From latife at astro.rug.nl Tue Jan 1 09:33:28 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Tue, 01 Jan 2008 16:33:28 +0100 Subject: [FLASH-USERS] problem in running FLASH2.5 for Pancake test problem Message-ID: <477A5D48.5050408@astro.rug.nl> Hi Dear all, Happy New Year to all, I am new user of FLASH2.5. I am trying to run FLASH for Zeldovich Pancake Test Problem(as given in setups directory) but it is not working. it gives the following error. i write this command in home directory. ./setup pancake -auto SUCCESS then while i write "make" it gives following error fortcom: Severe: GravPotentialAllBlocks.F90, line 35: **Internal compiler error: segmentation violation signal raised** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error. contains ^ compilation aborted for GravPotentialAllBlocks.F90 (code 3) make[1]: *** [GravPotentialAllBlocks.o] Error 3 make[1]: Leaving directory `/.auto make: *** [default] Error 2 ............................................................................... i have tried this {./setup pancake -2d -maxblocks=150 -auto} but it is still not working while it is working for all other problems except "orbit" test problem for orbit problem it gives following error. ./setup orbit -3d -maxblocks=150 -auto sucess "make" it gives following error fortcom: Severe: ReDistributeParticles.F90, line 12: **Internal compiler error: segmentation violation signal raised** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error. contains ^ compilation aborted for ReDistributeParticles.F90 (code 3) make[1]: *** [ReDistributeParticles.o] Error 3 make[1]: Leaving directory `/.automount/perenago/root/data/users/latife/FLASH2.5/object' make: *** [default] Error 2 ............................................ i think problem is only wheni use particle module bcz both of these problems are using particle module(particle test problems). i will be very thankfull to you for your kind suggestion in this regard. If some body can help me. My 2nd question is how to solve any problem for spherical Geometry using FLASH 2.5. Thanks in advance. kind regards M.A.Latife Kapteyn Astronomical Institute http://www.astro.rug.nl From cdaley at flash.uchicago.edu Thu Jan 3 15:35:43 2008 From: cdaley at flash.uchicago.edu (Chris Daley) Date: Thu, 03 Jan 2008 15:35:43 -0600 Subject: [FLASH-USERS] problem in running FLASH2.5 for Pancake test problem In-Reply-To: <477A5D48.5050408@astro.rug.nl> References: <477A5D48.5050408@astro.rug.nl> Message-ID: <477D552F.7060307@flash.uchicago.edu> Hi M.A. Latife, It would seem that the compilation failure is happening because of a bug in the version of the Intel compiler you are using. I compiled the Pancake problem for FLASH 2.5 using an Intel compiler on one of our machines. The compiler version I used was: Intel(R) Fortran Compiler for Intel(R) EM64T-based applications, Version 9.1 Build 20060707 Package ID: l_fc_c_9.1.036 The files compiled, but there was a warning message about the length of the procedure names in the 2 source files that are causing you problems. This warning message only appeared for ReDistributeParticles.F90 and GravPotentialAllBlocks.F90. fortcom: Warning: GravPotentialAllBlocks.F90, line 0: Global name too long, shortened from: modulegravpotentialallblocks_mp_gravpotentialallblocks_$BLK.modulemapparticlestomesh_mp_mpi_bottom_ to: ulegravpotentialallblocks_mp_gravpotentialallblocks_$BLK.modulemapparticlestomesh_mp_mpi_bottom_ It may be more than coincidental that your compilation problems are happening for the only 2 source files in which there are long (>31 characters) procedure names. I can offer the following suggestions: * Report the compilation failure to Intel. * Attempt to use a different compiler / different Intel compiler version. * Manually edit the source code so that the concatenated module and procedure names are always less than 31 characters in the 2 source files. Finally, you can run spherical simulations in FLASH2.5 by specifying the following option in the flash.par file: geometry = "spherical" Hope this of some help, Regards, Chris M.A. Latife wrote: > Hi > Dear all, > Happy New Year to all, > I am new user of FLASH2.5. I am trying to run FLASH for