAFOD4 F3DFA8 headerrows 2
5. 5 The physics modules 14 CRINE sop C ce be od E hod wy ae dpb te ie heats BES 14 5 0 16 Describing the size and chemistry of the dust 14 5 0 17 Setting up the initial conditions for the Lucy iterations 15 50 18 convergence gt ena a vo og a ee ee ENEE 15 5019 Common talk ss m ode dee RR eileen nite 64846 eden A 15 50 20 Calculating 6 de dede he inp ete of eae a et de et e d Ge ntl es Ge oh ee d 16 5 021 alculatig Wages e a turc eger gl E Rem EEN 17 54122 Qutput Hes uou 4s 2k oer ee ee x es Reus BE Ow dome 3 18 5 0 23 Molecular Line Transfer Calculations 2 a 18 5 0 24 Atomic Line Transfer Calculations 2 a 19 51 Photomnization Models o m o sex mx x OX 9 ed 20 5 11 Domain Decomposed Photoionization 5 20 DIZ The Aak ETT 20 5 1 3 Monochromatic Models domain decomposed models 21 5 14 On The Spot Approximation x c saaa sasad ELE EE athe x E Ra oes 22 5 1 5 Simplified Thermal Calculation 2 2 sanaa e sss 22 5 16 Variance Reduction sHs sese a ee aaa a a Ee de e e cn ee ate ee 22 5 1 7 Periodic Boundary Conditions domain decomposed models 23 5 1 8 Customizing Atomic Species ARR eee dde 23 5 1 9 Retrieving data from g
6. Once Profile Mode is changed from none to any other option a profile viewing window will pop up see next image for example To view spectral lines wherever your cursor is choose line as the profile mode and be sure the track cursor box is checked As you move the mouse around the main display window the line profile will auto update in the profile window 12 CHAPTER 4 PLOTTING TORUS DATA e KvisScreenShot3 Auto V Zoom scales the intensity axis on the fly depending on where your cursor is uncheck this and click Unzoom to fix the y axis display to the full range To adjust the display settings in the profile window click Overlay and select axis labels This box will pop up e KvisScreenShot4 Once enable is checked select display settings as desired Additional notes e To generate a box average of the spectral line in a given region change Profile Mode to average and middle click drag alt click drag on macs to define a rectangular region on the image After a moment the averaged line profile within your defined region will appear in the profile window e kvis can also generate movies click Movie in the view control window The default setting is for the X Y plane to be the slice direction with the movie progressing through z values in the data cube in our case velocities e The scroll wheel on the mouse can be used to zoom in out in the main display window and con trol left click drag can
7. 2 Protoplanetary disc geometry Use the shakara geometry to set up a flared protostellar disc The parameters of the disc can be specified using the keywords listed in the table below If a gap low density annulus is specified then the mass of the disc integrated over the disc volume will not match mdisc Parameter Units Description mdisc solar masses The total mass of the disc gas plus dust alphadisc Disc density goes as r alpha betadisc Disc scaleheight goes as r beta rinner stellar radius sourcel Inner disc radius router AU Outer disc radius height AU Disc scaleheight at 100 AU rgapinner AU Inner radius of gap rgapouter AU Outer radius of gap rhogap g cm 3 Minimum density in the gap smoothinneredge Exponential decay of density at the inner edge Complete Responsible Author Alicia Aarnio 07 Jan 2011 Contributors Alicia Aarnio Last significantly modified by Alicia Aarnio 07 Jan 2011 Not yet reviewed 8 0 3 ttauri geometry Setting geometry ttauri provides a geometry for protostellar accretion and outflow This geometry can be broken down into three components e he magnetosphere A dipole magnetosphere that can have a dipole offset compared to the rotation axis always the z direction The footprints of the magnetospheric accretion stream produce hotspots on the surface of the protostar as sumed to be source number one placed at the origin e he dis
8. Assuming that you complete compilation without error messages you can check that your torus executable is doing the right thing using the benchmark tests that torus runs on an nightly basis The parameter files and scripts to check the test results are located at torus benchmarks CHAPTER 2 BUILDING TORUS The tests themselves cover different aspects of the code including hydrodynamics torus benchmarks hy drold photoionization HIl region HIl_regionMPI and molecular line transport molebench Copy your torus executable into a benchmark folder run the code and check the results Complete Responsible Author David Acreman 13 Feb 2009 Contributors David Acreman Last significantly modified by David Acreman 13 Feb 2009 Not yet reviewed CHAPTER 3 RUNNING TORUS 3 0 5 Running torus The torus distribution contains a variety of data files such as model atmospheres grain refractive indices and stellar evolution tracks These are stored in the data subdirectory of the torus distribution In order for the code to access these the environment variables TORUS DATA must be set e g in csh setenv TORUS DATA mypathtotorus torus data Once you have compiled the code you will have an executable called torus xxx where xxx is defined by your SYSTEM type e g torus gfortran In order to run the code you need to execute the torus executable in a directory that contains a parameters input file
9. Romanova M M Harries Tim J 3D simulations of rotationally induced line variability from a classical T Tauri star with a misaligned magnetic dipole 2008 MNRAS 385 1931 Tannirkulam A Monnier J D Millan Gabet R Harries T J Pedretti E ten Brummelaar T A McAlister H Turner N Sturmann J Sturmann L Strong Near Infrared Emission Interior to the Dust Sublimation Radius of Young Stellar Objects MWC 275 and AB Aurigae 2008 ApJ 677L 51 Hatchell J Fuller A Richer J S Harries T J Ladd E F Star formation in Perseus Il SEDs classification and lifetimes 2007 A amp A 468 1009 Tannirkulam A Harries T J Monnier J D The Inner Rim of YSO Disks Effects of Dust Grain Evolution 2007 ApJ 661 374 42 CHAPTER 11 TORUS PUBLICATIONS 18 19 20 21 22 23 24 25 Complete Monnier J D Tuthill P G Danchi W C Murphy N Harries T J The Keck Aperture masking Experiment Near Infrared Sizes of Dusty Wolf Rayet Stars 2007 ApJ 655 1033 Kurosawa Ryuichi Harries Tim J Symington Neil H On the formation of H amp 945 line emission around classical T Tauri stars 2006 MNRAS 370 580 Kurosawa Ryuichi Harries Tim J Symington Neil H Time series Paschen amp 946 spectroscopy of SU Aurigae 2005 MNRAS 358 671 Symington Neil H Harries Tim J Kurosawa Ryuichi Emission line profile modelling of structured T Tauri magnetospheres 20
10. be indicated in the A list of Torus input parameters section Any comments problems or feedback email haworth astro ex ac uk Key Files e inputs modV2 F90 e units mod f90 10 1 1 Available units Case sensitive Distances Default value 1010 Centimetres e cm 10 0 centimetres e m 10 metres e au Astronomical units 39 10 1 UNIT DIRECTORY CHAPTER 10 TORUS INPUT PARAMETERS e parsecs e rSol solar radii Wavelengths e Default value Angstroms e A Angstroms e nm Nanometers e um Microns e mm Millimetres Dust grain sizes e Default value microns e A Angstroms e nm Nanometers e um Microns e mm Millimetres Angles e Default value radians e rad radians e deg degrees e arcmin arcminutes e arcsec arcseconds Masses e Default value grams e g grams e mSol solar masses e kg kilograms Time e Default seconds e s seconds e yr years kyr kiloyears e Myr megayears Temperature e Default Kelvin e K Kelvin Luminosity e Default Ergs per second e ergsec ergs per second e Sol solar luminosities 40 CHAPTER 10 TORUS INPUT PARAMETERS 10 1 UNIT DIRECTORY Velocities e Default cm s lems ems 1 ms ms 16715 kms 1 c fraction of light speed Densities 10 1 2 Unit Types for developers Torus performs calculations using cgs units each kind of physical quantity has different units that torus will use When addin
11. be used to define a rectangular zoom area in the image 4 0 14 Plotting spectral energy distributions The creation of dat files containing spectra is described here subsection Calculating SEDs SEDs can be plotted using IDL with relative ease will attach sample code here soon Complete __ Responsible Author Alicia Aarnio 26 Aug 2010 Contributors Alicia Aarnio Last significantly modified by Alicia Aarnio 26 Aug 2010 Not yet reviewed 13 CHAPTER 5 THE PHYSICS MODULES 5 0 15 Introduction Torus uses the radiative equilibrium method of Lucy 1999 to calculate dust temperatures for arbitrary distributions of dust illuminated by multiple radiation sources New users should read and become familiar with the photon packet algorithm described in Lucy s paper The following chapter details the input keywords required to compute a radiative equilibrium model and to calculate emergent SEDs and images Some pitfalls and gotchas are also described which should help the novice user 5 0 16 Describing the size and chemistry of the dust The dust grain size distribution and chemistry must be specified using keywords if dustphysics T is set An arbitrary number of different grain sizes and chemistries can be listed by the keyword ndustype which requires an integer value The minimum grain size in microns is set using amin the maximum using amax the power law index as qdist graintypel draine sil Draine 2003 s