Zeldovich > Pancake Test Problem(as given in setups directory) but it is not > working. it gives the following error. > i write this command in home directory. > ./setup pancake -auto > SUCCESS > then while i write "make" > it gives following error > > fortcom: Severe: GravPotentialAllBlocks.F90, line 35: **Internal > compiler error: segmentation violation signal raised** Please report > this error along with the circumstances in which it occurred in a > Software Problem Report. Note: File and line given may not be > explicit cause of this error. > contains > ^ > compilation aborted for GravPotentialAllBlocks.F90 (code 3) > make[1]: *** [GravPotentialAllBlocks.o] Error 3 > make[1]: Leaving directory `/.auto > make: *** [default] Error 2 > ............................................................................... > > i have tried this {./setup pancake -2d -maxblocks=150 -auto} but it is > still not working > > while it is working for all other problems except "orbit" test problem > for orbit problem it gives following error. > ./setup orbit -3d -maxblocks=150 -auto > sucess > "make" > it gives following error > fortcom: Severe: ReDistributeParticles.F90, line 12: **Internal > compiler error: segmentation violation signal raised** Please report > this error along with the circumstances in which it occurred in a > Software Problem Report. Note: File and line given may not be > explicit cause of this error. > contains > ^ > compilation aborted for ReDistributeParticles.F90 (code 3) > make[1]: *** [ReDistributeParticles.o] Error 3 > make[1]: Leaving directory > `/.automount/perenago/root/data/users/latife/FLASH2.5/object' > make: *** [default] Error 2 > ............................................ > i think problem is only wheni use particle module bcz both of these > problems are using particle module(particle test problems). > i will be very thankfull to you for your kind suggestion in this regard. > If some body can help me. > > My 2nd question is how to solve any problem for spherical Geometry > using FLASH 2.5. > > > Thanks in advance. > kind regards > M.A.Latife > Kapteyn Astronomical Institute > http://www.astro.rug.nl From woitke at strw.leidenuniv.nl Tue Jan 8 05:47:45 2008 From: woitke at strw.leidenuniv.nl (Peter Woitke) Date: Tue, 8 Jan 2008 12:47:45 +0100 (MET) Subject: [FLASH-USERS] 2D disk models Message-ID: Hi, I have a very basic question. I would like to setup a FLASH model in cylinder coordinates (r,z,phi) assuming 2D axisymmetry (d./dphi=0) with rotation v_phi<>0. My problem (or maybe just a misunderstanding) is that if I specify 2D geometry = "cylindrical", there seems to be no rotation in the model (v_phi=0?) and I cannot find/specity the v_phi variable. If you write down the hydro-equations in cylinder coordinates, the problem becomes very apparent: The equation of motion (r-direction) contains the centrifugal term -(rho v_phi^2)/r. In order to keep track of v_phi, however, one needs to solve the equation of motion (phi-direction), too, i.e. the angular momentum conservation. So what I want is a solution of the full 3D hydro-equtions in cylinder coordinates just under the condition of axisymmetry d./dphi=0. How do I do this with FLASH? Thank you, Peter Woitke From bernalcg at astroscu.unam.mx Tue Jan 8 16:42:24 2008 From: bernalcg at astroscu.unam.mx (Giovanny Bernal) Date: Tue, 8 Jan 2008 15:42:24 -0700 Subject: [FLASH-USERS] 2D disk models In-Reply-To: References: Message-ID: Hi, I have a similar problem... how I put rotation (v_phi <>0) in cylindrical and polar geometries in flash2.5?? In init_block? in hydro_sweep? thanks El 08/01/2008, a las 04:47 AM, Peter Woitke escribi?: > Hi, > > I have a very basic question. I would like to setup a FLASH model > in cylinder coordinates (r,z,phi) assuming 2D axisymmetry (d./dphi=0) > with rotation v_phi<>0. > > My problem (or maybe just a misunderstanding) is that if I specify 2D > geometry = "cylindrical", there seems to be no rotation in the model > (v_phi=0?) and I cannot find/specity the v_phi variable. If you > write down > the hydro-equations in cylinder coordinates, the problem becomes very > apparent: The equation of motion (r-direction) contains the > centrifugal > term -(rho v_phi^2)/r. In order to keep track of v_phi, however, > one needs > to solve the equation of motion (phi-direction), too, i.e. the angular > momentum conservation. > > So what I want is a solution of the full 3D hydro-equtions in cylinder > coordinates just under the condition of axisymmetry d./dphi=0. > > How do I do this with FLASH? > > Thank you, > > Peter Woitke > From tomek at scs.fsu.edu Tue Jan 8 16:24:41 2008 From: tomek at scs.fsu.edu (Tomasz Plewa) Date: Tue, 8 Jan 2008 17:24:41 -0500 Subject: [FLASH-USERS] 2D disk models In-Reply-To: References: Message-ID: <20080108172441.jwv39qkls8k0cko0@webmail.scs.fsu.edu> Peter, Giovanny - The simplest recipe is to advect the gas specific angular momentum. This is quite similar to advecting mass scalars (individual species) and you can use one of the mass scalars for that purpose. (I hope mass scalars are supported in the official 2.5 version.) Once you know actual distribution of specific angular momentum, centrifugal force can be easily calculated. Hope this helps. Cheers - Tomek -- School of Computational Science, DSL 443 ph: 850.644.1959 Florida State University fax: 850.644.0098 Tallahassee, FL 32306 web: people.scs.fsu.edu/~tomek/ -- Giovanny Bernal wrote: > Hi, > I have a similar problem... how I put rotation (v_phi <>0) in > cylindrical and polar geometries in flash2.5?? In init_block? in > hydro_sweep? > thanks > > El 08/01/2008, a las 04:47 AM, Peter Woitke escribi?: > >> Hi, >> >> I have a very basic question. I would like to setup a FLASH model >> in cylinder coordinates (r,z,phi) assuming 2D axisymmetry (d./dphi=0) >> with rotation v_phi<>0. >> >> My problem (or maybe just a misunderstanding) is that if I specify 2D >> geometry = "cylindrical", there seems to be no rotation in the model >> (v_phi=0?) and I cannot find/specity the v_phi variable. If you write down >> the hydro-equations in cylinder coordinates, the problem becomes very >> apparent: The equation of motion (r-direction) contains the centrifugal >> term -(rho v_phi^2)/r. In order to keep track of v_phi, however, one needs >> to solve the equation of motion (phi-direction), too, i.e. the angular >> momentum conservation. >> >> So what I want is a solution of the full 3D hydro-equtions in cylinder >> coordinates just under the condition of axisymmetry d./dphi=0. >> >> How do I do this with FLASH? >> >> Thank you, >> >> Peter Woitke >> > > > -- > This message has been scanned for viruses and > dangerous content by MailScanner, and is > believed to be clean. ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. From kruijssen at astro.uu.nl Thu Jan 10 09:25:20 2008 From: kruijssen at astro.uu.nl (Diederik Kruijssen) Date: Thu, 10 Jan 2008 16:25:20 +0100 Subject: [FLASH-USERS] bug output at test setup Sedov References: <4C4E8350-EC7A-43E0-B33A-B9788871192E@astro.uu.nl> Message-ID: <718EFB18-6D35-4011-8BD4-19FDA36F7FFF@astro.uu.nl> Hi, I am experiencing difficulties very early on when installing Flash3.0beta on my Mac PowerBook G4, with MPICH2v1.0.5 and HDF5v1.6.6. When setting up for the Sedov blast wave test problem, it gives the following output: =============================== Processing Shortcut file: /Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup_shortcuts.txt checking for needed files and directories A setup internal error has occured, if possible please email the following debugging info to flash-bugs at flash.uchicago.edu Arguments: ['/Users/Diederik/Documents/work/FLASH3.0_beta/bin/ setup.py', 'Sedov', '-auto'] Python Version: 2.3.5 Platform Details: darwin Traceback (most recent call last): File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup.py", line 232, in ? main() File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup.py", line 49, in main machDir = determineMachine() File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/utils.py", line 154, in determineMachine hostname = getHostName(siteDir) File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/utils.py", line 76, in getHostName temp = socket.gethostbyaddr(socket.gethostname()) socket.herror: (1, 'Unknown host') ============================== Of this, only the first two lines are as