12. error per pixel with adaptive pixel subsampling isinlte logical Initialise molecular level populations assuming LTE setmaxlevel integer Manually set maximum number of levels to consider during convergence Notes molAbundance will be set to a constant by this parameter There are no geometry specific routines for setting the abundance of any molecular species at the moment There is a drop model for 13CO that can be accessed by adding doChemistry T Some geometries have their own molecular abundance reading routines which will override this default setting As for molAbundance the vturb parameter adds a constant non thermal broadening parameter to the line profile function Doing anything more clever than this will require an extra function In addition to these recommended parameters there are a number of others that can be set that turn on off some of the diagnostic information or convergence enhancing code that has been written to make TORUS quick and easy to use Complete Responsible Author David Rundle 02 Sep 2010 Contributors David Rundle Last significantly modified by David Rundle 02 Sep 2010 Not yet reviewed 5 0 24 Atomic Line Transfer Calculations Setting atomicphysics T in the parameters file will allow the user to performing co moving frame atomic line calculations in Torus There are two keywords that need to be defined natom defines how many differen
13. grid files periodically throughout hydrodynamics and RHD calculations The vtu files are used for visualisation in Vislt and the grid files are used as checkpoints from which the calculation can be restarted In long high resolution calculations the storage demands and time spent writing all of these files can be a problem In particular the grid files can get very large It is often the case that visualisation files are required more frequently than restart checkpoints By using the vtuToGrid parameter you can specify the vtu to grid file ratio If you had a parameters dat file which dumps both grid and vtu files every time t 10 you could change tdump 10 0 to tdump 1 0 vtuToGrid 10 then you will get the same number of grid files as in the original but 10 times more vtu files If you had 26 CHAPTER 5 THE PHYSICS MODULES 5 4 RADIATION HYDRODYNAMICS tdump 10 0 vtuToGrid 10 then you would get the same number of vtu files as in the original parameters file but 10 time fewer grid files Complete Responsible Author Thomas Haworth 13 Dec 2010 Contributors Thomas Haworth Last significantly modified by Thomas Haworth 13 Dec 2010 Not yet reviewed 5 4 Radiation Hydrodynamics Radiation hydrodynamics models combine the domain decomposed photoionization and hydrodynamics routines in torus to perform simulations where the evolution of the radiat
14. test for errors without actually running the model Here is an example TORUS input parameter file for the Pascucci circumstellar disc benchmark Torus v2 parameter file for 2D benchmark disc See Pascucci al 2004 A amp A 417 793 dustphysics T use dust microphysics radeq T perform a radiative equilibrium calculation AMR grid parameters readgrid F we aren t reading a grid we will set one up from scratch writegrid E we don t need to write out the AMR file we just need SEDs amrgridsize 4 e6 the linear size of the top level AMR mesh in units of 10 10 cm amr2d T this is a 2d cylindical model grid smoothing switches smoothgridtau T smooths the grid for optical depth in order to resolve disc photosphere dosmoothgrid T smooth the grid for jumps in cell refinement 7 CHAPTER 3 RUNNING TORUS smoothfactor 3 0 make sure that neighbouring cells are not only one AMR depth apart Source parameters nsource 1 there is just one source radiusl 1 it has a radius of 1 solar radius teffl 5800 the source effective temperature contfluxl blackbody the continuum flux is assumed to be a blackbody massl 1 the source has a mass of one solar mass sourceposl 0 0 O it is located at the grid centre Geometry specific parameters geometry benchmark this is the Pascucci 2004 benchmark rinner 1 inner disc radius AU router 1000 outer disc radius heig
15. the others particularly donorcell can be useful when debugging and developing 5 3 2 Rhie Chow Interpolation In a scheme where pressure is calculated by considering the difference between the i 1th and i 1th cells we can be left with two decoupled pressure fields This gives rise to a amp 8220 checkerboard effect amp 7 8221 and essentially leaves one with half the resolution that they are otherwise computing To get around this as TORUS had to one can use Rhie Chow interpolation This uses the pressure difference at a cell s own interfaces calculated by averaging with its neighbours and thus removing any possibility of decoupling Rhie Chow interpolation is turned on by default but can be turned off by using the following rhiechow F 5 3 3 Boundary Conditions Hydrodynamics models require boundary conditions These must be specified in order for your model to run The following should be specified as appropriate in your parameters file xplusboundstring xminusboundstring yplusboundstring yminusboundstring zplusboundstring zminusboundstring If you are running a 1D model then you only need to specify the x boundaries 2D models require specification of the x and z boundaries and 3D models require all of them These boundary specifiers should be matched by your desired option for the boundary condition from one of the following strings Boundary string Notes reflecting impinging m
16. 05 MNRAS 358 671 Vink Jorick S Harries T J Drew J E Polarimetric line profiles for scattering off rotating disks 2005 A amp A 430 213 Kurosawa Ryuichi Harries Tim J Bate Matthew R Symington Neil H Synthetic infrared images and spectral energy distributions of a young low mass stellar cluster 2004 MNRAS 351 1134 Harries Tim J Monnier John D Symington Neil H Kurosawa Ryuichi Three dimensional dust radiative transfer models the Pinwheel Nebula of WR 104 2004 MNRAS 350 565 Harries T J Babler B L Fox G K The polarized spectrum of the dust producing Wolf Rayet O star binary WR137 2000 A amp A 361 273 Harries Tim J Synthetic line profiles of rotationally distorted hot star winds 2000 MNRAS 315 722 Responsible Author Tim Harries 05 May 2010 Contributors Tim Harries Last significantly modified by Tim Harries 05 May 2010 Not yet reviewed 43
17. 800 total buffer size is 504475t You will then have to manually increase the buffer size from its default of 5000 photon stacks not packets in the buffer To do this include bufferCap 100 in your parameters file replacing 100 with whatever number of stacks you want available space for in the buffer This issue should not arise very often if at all 5 1 3 Monochromatic Models domain decomposed models only You can run monochromatic models by including monochromatic T in the parameters file This uses photons with energy 13 6 10 5 eV only 21 5 1 PHOTOIONIZATION MODELS CHAPTER 5 THE PHYSICS MODULES 5 1 4 On The Spot Approximation To neglect the diffuse field photons due to recombination events include the following in your parame ters dat file nodiffuse T it otherwise treated by default 5 1 5 Simplified Thermal Calculation You can implement a faster simplified thermal calculation This should only really be used for Hydrogen only models since it assigns temperature by interpolating between pre assigned fully ionized and fully neutral gas temperatures TORUS uses 10000K and 10K for these states respectively To use this simplified thermal calculation include quickthermal T in your parameters file 5 1 6 Variance Reduction You can improve your signal to noise by propagating more ionizing photons energy gt 13 6eV wavelength lt 912A
18. For example if you have a directoy home myusername model then cd into that directory and run torus my path to torus torus xxx Torus will then run and will look for a file called parameters dat If you have called your parameters file something different e g mymodel dat then my path to torus torus xxx mymodel dat will run torus on mymodel dat Complete Responsible Author Tim Harries 24 Aug 2010 Contributors Tim Harries Last significantly modified by Tim Harries 24 Aug 2010 Not yet reviewed 3 0 6 Creating a parameters file The parameters file basically consists of a list of keywords and associated values The keywords are case sensitive and should appear as the first non whitespace characters on a given line of the input file The next non whitespace characters should be the value associated with that keyword Characters after the keyword value pair are ignored and lines that start with a or keyword are also ignored Keywords may appear in any order in the input file although for clarity you will wish to collect keywords referring to particular part of the model together This gives significant flexibility in formatting the TORUS parameter file for maximum clarity Note that each unique keyword should appear only once in the parameters file As an example say we have a keyword nphotons which we wish to assign a value of 1000 We could write this as CHAPTER 3 RUNNING TORUS nphotons 1000 as one l
19. Torus User Manual 02 Aug 2013 Contributors Alicia Aarnio David Acreman Tim Harries Thomas Haworth David Rundle TWiki to PDF Converter TWiki2pdf Version 2 3 4 by Steve Lloyd Copyright 2004 2013 by the contributing authors CONTENTS Preface 1 Introduction 2 Building Torus 20 1 Get ng the Torus code cn ea ee ck rae erae E eden ete igs 205 Compie x Bede eer so Gace Seva cae Sp GEE de ay oe deve LEGE EE 203 Building TOUS A ao oom ox o n RP Dee EG EORR Rue 204 Testing FOUR RUNG du a eek cde E ip ine Roe Sp Rm RE APR IN Ge get tees 3 Running Torus 306 R nning DORMS uus eom eR Ge RU Ron Ux Ree wow kt ex UR OR Ge OR mre F 306 Creatine parameters file uo oue uuo eat Red a ee A n 3 0 Top level keywords gt gt 2 622 484 op eoe ERR ER REOR R moy r 30 3 Settingupthe AMR grid Ls be cia be Rok ook ERR Pe eed SOG Geen ee hah be 4 ee ee eb ee GG h T 3 10 Physics bo ee 3e mex ER ee Ee ER a ee SOIL Type calcula de Lese og cee ks ox d A e er e ee EE GE 3 0 12 Checking your parameters file ss ee ee 4 Plotting Torus Data 4 0 13 Plotting images 4 0 14 Plotting spectral energy distributions 10 10 10 12 CONTENTS CONTENTS