they should be. It appears setup can't find the hostname, but if this is true, how should it be fixed? It requests to send an email to flash-bugs, but since only developers are members of that list, I'm trying it this way. Thank you very much for your help in advance, Sincerely, Diederik Kruijssen From nhearn at uchicago.edu Thu Jan 10 09:41:58 2008 From: nhearn at uchicago.edu (Nathan Hearn) Date: Thu, 10 Jan 2008 09:41:58 -0600 Subject: [FLASH-USERS] bug output at test setup Sedov In-Reply-To: <718EFB18-6D35-4011-8BD4-19FDA36F7FFF@astro.uu.nl> References: <4C4E8350-EC7A-43E0-B33A-B9788871192E@astro.uu.nl> <718EFB18-6D35-4011-8BD4-19FDA36F7FFF@astro.uu.nl> Message-ID: <2467fdc0801100741k6707d7c7ta0ea737dd6a53255@mail.gmail.com> Hi Diederik, Mac OS X tends to use the name under the Networks preferences (typically a name followed by ".local"), rather than the DNS entry for the machine. The easiest way to correct this problem is to specify an explicit site name when running Flash's setup script. Adding the option -site=myhostname to the setup call (where "myhostname" is the name of the appropriate directory in the FLASH3/sites directory) should work. - Nathan On Jan 10, 2008 9:25 AM, Diederik Kruijssen wrote: > Hi, > > I am experiencing difficulties very early on when installing > Flash3.0beta on my Mac PowerBook G4, with MPICH2v1.0.5 and HDF5v1.6.6. > When setting up for the Sedov blast wave test problem, it gives the > following output: > > =============================== > Processing Shortcut file: > /Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup_shortcuts.txt > checking for needed files and directories > > A setup internal error has occured, if possible please email the > following > debugging info to flash-bugs at flash.uchicago.edu > Arguments: ['/Users/Diederik/Documents/work/FLASH3.0_beta/bin/ > setup.py', 'Sedov', '-auto'] > Python Version: 2.3.5 > Platform Details: darwin > Traceback (most recent call last): > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup.py", > line 232, in ? > main() > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup.py", > line 49, in main > machDir = determineMachine() > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/utils.py", > line 154, in determineMachine > hostname = getHostName(siteDir) > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/utils.py", > line 76, in getHostName > temp = socket.gethostbyaddr(socket.gethostname()) > socket.herror: (1, 'Unknown host') > ============================== > > Of this, only the first two lines are as they should be. It appears > setup can't find the hostname, but if this is true, how should it be > fixed? > It requests to send an email to flash-bugs, but since only developers > are members of that list, I'm trying it this way. > > Thank you very much for your help in advance, > > Sincerely, > > Diederik Kruijssen From dubey at flash.uchicago.edu Thu Jan 10 09:42:12 2008 From: dubey at flash.uchicago.edu (Anshu Dubey) Date: Thu, 10 Jan 2008 09:42:12 -0600 (CST) Subject: [FLASH-USERS] bug output at test setup Sedov In-Reply-To: <718EFB18-6D35-4011-8BD4-19FDA36F7FFF@astro.uu.nl> References: <4C4E8350-EC7A-43E0-B33A-B9788871192E@astro.uu.nl> <718EFB18-6D35-4011-8BD4-19FDA36F7FFF@astro.uu.nl> Message-ID: <50584.75.3.90.212.1199979732.squirrel@flash.uchicago.edu> Dear Diederik, In flash distribution, the prototype Makefile.h for Mac platforms is under the directory sites/Prototypes/Darwin, and is designed to work with the IBM xlf compiler. Your best option is to create a specific site directory for your machine as described in User's guide, and then always specify -site="your sitedir" option in the setup command. You can use the prototype as an example of what your Makefile.h should be like. Anshu > Hi, > > I am experiencing difficulties very early on when installing > Flash3.0beta on my Mac PowerBook G4, with MPICH2v1.0.5 and HDF5v1.6.6. > When setting up for the Sedov blast wave test problem, it gives the > following output: > > =============================== > Processing Shortcut file: > /Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup_shortcuts.txt > checking for needed files and directories > > A setup internal error has occured, if possible please email the > following > debugging info to flash-bugs at flash.uchicago.edu > Arguments: ['/Users/Diederik/Documents/work/FLASH3.0_beta/bin/ > setup.py', 'Sedov', '-auto'] > Python Version: 2.3.5 > Platform Details: darwin > Traceback (most recent call last): > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup.py", > line 232, in ? > main() > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/setup.py", > line 49, in main > machDir = determineMachine() > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/utils.py", > line 154, in determineMachine > hostname = getHostName(siteDir) > File "/Users/Diederik/Documents/work/FLASH3.0_beta/bin/utils.py", > line 76, in getHostName > temp = socket.gethostbyaddr(socket.gethostname()) > socket.herror: (1, 'Unknown host') > ============================== > > Of this, only the first two lines are as they should be. It appears > setup can't find the hostname, but if this is true, how should it be > fixed? > It requests to send an email to flash-bugs, but since only developers > are members of that list, I'm trying it this way. > > Thank you very much for your help in advance, > > Sincerely, > > Diederik Kruijssen > From woitke at strw.leidenuniv.nl Fri Jan 11 03:42:52 2008 From: woitke at strw.leidenuniv.nl (Peter Woitke) Date: Fri, 11 Jan 2008 10:42:52 +0100 (MET) Subject: [FLASH-USERS] 2D disk models Message-ID: Dear Tomasz, > The simplest recipe is to advect the gas specific angular momentum. This > is quite similar to advecting mass scalars (individual species) and you > can use one of the mass scalars for that purpose. (I hope mass scalars > are supported in the official 2.5 version.) Once you know actual > distribution of specific angular momentum, centrifugal force can be > easily calculated. thank you very much for this comment! In fact, in (r,z,phi) cylinder coordinates with axisymmetry (d./dphi=0), the phi-component of the equation of motion is equivalent to D/Dt(rho L3) = 0 (1) where D/Dt means the Lagrangian derivative, r=sqrt(x^2+y^2), and L3 = r vphi, (2) i.e. the angular momentum density rho*L3 in an invariant of the flow. So I just have to define and initialize the mass scalar L3 and FLASH will solve Eq.(1) automatically for me. The z-comonent of the equation of motion is already correct, even with phi-rotation. Only the r-component needs a modification, namely the additional source term +rho vphi^2/r (the centrigugal force) on the r.h.s. I will implement this by a modified gravity module which uses L3 as: gr = -GM/(r^2+z^2)^(3/2)*r + L3/r^3 (3) gz = -GM/(r^2+z^2)^(3/2)*z (4) >From my experience with radiation pressure as additional force, I can confirm that mass scalars work properly with Flash 2.4. A modified gravity module might be more accurate, since it is called several times during the hydro-integration, than an additional force term which would be treated by operator splitting, wouldn't it? Thank you, Peter Woitke ============================================ Dr. Peter Woitke -------------------------------------------- School of Physics & Astronomy University of St Andrews North Haugh St Andrews KY16 9SS Scotland UK -------------------------------------------- Room 329a Tel.: (+44) 1334 463068 Fax.: (+44) 1334 463104 URL.: http://www.strw.leidenuniv.nl/~woitke ============================================ From woitke at strw.leidenuniv.nl Fri Jan 11 07:22:09 2008 From: woitke at strw.leidenuniv.nl (Peter Woitke) Date: Fri, 11 Jan 2008 14:22:09 +0100 (MET) Subject: [FLASH-USERS] 2D disk models Message-ID: Hi once more, one remaining question is about the kinetic energy and the equation of state. The specific internal energy should be e = e_int + 1/2 (vr^2 + vz^2 + vphi^2), So I seems neccessary to include the vphi^2-term "everywhere". But where is "everywhere" in the FLASH-code? Thank you, Peter Woitke From tomek at scs.fsu.edu Fri Jan 11 07:23:49 2008 From: tomek at scs.fsu.edu (Tomasz Plewa) Date: Fri, 11 Jan 2008 08:23:49 -0500 Subject: [FLASH-USERS] 2D disk models In-Reply-To: References: Message-ID: <20080111082349.2tltfwh04ko40cs8@webmail.scs.fsu.edu> Peter - Keeping all forces together is a good first approximation. It