20. aterial will be reflected from the boundary periodic material leaving through one boundary e g x will re appear at the opposite boundary e g x shock freeOutNoln material is allowed to flow freely off the grid but none is allowed to flow onto the grid constDenNoVel a constant density zero velocity boundary AKA a brick wall inflow material streams onto the grid from the inflow boundary inflowGrad Apply a gradient to the material streaming onto the grid an example use is as follows 25 5 3 HYDRODYNAMICS CHAPTER 5 THE PHYSICS MODULES xplusboundstring periodic In order to use the inflow boundary condition with a gradient inflowGrad you will need to specify along which axis there is a gradient e g in 3D is the gradient along the x boundary in the y or z directions Do this by setting one of xslope yslope zslope to T or F in your parameters file At present the gradients are geometry specific and you cannot customise them in your parameters file This will be updated before too long 5 3 4 Artificial Viscosity By default artificial viscosity is turned on You can turn it off by including useviscosity F in your parameters file And to provide a custom numerical value for the artificial viscosity default is 3 d0 use etaviscosity 9 d0 Where 9 40 is whatever value you wish to use 5 3 5 Saving time and space in large calculations Torus dumps vtu and
21. by one or more radiation sources normally stars These sources provide what is essentially a boundary condition to the radiation field Sources are described by a position radius effective temperature mass and a spectral energy distribution The number of radiation sources in the defined by the nsource integer keyword which must be specified in the input deck Each source is then defined by a set of keywords appended with the source number i e radius1 refers to the radius of source number 1 radius2 is the radius of the second source etc The source position needs to be specified The sourcepos keyword expects three double precision numbers corresponding to the source s x y z coordinate remember the code distance units are 1010 If the source is placed on the edge or corner of a grid then all photon packets emerge in the grid space and are appropriately weighted The source mass is specified by mass and the units are solar masses Similar the source radii are specified using the radius keyword with units of the solar radius The effective temperature of the source is given in Kelvins with teff The SED of the source can be defined in a number of ways by using the contflux keyword Setting contflux blackbody gives a Planck function SED Setting contflux somefilename dat will read the SED from somefilename dat The SED file should contain two columns the first of which should be wavelength or frequency and the second flux Wavelengths shoul
22. c ttauri magnetospheric accretion model cmfgen Uses input opacities from Hillier s cmfgen code cluster Uses a cluster from Bate s SPH models as input theGalaxy Uses Dobb s SPH galaxy simulation as input molcluster Uses one of Bate s SPH simulations as input wrl04 Spiral WR nebula model clumpyagb A clumpy AGB wind starburst A starburst simulation runaway Uses VH1 hydro simulation as input protobin Protostellar binary formation model a la Bate et al 1995 Specifying a geometry usually defines a set of other keywords such as the disc scaleheight the microtur bulence etc etc than must also be assigned values Specific cases are dealt with later in the manual 3 0 10 Physics There are several different microphysical processes that may be considered Some of these can be considered simultaneously e g dust and photoionization or dust and molecules Some processes are mutually exclusive for example molecular physics and atomic physics Setting any of these keywords to true will then mean that a further subset of keywords must be specified which are detailed later Physics Description dustphysics This will include dust emission absorption processes molecularphysics This will include molecular opacities atomicphysics Atomic physics currently limited to H He photoionphysics Photoionization physics H He C N O S Si Ne etc 3 0 11 Type of calculati
23. c wind A Blandford Payne style outflow whose direction and launch annulus can be altered e A dusty accretion disc This is a standard alpha disc with a curved Isella and Natta style inner rim Each component can be present absent from the geometry Radiative equilibrium can be solved for the dusty disc while statistical equilibrium for pure hydrogen can be solved for the wind and magnetosphere Subsequently data cubes for particular line transitions can be calculated along with emission line spectra The continuum emission from the inner rim can be included as a boundary condition for the radiation field 31 CHAPTER 8 GEOMETRIES 8 0 4 Magnetosphere parameters Parameter Units Description ttaurimag Set to T to include a T Tauri magnetosphere mdotparl solar masses per year The accretion rate ttaurirstar solar radii The radius of the protostar must be same as radius1 ttaurimstar solar masses The mass of the protostar must be the same as 1 ttauririnner stellar radii The inner radius of the magnetosphere at the midplane ttaurirouter stellar radii The outer radius of the magnetosphere at the midplane ttauridiskheight stellar radii mdottype constant constant accretion rate isotherm Set to T to use an isothermal temperature structure for the magnetosphere isothermtemp Kelvin Isothermal temperature of the magnetosphere usehartmanntemp Set to T to use a Hartmann
24. chapters that deal with the individual physical models that can be computed 8 CHAPTER 3 RUNNING TORUS 3 0 8 Setting up the AMR grid There is a choice of AMR dimensionality and coordinate system This is done by setting one and only one of amrid or amr2d or amr3d to T The standard 3 d cartesian grid is specified by amr3d T grid is three dimensional and gives a grid of 2x2x2 cubes each of which can be further sub divided leading to an AMR structure stored as an octree Should your problem contain a high degree of rotational symmetry then you may wish to employ the 3 d cylindrical grid specified as follows amr3d T grid is three dimensional cylindrical T use cylindrical coordinates In the cylindrical system x is radius z is height and the third dimension phi is azimuthal angle Each cell in the grid can have either four children the cell is split in x and z but not phi or eight children cell is split in x z and azimuthal extent Should your model be axisymmetric such as a disc the grid should be specified as 2d using amr2d T grid is 2 4 cylindrical which gives a quad tree grid where x is the radius and z is height Note that in the hydrodynamical case only the 2d geometry is a cartesian surface One dimensional grids are specified by amrid T grid is one dimensional This provides an AMR bi tree where the x coordinate becomes radius This is suitable for
25. contains a Makefile which can be used to build the code for a number of different systems System specific options e g compiler flags are specified in the Makefile so you should not need to set these up unless you are using a new system The environment variable SYSTEM is used to tell the Makefile which configuration to use For example to use the gfortran configuration suitable for the gfortran compiler on Mac OS X and Linux e g in csh type setenv SYSTEM gfortran Alternatively you can set this environment variable in your cshrc file The this table describes the supported SYSTEM types relating to specific compilers SYSTEM Description Can use mpi yes option gfortran The GNU gfortran compiler gcc gnu org fortran Yes can also use ompiosx system ifort The intel compiler Yes g95 The g95 compiler www g95 org No use ompi system nagfor The Nagware fortran compiler No The this table describes the currently supported SYSTEM types relating to specific machines 3 CHAPTER 2 BUILDING TORUS SYSTEM Description complexity The Leicester Dirac system AKA complexity cluster zen intel MPI compiler on the SGI Altix system at Exeter zensingle single processor compilation on SGI Altix system zenmpt Use SGI s MPI Toolkit on zen You can now go into the torus directory and build the code cd torus make depends make Several options can be passed to the make proce
26. culation The logical parameter discardsinks tells Torus to not store sink star particles 34 CHAPTER 9 USING TORUS WITH SPH DATA 9 2 SPH TORUS Parameter Default value Description sphdatafilename None Name of file containing SPH data inputFileFormat ascii Format of SPH file ascii or binary sphNormLimit 0 5 Normalise by sum of weights above this threshold kerneltype 0 0 is exponential 1 is spline useHull false Use hull particle method hcritPercentile 0 80 Percentile for hcrit hmaxPercentile 0 99 Percentile for hmax discardsinks false Sink particles are not stored Complete Responsible Author David Acreman 13 Feb 2009 Contributors David Acreman Last significantly modified by David Acreman 13 Feb 2009 Not yet reviewed 9 2 SPH Torus SPH Torus is a combination of Torus with Matthew Bate s SPH code The benchmarking and methods paper applied the code to modelling a circumstellar disc Complete Responsible Author David Acreman 13 Feb 2009 Contributors David Acreman Last significantly modified by David Acreman 13 Feb 2009 Not yet reviewed 35 CHAPTER 10 TORUS INPUT PARAMETERS All of the non geometry specific parameters TABLE sort on tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 666666 databg 4FAFOD4 F3DFA8 headerrows 2 Torus parameters