might be done somewhat better if more force-specific information is available than for gravity. Tomek -- School of Computational Science, DSL 443 ph: 850.644.1959 Florida State University fax: 850.644.0098 Tallahassee, FL 32306 web: people.scs.fsu.edu/~tomek/ -- Peter Woitke wrote: > Dear Tomasz, > >> The simplest recipe is to advect the gas specific angular momentum. This >> is quite similar to advecting mass scalars (individual species) and you >> can use one of the mass scalars for that purpose. (I hope mass scalars >> are supported in the official 2.5 version.) Once you know actual >> distribution of specific angular momentum, centrifugal force can be >> easily calculated. > > thank you very much for this comment! In fact, in (r,z,phi) cylinder > coordinates with axisymmetry (d./dphi=0), the phi-component of the > equation of motion is equivalent to > > D/Dt(rho L3) = 0 (1) > > where D/Dt means the Lagrangian derivative, r=sqrt(x^2+y^2), and > > L3 = r vphi, (2) > > i.e. the angular momentum density rho*L3 in an invariant of the flow. So I > just have to define and initialize the mass scalar L3 and FLASH will solve > Eq.(1) automatically for me. The z-comonent of the equation of motion is > already correct, even with phi-rotation. Only the r-component needs a > modification, namely the additional source term +rho vphi^2/r (the > centrigugal force) on the r.h.s. > > I will implement this by a modified gravity module which uses L3 as: > > gr = -GM/(r^2+z^2)^(3/2)*r + L3/r^3 (3) > gz = -GM/(r^2+z^2)^(3/2)*z (4) > >> From my experience with radiation pressure as additional force, > I can confirm that mass scalars work properly with Flash 2.4. > A modified gravity module might be more accurate, since it is > called several times during the hydro-integration, than an additional > force term which would be treated by operator splitting, wouldn't it? > > Thank you, > > Peter Woitke > > > ============================================ > Dr. Peter Woitke > -------------------------------------------- > School of Physics & Astronomy > University of St Andrews > North Haugh > St Andrews KY16 9SS > Scotland > UK > -------------------------------------------- > Room 329a > Tel.: (+44) 1334 463068 > Fax.: (+44) 1334 463104 > URL.: http://www.strw.leidenuniv.nl/~woitke > ============================================ > > -- > This message has been scanned for viruses and > dangerous content by MailScanner, and is > believed to be clean. ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. From tomek at scs.fsu.edu Fri Jan 11 07:35:08 2008 From: tomek at scs.fsu.edu (Tomasz Plewa) Date: Fri, 11 Jan 2008 08:35:08 -0500 Subject: [FLASH-USERS] 2D disk models In-Reply-To: References: Message-ID: <20080111083508.h64mfa1bkocw00wk@webmail.scs.fsu.edu> Peter - That is correct and it would be the specific total energy, of course. Extra contribution should be automatically accounted for in update_soln (provided the third velocity component is not eliminated given the problem is 2D and not 3D). Cheers - Tomek -- School of Computational Science, DSL 443 ph: 850.644.1959 Florida State University fax: 850.644.0098 Tallahassee, FL 32306 web: people.scs.fsu.edu/~tomek/ -- Peter Woitke wrote: > Hi once more, > > one remaining question is about the kinetic energy and the equation > of state. The specific internal energy should be > > e = e_int + 1/2 (vr^2 + vz^2 + vphi^2), > > So I seems neccessary to include the vphi^2-term "everywhere". > But where is "everywhere" in the FLASH-code? > > Thank you, > > Peter Woitke > > > -- > This message has been scanned for viruses and > dangerous content by MailScanner, and is > believed to be clean. ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. From latife at astro.rug.nl Mon Jan 14 10:34:57 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Mon, 14 Jan 2008 17:34:57 +0100 Subject: [FLASH-USERS] Problem in solving jenas instability for spherical geometry using FLASH2.5 Message-ID: <478B8F31.1080309@astro.rug.nl> Hi Dear all, Hope you will be fine and enjoying good health. I have very basic questions. 1)i am trying to solve the jeans instability problem for spherical geometryusing FLASH 2.5. when i run this problem for one dimensional spherical geometry it gives the following error conditons for this problem are given below ./setup testsph -1d -auto result success make result success cd object ./flash2 WARNING: desired timestep < dtmin, using dtmin 24 1.4911E-01 1.0000E-08 | 0.000E+00 abort_flash called. See log file for details. Calling MPI_Abort() for immediate shutdown! [0] MPI Abort by user Aborting program ! [0] Aborting program! p0_18386: p4_error: : 1 in log file there is following error abort_message: FATAL: find_center_of_mass: Mtot can any body tell me why its is happening? 2) My second question is when i run it for two dimensional spherical goemetry. For this case when i choose r=1.5 cm there is more density along the boundaries of circle (2-d spherical geometry. when i choose r100 cm there is high density on centre in form of peak but difference between maximum density and minimum density is small(like max den=4.3*10^7 and min 9.9*10^6, while inital density is 1.5*10^7, for cartesian coordinates max den=1.1*10^8, min 7.5*10^5, while inital density is 1.5*10^7). Intial conditions for problem are given below. Can any one help me in this regard? I am not good user of FLASH, trying to learn the FLASH. For two dimensional sperical geometry i am expcting centre collapse p0 rho0 lambdax lambday lambdaz amplitude Computational domain xmin xmax ymin =-1.0 ymax zmin zmax geometry xl_boundary_type xr_boundary_type yl_boundary_type yr_boundary_type zl_boundary_type zr_boundary_type nblockx nblocky dtini dtmin dtmax Gravity igrav grav_boundary_type mpole_lmax rho=rho0(1+amplitude(cos(krcos(theta)+krsin(theta))) p=p0(1+gamma*amplitude(cos(krcos(theta)+krsin(theta))) kj=sqrt(4*pi*G*rho0)/c0 c0=sqrt(gamma*p0/rho0) I will be very thankful to you for ur kind help in this regard thanks in advance cheers M.A.Latife -------------- next part -------------- An HTML attachment was scrubbed... URL: http://flash.uchicago.edu/pipermail/flash-users/attachments/20080114/4e64bcb8/attachment.html From zuhone at flash.uchicago.edu Mon Jan 14 14:56:27 2008 From: zuhone at flash.uchicago.edu (John ZuHone) Date: Mon, 14 Jan 2008 14:56:27 -0600 Subject: [FLASH-USERS] Problem in solving jenas instability for spherical geometry using FLASH2.5 In-Reply-To: <478B8F31.1080309@astro.rug.nl> References: <478B8F31.1080309@astro.rug.nl> Message-ID: Hello M.A., With regards to your first question, are you using the jeans setup in the FLASH 2.5 source tree? There doesn't seem to be a setup called "testsph" in the setups directory. Also, it would be good to know whether you are using the multigrid or the multipole solver for gravity. Both the timestep being zero and the fact that the gravity solver complaining about Mtot = 0 suggest that something has gone wrong with the initial setup of density on the grid. I would have to think more carefully about your second question. Best, John ZuHone On Jan 14, 2008, at 10:34 AM, M.A. Latife wrote: > Hi > Dear all, > Hope you will be fine and enjoying good health. > I have very basic questions. > 1)i am trying to solve the jeans instability problem for spherical > geometryusing FLASH 2.5. when i run this problem for one dimensional > spherical geometry it gives the following error > conditons for this problem are given below > ./setup testsph -1d -auto result success > make result success > cd object > ./flash2 > WARNING: desired timestep < dtmin, using dtmin > 24 1.4911E-01 1.0000E-08 | 0.000E+00 > > abort_flash called. See log file for details. > Calling MPI_Abort() for immediate shutdown! > > [0] MPI Abort by user Aborting program ! > [0] Aborting program! > p0_18386: p4_error: : 1 > in log file there is following error > abort_message: FATAL: find_center_of_mass: Mtot = 0 > can any body tell me why its is happening? > 2) My second question is when i run it for two dimensional spherical > goemetry. For this case when i choose r=1.5 cm there is more > density along the boundaries of circle (2-d spherical geometry. when > i choose r= 100 cm there is high density on centre in form of peak > but difference between maximum density and minimum density is > small(like max den=4.3*10^7 and min 9.9*10^6, while inital density > is 