27. d be in angstroms and fluxes should be in either erg s cm ensstrem for wavelength space or erg s cm P7 for frequency space files The code works out itself whether the file is wavelength or frequency space and will warn you if it has to perform a conversion The flux file should be space delimited and should not contain tabs The final option for the contflux keyword is contflux kurucz which will interpolate in the Kurucz LTE model atmosphere grid using the given teff and the mass and radius to get log g In some modes each source will be defined by a surface across which the temperature and therefore surface brightness can vary This is useful to simulate accretion hot spots for example Parameter Units Description nsource The number of radiation sources sourcepos 10 10cm x y z coordinate of the source mass solar masses The source mass radius solar radii The source radius teff K The source effective temeprature contflux SED distribution either blackbody kurucz or a filename 28 CHAPTER 6 DEFINING RADIATION SOURCES Complete Responsible Author Tim Harries 06 Jan 2011 Contributors Tim Harries Last significantly modified by Tim Harries 06 Jan 2011 Not yet reviewed 29 CHAPTER 7 ADAPTIVE MESH REFINEMENT Complete 1 Responsible Author Contributors Last significantly modified by Not yet reviewed 30 CHAPTER 8 GEOMETRIES 8 0
28. d using temperature gradient refineonionization false N A Refine AMR grid using ionization fraction gradient dophotorefine false N A Refine AMR grid between photoionization iterations amrtolerance 1 d 3 N A Maximum gradient before refinement amrtemperaturetol 1 d 3 Maximum temperature gradient before refinement amrspeedtol 1 d 3 Maximum speed gradient before refinement amrionfractol 1 d 3 Max ionization fraction gradient before refinement amrrhoetol 1 d 3 Maximum temperature gradient before refinement masstol 1 d 5 solar masses g mass Maximum allowed mass contained in one cell 36 CHAPTER 10 TORUS INPUT PARAMETERS TABLE on tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 4666666 databg ZZFAFODA F3DFA8 headerrows 2 timeunit 1 d0 N A Code unit of time lengthunit 1 d0 N A Code unit of length massunit 1 d0 N A Code unit of mass TABLE tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 666666 databg 4FAFOD4 F3DFA8 headerrows 2 hydrodynamics false N A Perform a hydrodynamics calculation cfl 0 3 N A Courant Friedrichs Lewy constant limitertype superbee N A Flux limiter to use rhieChow true N A Use Rhie Chow interpolation fluxinterp false N A Do fine to coarse flux interpolation incomplete usevi
29. developed by Tim Harries and co workers Originally developed at St Andrews and UCL the code developers are now primarily based at the University of Exeter in the sunny southwest of the UK near the English Riviera The basic philosophy of TORUS is one of flexibility The code has a basic infrastructure that includes the AMR mesh scheme first developed by Neil Symington that is used by several physics modules including atomic line transfer in a moving medium developed by Tim Harries Ryuichi Kurosawa and Neil Symington molecular line transfer Dave Rundle and Tim Harries photoionization Tim Harries radiation hydrodynamics Tim Harries Dave Acreman Tom Haworth and radiative equilibrium Tim Harries Dave Acreman Ryuichi Kurosawa The code has been used to tackle a wide variety of problems from magnetospheric accretion onto T Tauri stars spiral nebulae around Wolf Rayet stars discs around Herbig AeBe stars structured winds of O supergiants and Raman scattered line formation in symbiotic binaries and dust emission and molecular line formation in star forming clusters As of today 6th January 2011 the code consists of approximately 145000 lines of Fortran 2003 The code itself is written to fairly tight coding guidelines and should be fairly readable although it is admittedly sparsely commented we use self explanatory variable and subroutine names as much as possible The code also makes extensive use of fortran modules and is compil
30. e inclination ninc 4 the number of inclination inclinations 0 90 the range of inclinations in degrees Alternatively you can specify the first and last inclinations again separated in uniform steps of cosine inclination ninc 4 the number of inclinations firstinc O the first inclination lastinc 90 the last inclination If you want even spacing in inclination angle rather than cos inclination set sedcosspaced F in the parameters file The output filenames are then inclination dependent with incxx appended to filename where xx is the inclination of that model in degrees Ancillary files are also output stellar direct is the spectrum produced 16 CHAPTER 5 THE PHYSICS MODULES photons that arrive at the observer directly from the stellar photons The stellar_scattered spectrum arises solely from stellar photons that are scattered one or more times towards the observer thermal direct is the spectrum of photons that are emitted by the disc and arrive without scattering at the observer thermal scattered are photons produced by dust that have been scattered one or more times before arriving at the photosphere For 3D geometries the SED will depend on a second angle in addition to the inclination A position angle can be set using the firstPA and lastPA parameters which results in evenly sampled angles between firstPA and lastPA over ninc steps To scan through position angles whilst keepi
31. ed using a standard Gnu makefile The code is parallelized using both MPI and OMP and can uses these parallel sections either separately or in a hybrid mode Most physics modules have quite low communication overhead TORUS has been compiled on a wide variety of system architectures We regularly run the code at Exeter using ifort and compile the code nightly using g95 ifort gfortran and nagfor A nightly test suite developed by Dave A is run from the CVS repository ensuring that the current CVS version at least compiles and runs basic benchmarks for radiative equilibrium molecular statistical equilibrium and hydrodynamics a Sod shock tube test We rely very little on external libraries If you want to have your output files as FITS which we recommend then you need to have the cfitsio library installed and in your linking path when compiling You need MPI libraries we use the intel one or openMPI here if you want to use the MPI parallelization and you need an OMP directive compatible fortran compiler to access the OMP or hybrid modes Visualization of the torus models can be achieved via any suitable 3rd party program that can read the VTK format files that TORUS produces we prefer visit here but paraview can be good too If you are reading this then you have been given access to TORUS please respect our intellectual property and do not distribute the code We are of course really happy for groups to use the code and are excited to see re
32. fundamentally the resolution of the spectrum is limited by the resolution at which the grain refractive indices were originally computed The SEDs can be output for a variety of units To produce a spectrum set spectrum T in the input file along with filename which is the name for your SED file Setting sed T will produce a lambda vs lambda F_lambda file The wavelength range of the SED will run from sedlammin to sedlammax with sednumlam points Note that the wavelength spacing is normally logarithmic sedwavlin T will produce linearly spaced wavelength points The SED production is Monte Carlo in nature this allows for anistropic scattering and polarization to be computed so one can specify the number of photon packets used nphotons this number of packets is divided up evenly amongst the number of wavelength bins so the higher your resolution the more packets you need for the same signal to noise The observer distance is set using distance xx where xx is the distance from the grid origin to the observer in units of parsecs For non spherical geometries the SED will be inclination dependent and torus can produce a series of SEDs for different inclinations The number of inclinations is specified using ninc and the individual inclinations can then be specified as an array ninc 4 the number of inclination inclinations 0 30 60 90 the different inclinations in degrees or as a range of values separated in uniform steps of cosin
33. g new quantities to inputs mod you are therefore required to specify a unit type this is the third quantity passed in a call to getUnitDouble for example call getUnitDouble keyword sourceTeff i temperature cLine fLine nLines amp Source temperature K a f8 0 a 1 d0 ok true where the unit type is temperature Other unit types are e distance wavelength dust angle mass time temperature luminosity More may be required Complete Responsible Author Thomas Haworth 18 Aug 2011 Contributors Thomas Haworth Last significantly modified by Thomas Haworth 18 Aug 2011 Not yet reviewed 41 CHAPTER 11 TORUS PUBLICATIONS 10 11 12 13 14 15 16 Haworth Thomas Harries J Acreman David M Rundle David A 2013 Assessing molec ular line diagnostics of triggered star formation using synthetic observations 2013 Accepted for publication in MNRAS arXiv 1303 1322 Haworth Thomas J Harries Tim J Acreman David M 2012 Testing diagnostics of triggered star formation MNRAS 426 203 Acreman David M Dobbs Clare L Brunt Christopher M Douglas Kevin A The structure of HI in galactic disks Simulations vs observations MNRAS 422 241 Haworth Thomas J Harries Tim J 2011 Radiation hydrodynamics of triggered star formation the effect of the diffuse radiation field MNRAS 420 562 Harries Tim J 2011 An algorithm fo
34. he grid Firstly the grid can be split if the mass within a cell exceeds a specified threshold the threshold is set by the parameter limitscalar Secondly a cell can be split if the density contrast within the cell exceeds a given threshold specified by limitscalar2 Thirdly it is possible to take into account the velocity difference between particles in a cell and split that cell if a threshold is exceeded A logical variable called doVelocitySplit can be read in from the parameter file to control whether this splitting condition is applied The default is true but will be set to false if velocities from SPH particles are not available e g in the SPH Torus configuration Cells on the AMR grid are populated with values using an SPH interpolation sum The interpolation kernel can be selected kerneltype and it is possible to control when interpolated values are normalised by the sum of the kernel weights either with sphNormLimit or useHull Normalisation is required to reduce noise at points between SPH particles but should not be done at points significantly outside the particle distribution To speed up the calculation of the interpolated values not all the SPH particles are used The parameters hcritPercentile and hmaxPercentile control the selection of which particles are included in the interpolation sum Sink and star particles present in the SPH dump can be read in and stored with the possibility of using them as sources in a subsequent cal
35. ht 125 disc scaleheight at 100 AU in AU rho 8 16136e 18 density at inner edge midplane g cc Dust grain properties iso_scatter T Assume isotropic scattering assumed by benchmark graintypel sil dl Drain and Lee silicates aminl 0 12 minimum grain size microns amaxl 0 1201 maximum grain size microns qdistl 0 01 power law index flat Output SEDs spectrum produce a spectrum SED parameters number of inclinations the first inclination degrees the last inclination degrees the number of photon packets in each SED the root of the output filename Write spectrum as lambda vs F lambda in SI units Distance to observer minimum wavelength in SED file maximum wavelength in SED file Linear spacing in SED file ninc 2 firstinc 12 5 lastinc 77 5 nphotons 500000 filename test sised T distance 2 25558e 8 sedlammin 0 12 sedlammax 2000 sedwavlin F 3 0 7 Top level keywords The basic architecture of the parameters file follows three stages The first is to setup up the AMR mesh the second is to perform some physics radiative equilibrium hydrodynamics etc and the third is to produce some outputs dust continuum images molecular line datacubes Halpha spectra etc Here we describe the most important keywords for setting up the model Naturally each choice of grid setup physics and outputs leads to a further set of parameter keywords that must be specified which are detailed in the
36. ied by Tim Harries 07 May 2010 Not yet reviewed 11 CHAPTER 4 PLOTTING TORUS DATA TORUS used to produce graphical output using internal PGPLOT calls but these calls have now been removed in order to increase TORUS portability 4 0 13 Plotting images VTK files Output of grid data for plotting may now be made using calls to vtk subroutines which produce VTK format files VTK files produced by Torus on the fly include filename Description lucy vtk Overwritten with each Lucy iteration contains crossings deltat dust1 etacont etaline fixedtemp mesh qualit bias vtk Contains chiline mesh quality and temperature beforesmooth vtk aftersmooth vtk Contains inflow mesh quality rho temperature and velocity magnitude rho vtk Contains inflow mesh quality rho temperature velocity magnitude VTK files can be plotted with the VisltVisualization Tool You may also like to try paraview FITS images See the Dust Continuum Models section subsection Calculating Images for more information on how to produce image files The SAO DS9 GUI can be used to plot FITS files To view spectral line profiles the program kvis handles data cubes output as described in the atomic line transfer calculations section effectively The following is an example of the fits output of an atomic line transfer model run e KvisScreenShot1 Clicking View will open this window e KvisScreenShot2
37. ilicates grainfracl 0 01 1 1 100 dust to gas mass ratio 1 0 1 minimum grain size microns amaxl 1 0 maximum grain size microns qdistl 3 5 power law index a 3 5 The following table gives details of the different grain chemistries currently coded graintype Description draine sil Draine 2003 silicate grains Alternatively you can provide your own grain optical constants in a text file This file must have three space delimited columns Wavelength microns Real part of n imaginary part of n Comments starting in the first column with a hash symbol are ignored You must also specify the density of the grain material in grams per cc graintypel mygrains dat graindensityl 3 5 density of grain material in grams cc grainfracl 0 01 1 100 dust to gas mass ratio 1 0 1 minimum grain size microns amaxl 1 0 maximum grain size microns qdistl 3 5 power law index a 3 5 14 CHAPTER 5 THE PHYSICS MODULES 5 0 17 Setting up the initial conditions for the Lucy iterations There are several keywords associated with the numerics of the Lucy algorithm that can be tweaked but for which the default values will usually give satisfactory results The number of photon packets is by default set to be 10 times the number of grid cells in order to provide adequate sampling of the radiation field in each cell but this can be adjusted using the nlucy keyword Solving
38. ine in the parameters file But this doesn t contain any comments to remind the user what this is for so it may be better to write nphotons 1000 The number of photons used to produce the SED The above is more user friendly We can also put comments on their own lines but make sure they start with a ora eg The following parameters describe the image nphotons 1000 The number of photons used to produce the SED imagesize 100 The linear image size in AU npixels 100 Produce a 100x100 pixel image noscat dk calculate image without scattered light The values associated with keywords may be integer logical T or F or floating point Normally only one value is associated with each keyword although a limited number require an array of values separated by whitespace e g Sourcepositioni 0 12 7 first source position in units of 1010 cm Only a subset of keywords must be specified at run time Others will take on a default value which is normally sensible If a keyword is not in the parameters file but it is required for the model the code will exit and tell you that a particular keyword value pair must be specified TORUS will stop with a fatal error if a particular keyword appears more than once in an parameters file It will also report keywords that were unused on input To check your parameters file run TORUS with the word check afterwards This causes TORUS to read your parameters file then exit so that you can
39. ion field and material dynamics are not independent The algorithm that torus uses performs an initial radiative transfer calculation to give the starting state of the grid It then performs sequential hydrodynamics and radiative transfer steps To run a radiation hydrodynamics model the following needs to be included in your parameters file radiationhydro T hydrodynamics T splitovermpi T photoionphysics T Radiation hydrodynamics models are domain decomposed and thus require the same number of processors as specified in Running photoionization models It is also possible to scale up the number of processors used by running a hybrid MPI openMP model This requires one node for each of the processors that would normally be required For example a 3D job using 9 processors will now require 9 nodes and a 3D model using 65 processors will now require 65 nodes In hybrid mode the initial processors use one thread on the node and the rest contribute to the openMP part of the calculation To run in hybrid mode just compile with openMP yes you will also need to modify your jobfile contact your system administrator for help with this Complete Responsible Author Thomas Haworth 09 Jun 2011 Contributors Thomas Haworth Last significantly modified by Thomas Haworth 09 Jun 2011 Not yet reviewed 27 CHAPTER 6 DEFINING RADIATION SOURCES 6 0 1 Defining radiation sources The torus grid may be illuminated
40. ltiple images a number can be appended to specify offsets for individual images e g imagecentrex3 nimage 2 17 CHAPTER 5 THE PHYSICS MODULES imagefilel onemicron fits lambdaimagel 10000 npixelsl 256 inclinationl 60 imagefile2 twomicron fits lambdaimage2 20000 npixels2 256 inclination2 60 5 0 22 Output files As well as the images and SEDs the radiative equilibrium calculation produces some diagnostic files e tune dat This file contains timing information on various stages of the calculation and may help you judge how long a particular model is going to take to run and may help with code system optimization e convergence lucy dat This file stores details of each iteration of the Lucy algorithm The columns are the iteration number the mean maximum and minimum changes in the temperature in the grid at the end of the that iteration the percentage of bad cells those that are deemed to have an undersampled radiation field while not being in the diffusion regime the dust emissivity in cgs units and the same in units of the source luminosity the maximum fractional change in temperature across all cells and finally the number of photon packets used in that iteration Remember the code is looking for a less than 196 change in the dust emissivity between iterations for convergence albedo dat The dust albedo computed for the dust mixture The column description is given in the file header info g
41. mum linear size of the grid npixels defines the number of pixels along one side of the image i e the image has npixels squared pixels in total The size of the images is controlled by imagesize but can be overridden for one or more images for example imagesize3 changes the size of the third image By default images are square but this can be changed using the imageaspect keyword The y axis is of size imagesize with npixels pixels but the x axis size and number of pixels will be scaled by imageaspect For a single image the observer s viewing angle is set by inclination in degrees and positionangle once again in degrees If the inclination and position angle are zero then the observer will be looking down the z axis of the grid The inclination angle moves the viewing vector in the x z plane and the position angle rotates the viewing angle about the z axis Hence inclination corresponds to the polar angle in a spherical co ordinate system and position angle corresponds to the azimuthal angle When generating multiple images the inclination and positionangle keywords can be used to specify values which apply to all images but they can be overridden e g inclination3 45 0 sets the inclination of the third image to 45 degrees The centre of the image will be the origin of the Torus co ordinate system unless you specify otherwise imagecentrex and imagecentrey can be used to move the centre of the image units are Torus units i e 1019 cm For mu
42. nces of metals are set to those used in the Lexington benchmark relative to Hydrogen Species Default abundance relative to Hydrogen H 1 He 0 1 C 22 e 5 N 4 e 5 33 e 5 Ne 5 e 5 5 0 9e 5 To customize these abundances modify the above figures using the following names in your parameters file lt verbatim gt h abund he_abund c_abund n_abund o_abund ne_abund s_abund lt verbatim gt 5 1 9 Retrieving data from grid files grid files from domain decomposed photoionization radiation hydrodynamics models can be used to dump vtk vtu files by including justdump T 23 5 2 CREATING SYNTHETIC IMAGES CHAPTER 5 THE PHYSICS MODULES in the parameters file When executed the grid will be read in and immediately dump a vtk vtu file for that grid upon completion of the dump torus will abort Complete Responsible Author Thomas Haworth 05 Apr 2011 Contributors Thomas Haworth Last significantly modified by Thomas Haworth 05 Apr 2011 Not yet reviewed 5 2 Creating Synthetic Images See the dust continuum model page for how to set up generic information about images e g size inclination For images generated by the photoionisation routines there is an additional imagetype option which specifies the type of image to be generated imagetype options are e freefree e forbidden e recombination e dustonly Complete Responsible Autho
43. ng the inclination constant just set firstinc equalto lastinc Non zero position angles will result in PAxx being appended to filename where xx is the position angle of that model in degrees Alternatively you can use the posangs parameter to specify a list of angles or a range in the same way that the inclinations parameter is used If you are specifying both inclinations and position angles then this should be done using the same method e g if you use the inclinations parameter then the position angles should be specified using posangs 5 0 21 Calculating images Images are calculated using the same code that produces the SEDs First one must set image T in the input file Torus can produce a sequence of images In order to do this set nimage xx where xx is the number of images and append each keyword with the appropriate image number The observer distance is set using distance xx where xx is the distance between the observer and the grid origin in parsecs The name of the image file is set via the imagefile keyword and the wavelength of the monochromatic image is set by lambdaimage where the units of wavelength is angstroms The nphotons keyword allows one to control the signal to noise of the image The axis units of the fits images are set using imageaxisunits xx where xx can be au pc cm or arcsec the default is au imagesize sets the size of one side of the square image in the same units as the image axes the default size is the maxi
44. ngs 0 0 Position angles sed false Write spectrum as lambda vs lambda F_lambda sised false Write spectrum as lambda microns vs lambda F_lambda microns W m 2 jansky false Write spectrum in janskies Same parameter name nphotons used when generating an image but a different default value TABLE sort on tableborder 0 cellpadding 1 cellspacing 3 headerbg D5CCB1 header color 666666 databg FAFOD4 F3DFA8 headerrows 2 dustphysics false Include dust physics in the calculation dusttogas 0 01 Dust to gas ratio ndusttype 1 Number of different dust types grainTypeLabel sil dl Dust grain type grainFracLabel 1 Grain fractional abundance 0 005 aminLabel microns Min grain size amaxLabel 0 25 microns Max grain size qDistLabel 3 5 Grain power law a0Label 1 0e20 microns Scale length of grain size pdistLabel 1 Exponent for exponential cut off iso_scatter false Isotropic scattering Complete L_ Responsible Author Thomas Haworth 09 Jun 2011 Contributors Tim Harries Thomas Haworth Last significantly modified by Tim Harries 04 May 2010 Not yet reviewed 10 1 Unit Directory It is now possible to specify the units for certain variables in parameters dat This is still in the early stages more quantities are being updated to include this all the time and the quantities for which units can be specified should
45. ngstrom and weighting them accordingly to restore the correct result This can now be automatically applied to any spectrum that you use by specifying biasToLyman T in the parameters file By default this VR is turned off If you want to customise the extent to which the bias takes place simply use the following in the parameters file biasMagnitude 100 d0 and vary 100 40 to whatever you want it to be Note that biasMagnitude 1 40 is the same as running the non VR model Caution should always be exercised when using VR techniques If you want to check that your photons are still reproducing the correct spectrum use binPhotons T in the parameters file This sums the number of photon packets multiplied by their weight and divided by the total number of photon packets and dumps them to bins dat By running this procedure for biasMagnitude 1 d0 and then again with e g biasMagnitude 100 40 you can check that you are reproducing the correct spectrum by comparing the two output files You should expect the non ionizing part of the spectrum to be much noisier than the ionizing part For those who use photoionAMR mod i e domain decomposed models splitovermpi T in the log file you will also be given the ratio of the source luminosity to the sum of the energy per photon packet times its weight near the start of every iteration If everything is working correctly this should be close to 1 For fur
46. on Once the geometry has been specified and the different microphysical processes that need to be considered have been detailed the user must select the type of calculation they wish to perform e Statistical equilibrium stateq T Conduct a statistical equilibrium calculation by solving the rate equations If you have molecularphysics T then an algorithm described in Rundle et al 2010 is used If atomicphysics T then a comoving frame algorithm is employed e Radiative equilibrium radeq T This requires that dustphysics T and uses the Monte Carlo radia tive equilibrium method of Lucy 1999 e Photoionization equilibrium photoion eq Solves photoionization equilibrium including thermal equi librium using a method similar to that of Ercolano xxxx Dust may be included e Hydrodynamics hydrodynamics T This solves hydrodynamics using a TVD scheme and superbee flux limiter 3 0 12 Checking your parameters file Once you have written a parameters file you can get Torus to check it for you Run the Torus executable with the word check afterwards optionally followed by the name of your parameters file if it is not called parameters dat For example you can use the command torus openmp check mypar dat to check a parameters file called mypar dat using a torus executable called torus openmp 10 CHAPTER 3 RUNNING TORUS Complete Responsible Author Tim Harries 07 May 2010 Contributors Tim Harries Last significantly modif
47. r Thomas Haworth 10 Jun 2011 Contributors Thomas Haworth Last significantly modified by Thomas Haworth 10 Jun 2011 Not yet reviewed 5 3 Hydrodynamics TORUS performs hydrodynamics calculations using a Eulerian grid based algorithm It is flux conserving uses Rhie Chow interpolation and is total variation diminishing TVD How To Include Hydrodynamics To perform a Hydrodynamics calculation simply include the line hydrodynamics T in your parameters dat file Because it only works using multiple threads in a domain decomposition as outlined in Running photoionization models you also need to include the following splitovermpi T 5 3 1 Flux limiting A flux limiter prevents the formation of oscillations near shocks whilst attempting to retain physical oscillations in piecewise linear schemes Use of a flux limiter is turned on by default and there are a number available for use 24 CHAPTER 5 THE PHYSICS MODULES 5 3 HYDRODYNAMICS Limiter type Notes superbee This is the default flux limiter minmod MC fromm vanleer donorcell This is equivalent to turning the flux limiter off To select a specific flux limiter include the following in your parameters file limitertype minmod and replace minmod with whichever of the limiters you want in the table above Generally there should never be any need to use anything other than the superbee however
48. r Monte Carlo time dependent radiation transfer Mayne Nathan J Harries Tim J 2010 On the properties of discs around accreting brown dwarfs MNRAS 409 1307 Rundle David Harries Tim J Acreman David M Bate Matthew R Three dimensional molecular line transfer a simulated star forming region MNRAS 407 986 Acreman David M Harries Tim J Rundle David A Modelling circumstellar discs with three dimensional radiation hydrodynamics 2010 MNRAS 403 1143 Douglas Kevin A Acreman David M Dobbs Clare L Brunt Christopher M A Synthetic 21 cm Galactic Plane Survey of a smoothed particle hydrodynamics Galaxy Simulation 2010 MNRAS 407 405 Acreman David M Douglas Kevin A Dobbs Clare L Brunt Christopher M Synthetic HI observations of a simulated spiral galaxy 2010 MNRAS 406 1460 Pinte C Harries T J Min M Watson A M Dullemond C P Woitke P Mnard F Durn Rojas M C Benchmark problems for continuum radiative transfer High optical depths anisotropic scattering and polarisation 2009 A amp A 498 967 Tannirkulam A Monnier J D Harries T J Millan Gabet R Zhu Z Pedretti E Ireland M Tuthill P ten Brummelaar T McAlister H Farrington C Goldfinger P J Sturmann J Sturmann L Turner N A Tale of Two Herbig Ae Stars MWC 275 and AB Aurigae Comprehensive Models for Spectral Energy Distribution and Interferometry 2008 ApJ 689 513 Kurosawa Ryuichi
49. radiative equilibrium satisfactorily requires adequate spatial resolution and this is particularly im portant for the transition between optically thin and optically thick regions such as the surface of a disc or the inner radius of a dust shell If smoothgridtau is set to true in the input file an iterative sweep is performed over the grid comparing the optical depth across a given cell at a given wavelength with those of its neighbours If one of the cells has an optical depth of greater than taumax and its neighbour has an optical depth of less than taumin then the more optically thick cell is split The sweep is performed at wavelengths between lambdasmooth and the maximum wavelength under consideration normally 2 millimetres The dosmoothgrid parameter may also be set This invokes a smooth that ensures that the size ratio of neighbouring cells does not exceed smoothfactor by default 3 Large differences in cell AMR depths for adjacent cells can lead to numerical problems so it is a good idea to have dosmoothgrid T in your input file 5 0 18 On convergence Lucy s method is an iterative one that will gradually converge towards radiative equilibrium Since the method is a Monte Carlo one the temperatures of cells will have noise associated with them so convergence must be defined in a statistical sense It should be remembered that the more photon packets that are run the better the temperature estimate will be and the more likely it is
50. rid dat Details about the size depth number of cells etc in the AMR grid lucy vtk Visualization data for the grid output at the end of each iteration lucy_grid_tmp dat The grid is dumped after each iteration in case the code crashes the computer goes down or there s a nuclear attack etc Complete Responsible Author Tim Harries 07 May 2010 Contributors Tim Harries Last significantly modified by Tim Harries 07 May 2010 Not yet reviewed 5 0 23 Molecular Line Transfer Calculations Setting molecularphysics T in the parameters file will allow the user to perform co moving frame non LTE molecular line calculations in Torus There are several keywords that need to be defined moleculefile defines the name of the molecule to for which level populations are to be found and or the expected emission to be calculated for e g co mol If the user wishes to calculate statistical equilibrium level populations for the molecule using the algorithm set out in Rundle 2010 by setting stateq T with molecularphysics T then it is recommended that the parameters file for TORUS includes sensible values for at least the following parameters 18 CHAPTER 5 THE PHYSICS MODULES Parameter Type Comment molAbundance real Relative abundance of molecule H_2 Overridden by doChemistry vturb real Constant non thermal turbulent velocity tolerance real RMS Convergence level for level populations or standard
51. rid fes 2 2 s ss do a more mr d RE OE Ro Oe RO 23 5 2 Creating Synthetic Images oo e be Ee ok ba RARE ebook Rob dh ge red 24 53 Hydrodynamics s i s amp x E Re oomen as Loeb ee te e Die ee ES 24 534 osse Ee RO ee ee BUR ROB ROS OR OR ed a 24 5 3 2 Rhie Chow Interpolation 84 ee dO te aaa RO RP Re G9 E 5 25 5 3 3 Boundary Conditions sss kem x RA AUREUS ROAD SOR UR 25 534 Jtifieal Viscosity lt ea uo ooo Room o e ko xoxo kom RR ek ea p dem 26 5 3 5 Saving time and space in large calculations 26 54 Radiation Hydrodynamics pa mos xong ko kom emm Rok a RR Exe Red 27 CONTENTS CONTENTS 6 Defining radiation sources 6 0 1 7 Adaptive Mesh Refinement 8 Geometries 9 Using Torus with SPH data 9 1 Generating Torus grids from SPH data 9 2 SPH Torus 10 Torus input parameters 10 1 Unit Directory 10 1 1 Available units 10 1 2 Unit Types for developers 11 Torus Publications Defining radiation sources Protoplanetary disc geometry ttauri geometry Magnetosphere parameters Disc wind parameters Dusty disc parameters PREFACE Exeter is home to the 3 dimensional radiative transfer code TORUS which is either an acronym for Transport of Radiation Under Sobolev or Transport of Radiation Using Stokes CHAPTER 1 INTRODUCTION TORUS is a radiation transfer and radiation hydrodynamics code
52. s disctemp Disc temperature inside sublimation radius height AU The disc scale height at 100 AU mdisc solar masses The disc mass alphadisc The density goes as rho alpha betadisc The scaleheight goes as h beta epsilon ratio of Planck mean opacity at stellar and inner disc temperatures ttauriwarp Set to T to include warped disc around magnetosphere hoverr Warped disc H R 32 CHAPTER 8 GEOMETRIES Complete Responsible Author Tim Harries 11 Jan 2011 Contributors Tim Harries Last significantly modified by Tim Harries 11 Jan 2011 Not yet reviewed 33 CHAPTER 9 USING TORUS WITH SPH DATA 9 1 Generating Torus grids from SPH data The Torus AMR grid can be set up from density temperature and velocity fields derived from SPH particles This can either be from an SPH dump file which is read in by Torus or can be passed through the subroutine interface in TorusMod SPH dump files can be either binary files from the SPH NG code or ascii files generated by splash to ascii The geometry sphfile will set up a grid without velocities stored at octal corners whereas the geometry theGalaxy will set up a grid with velocities stored at octal corners e g for line calculations The grid is constructed in two stages Firstly the grid is split to achieve the required resolution then the cells are populated with values calculated from the SPH particles There are three main conditions used to split t
53. scosity true N A Use artificial viscosity xplusboundstring null Positive x direction boundary condition xminusboundstring null Negative x direction boundary condition yplusboundstring null Positive y direction boundary condition yminusboundstring null Negative y direction boundary condition zplusboundstring null Positive z direction boundary condition zminusboundstring null Negative z direction boundary condition tstart 0 d0 s time Start time in simulation tend 1 d10 s time End time in simulation tdump 0 d0 s time Time between data dumps in simulation TABLE tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 4666666 databg ZZFAFODA F3DFA8 headerrows 2 nsource 1 Number of sources stars pointsource false Source object is pointlike radiusj 1 40 Solar radii Radius of jth source teffj 1 d0 K temperature Effective temperature of jth source massj 1 d0 Solar masses Mass of jth source contfluxj none Continuum flux file for jth source sourceposj 0 d0 0 d0 0 d0 10 10cm Position of jth source velocityj 0 d0 0 d0 0 d0 km s velocity of jth source probsourcej 0 40 Probability of photon packet from jth source TABLE tableborder 0 cellpadding 1 cellspacing 3 headerbg 4D5CCB1 header color 666666 databg FAFOD4 F3DFA8 headerrows
54. spherically symmetric models the molecular benchmark is an example of this type of mesh Note that in the hydrodynamics case only the amrld mode has x as a linear coordinate not a radius The size of the AMR mesh is give by amrgridsize which refers to the full extent of the root cell The distance is given in units of 101 cm which are the general code units for distance The centre of the AMR mesh is assumed to be at 0 0 0 for the three d cases and r amrgridsize 2 z 0 for the 2 d cylindrical mode The cell depth in TORUS is measured from 1 for the root cell to N where amrgridsize 2 is the dimension of the smallest cell in the AMR mesh The grid is always split to a depth specified by mindepthamr default 5 and the maximum depth is set using maxdepthamr which defaults to 31 Note that the memory footprint of the AMR mesh is strongly dependent on these two parameters and these should be adjusted with caution and maxdepthamr should certainly not be revised upwards only downwards 3 0 9 Geometry The geometry keyword specifies the type of model you wish to run be it a protostellar disc a stellar wind a Bonnor Ebert sphere etc etc The following non exhaustive list provides the most commonly used geometries in TORUS 9 CHAPTER 3 RUNNING TORUS Geometry Description molbench The molecular benchmark test of xxxx benchmark The Pascucci 2004 disc benchmark shakara An alpha dis
55. ss Option Description Default Notes cfitsio Link with cfitsio libraries yes debug Switch on debugging flags no profile Enable profiling no static Perform static linking no Not available on Mac OS X trace Use Intel trace collector no Intel compilers only openmp Build with OpenMP parallelism no coverage Enable coverage analysis with gcov no gnu compilers only hydro Include hydrodynamics yes photoion Include photoionisation yes molecular Include molecular physics yes sph Include code for making grids from SPH yes atomic Include atomic physics yes hdf Build with hdf5 support required for reading flash dumps no For example to build torus without linking to the cfitsio libraries and with debug flags enabled use make cfitsio no debug yes Building without the photoion or sph options will result in a slightly smaller memory footprint for the grid there are statically allocated octal components which are excluded You may need to remove all files from a previous build and start from scratch for example if you have previously compiled with debug flags and need to recompile without debug flags The Makefile has an option to clean up files from the previous build make clean Building Torus on Mac OS X To link with the FITS library under OS X make sure that libcfitsio a is in the default search path e g usr local lib or usr lib 2 0 4 Testing your build
56. style temperature structure for the magnetosphere maxharttemp Kelvin Maximum of Hartmann temperature dipoleoffset degrees Dipole offset thotspot Kelvin Hot spot temperature Ixoverlbol Bolometric Luminosities X ray luminosity maxcellmass grams Maximum cell mass after splitting mdottype Default is constant enhance Set to T to include accretion enhancement mdotparN solar masses per year N ranges from 1 6 for constant accretion only set mdotpar1 8 0 5 Disc wind parameters Parameter Units Description ttauriwind Set to T to add a disc wind DW Rmin ttaurirouter Minimum radius for launch annulus DW Rmax DW Rmin Maximum radius for launch annulus DW theta degrees Launch angle for wind 90 is perpendicular to disc DW Twind Kelvin Isothermal temperature of the wind DW Mdot solar masses per year Wind mass loss rate DW alpha Exponent in the mass loss rate per unit area DW beta Exponent in beta velocity law DW gamma Exponent in the disc wind temperature law DW Rs 10 10cm Constant in beta velocity law usually 50xRmin DW f Scaling of asymptotic terminal velocity usually 2 0 8 0 6 Dusty disc parameters Parameter Units Description ttauridisc Set to T to add a dusty alpha disc rinner stellar radii The inner disc radius router AU The outer disc radius rsub stellar radii Sublimation radiu
57. sults from it appearing in the refereed literature Finally the TORUS code is the result of the intellectual effort of a small group of dedicated astronomers most of whom are at a fairly early career stage It is therefore extremely important that they get credit for their hard work Tim Harries 6th January 2011 CHAPTER 1 INTRODUCTION May ask what you expected to see out of a Torquay hotel bedroom window Sydney Opera House perhaps The Hanging Gardens of Babylon Herds of wildebeest sweeping majestically Basil Fawlty Complete Responsible Author David Acreman 13 Feb 2009 Contributors David Acreman Last significantly modified by David Acreman 13 Feb 2009 Not yet reviewed CHAPTER 2 BUILDING TORUS 2 0 1 Getting the Torus code The Torus code is managed using SVN To download the code from the repository type where USERNAME is your username on mail astro ex ac uk 2 0 2 Compilers Torus is known to build using the freely available gfortran compiler and the commercial Intel Fortran com piler These compilers are routinely used to run Torus so you can expect them to run typical configurations successfully The NAG Fortran compiler is also used to build Torus so it should work but is not so frequently tested g95 has been removed from the Torus test suite as the compiler is no longer being developed and building with gfortran gives better performance 2 0 3 Building Torus The Torus repository
58. t Photon packets are communicated between threads in stacks to reduce the communication overhead By default a single stack size is used for all threads this can be specified using stacklimit xx It is also possible to use a different stacklimit for each MPI thread To do this include customstacks T numMPlthreads xx stacklimit 200 20 CHAPTER 5 THE PHYSICS MODULES 5 1 PHOTOIONIZATION MODELS stacklimarrayO 100 stacklimarray21 50 stacklimarray33 50 where numMPlthreads is the number of mpi threads stacklimit acts as the default for unspecified threads and stacklimarray followed by the thread number gives the specified value for that thread Any questions or problems email haworth astro ex ac uk An Important Note The domain decomposed photoionization routine now uses buffered sending between threads This requires the specification of a buffer size that is difficult to predetermine and cannot be modified within a single iteration If the buffer size is exceeded then your job will crash and give the following in your stderr_torus file when not in debug mode cli 1 aborting job Fatal error in MPI_Bsend Invalid buffer pointer error stack MPI Bsend 184 MPI_Bsend buf 0x647c680 count 200 dtype USER lt struct gt dest 2 tag 41 MPILCOMM WORLD failed MPIR Bsend isend 338 Insufficient space in Bsend buffer requested 100
59. t atoms will be used simultaneously Subsequently each of the atomic data files need to be defined natom 2 atoml H atm atom2 Hel atm Note that only the hydrogen atom is currently fully implemented Complete Responsible Author Tim Harries 24 Aug 2010 Contributors Tim Harries Last significantly modified by Tim Harries 24 Aug 2010 Not yet reviewed 19 Parameter Type Comment Range de dongstep logical Perform Ng acceleration steps at specified intervals T F T plotlevels logical Write VTK file of diagnostic data level populations in plots T F F outputconvergence logical Output files containing global relative change in level populations for each iteration T F F quasi logical Use Sobol quasi random number generator to determine ray direction and frequency T F F usedust logical Includes continuum emission absorption from dust T F F densitysubsample logical Interpolate density where possible in order to make smoother image T F F 5 1 PHOTOIONIZATION MODELS CHAPTER 5 THE PHYSICS MODULES 5 1 Photoionization Models 5 1 1 Domain Decomposed Photoionization Models Domain decomposed photoionization models are run using multiple processors threads The computa tional domain is split into equal sized subdomains each of which is handled by an individual thread An additional thread is also required to perform governing operations At present models can be run using one of the following
60. that cells with propagate into optically thick regions However for some models such as a canonical T Tauri star disc the optical depths to the midplane are such that the chance of an individual photon packet penetrating to the midplane are vanishingly small and even very large numbers of photon packets will leave such cells undersampled Undersampled in this context means that the number of photons passing through a given cell is smaller than mincrossings so the algorithm cannot be used to adequately estimate the mean intensity for that cell At the end of a particular iteration there may be some volumes within the grid that contain many such undersampled cells Torus employs the diffusion approximation in these regions using the surround cells that do have well sampled radiation fields as a boundary condition The use of the diffusion approximation is controlled by a number of keywords that are initialised using sensible defaults Cells with a Rosseland optical depth taudiff are always part of the diffusion calculation Cells that are diffdepth from the effective surface of the structure e g the disc are always in the diffusion zone The iterative scheme proceeds as follows The photon packets are run through the grid and the new temperatures are calculated either from the MC estimators or the diffusion approximation The number of undersampled cells is found these are cells that have been sampled by fewer than mincrossings photon packets and
61. ther information or if it not working properly please contact Tom Haworth haworth astro ex ac uk 22 CHAPTER 5 THE PHYSICS MODULES 5 1 PHOTOIONIZATION MODELS 5 1 7 Periodic Boundary Conditions domain decomposed models only In some cases you might want periodic boundary conditions That is photons that leave the grid at one side re enter the grid at the opposite side An example of a case where this would be useful is to provide better signal to noise for cells near the edge that are otherwise not symmetrically sampled To avoid infinite looping of low energy photons packets are only allowed to cross a periodic boundary once To include periodic boundary conditions which are turned off by default simply include the following in your parameters file periodicX T periodicY T periodicZ T for the x y and z boundaries respectively Note that this only works for domain decomposed photoionization models splitovermpi T 5 1 8 Customizing Atomic Species You can specify what species you want present in a model as well as their relative abundances At present you have to include Hydrogen if this is all that you want then include hOnly T usemetals F in the parameters file To include just Hydrogen and Helium you need the following usemetals F hOnly is defaulted as False Finally to include metals as well simply to nothing as they are included by default Default abunda
62. yet have a Rosseland optical depth of less than taudiff If the percentage of undersampled cells is greater than lucy_undersampled then the number of photon packets is doubled for the next iteration The convergence criterion considers the emissivity of the dust If the percentage change of the emissivity is less than one percent compared to the previous iteration then the Lucy algorithm finishes The file lucy_convergence dat provides detailed information on an iteration by iteration basis of the convergence of the radiative equilibrium 5 0 19 Common pitfalls Here we list some common problems 15 CHAPTER 5 THE PHYSICS MODULES e If mincrossings is set too small then the MC estimator for the radiation field and therefore the temperature etc for some cells can be very noisy and although the percentage of undersampled cells is small the error on the dust emissivity will be large and the convergence criterion will never be triggered e Dust sublimation If your inner disc radius is too small then the dust temperature at the inner disc can exceed the dust sublimation temperature In order to ensure you have a consistent inner disc temperature you should set vardustsub T in your model while will invoke the routines to systematically produce the correct inner disc shape 5 0 20 Calculating SEDs Once the radiative equilibrium is complete one can produce spectra across an arbitrary wavelength region at an arbitrary resolution although

Torus User Manual - University of Exeter

Contents

Download Pdf Manuals

Related Search

Related Contents