1.5*10^7, for cartesian coordinates max den=1.1*10^8, min > 7.5*10^5, while inital density is 1.5*10^7). Intial conditions for > problem are given below. > Can any one help me in this regard? I am not good user of FLASH, > trying to learn the FLASH. > For two dimensional sperical geometry i am expcting centre collapse > > p0 = 1.5E0 > rho0 = 1.5E7 > lambdax = 5.0 kx=2pi/lambdax > lambday = 1.E10 ky=2pi/lambday > lambdaz = 1.E10 kz=2pi/lambdaz > amplitude = 0.001 > Computational domain > > xmin = 0.0 > xmax = 1.0E2 > ymin =-1.0 > ymax = 1.0 > zmin = 0. > zmax = 0. > > geometry = "spherical" > > xl_boundary_type = "reflect" > xr_boundary_type = "periodic" > yl_boundary_type = "periodic" > yr_boundary_type = "periodic" > zl_boundary_type = "periodic" > zr_boundary_type = "periodic" > nblockx = 2 > nblocky = 2 > dtini = 1.0E-5 > dtmin = 1.0E-8 > dtmax = 1.0E-2 > Gravity > > igrav = 1 > grav_boundary_type = "isolated" > mpole_lmax = 0.0 > rho=rho0(1+amplitude(cos(krcos(theta)+krsin(theta))) > p=p0(1+gamma*amplitude(cos(krcos(theta)+krsin(theta))) > kj=sqrt(4*pi*G*rho0)/c0 > c0=sqrt(gamma*p0/rho0) > > I will be very thankful to you for ur kind help in this regard > thanks in advance > cheers > M.A.Latife > > > From Daniel.Pfenniger at obs.unige.ch Tue Jan 15 13:22:41 2008 From: Daniel.Pfenniger at obs.unige.ch (Daniel Pfenniger) Date: Tue, 15 Jan 2008 20:22:41 +0100 Subject: [FLASH-USERS] Total energy in periodic self-gravitating fluids Message-ID: <478D0801.4040403@obs.unige.ch> Hello, I am playing with the Jeans problem as given in Flash 2.5 in 3D by using a slightly lower initial pressure in order to explore a mildly unstable regime. Obviously the total energy as outputted in the file flash.dat is not the total energy, but the kinetic plus thermal energy. But adding the potential energy does not give a conserved quantity, even in the slightly inhomogeneous regime. I suspect a miscalculation of the potential energy in the periodic case because in the file io/wr_integrals.F90 the total gravitational contribution is calculated as -------------------- ! po! potential energy if (igpot >= 0) & & lsum(8) = lsum(8) + 0.5*solnData(idens,i,j,k) * & & solnData(igpot,i,j,k)*dvol -------------------- while I would expect to see the potential solnData(igpot,i,j,k) not multiplied by half the density 0.5*solnData(idens,i,j,k), but by half the density difference with respect to the average density. In the input file flash.par the reference density is given but I am not sure how to pass this quantity properly in wr_integrals.F90. Will the Jeans problem be included in FLASH 3.0? Thanks for any tips, Dan From zuhone at flash.uchicago.edu Tue Jan 15 13:50:35 2008 From: zuhone at flash.uchicago.edu (zuhone at flash.uchicago.edu) Date: Tue, 15 Jan 2008 13:50:35 -0600 (CST) Subject: [FLASH-USERS] Total energy in periodic self-gravitating fluids In-Reply-To: <478D0801.4040403@obs.unige.ch> References: <478D0801.4040403@obs.unige.ch> Message-ID: <37163.128.135.152.57.1200426635.squirrel@flash.uchicago.edu> Dan, I think that you are correct, it should be the one-half the density difference. In wr_integrals the "reference_density" parameter would be included by adding the following line to wr_integrals.F90: call get_parm_from_context('reference_density', reference_density) and by making sure you have the real variable reference_density defined somewhere. I recommend putting your own copy of wr_integrals.F90 into your setup directory and doing this. The Jeans problem will be included with FLASH 3.0. Best, John ZuHone > Hello, > > I am playing with the Jeans problem as given in Flash 2.5 in 3D by > using a slightly lower initial pressure in order to explore a > mildly unstable regime. > > Obviously the total energy as outputted in the file flash.dat > is not the total energy, but the kinetic plus thermal energy. > But adding the potential energy does not give a conserved > quantity, even in the slightly inhomogeneous regime. > > I suspect a miscalculation of the potential energy in > the periodic case because in the file io/wr_integrals.F90 > the total gravitational contribution is calculated as > -------------------- > > ! po! potential energy > if (igpot >= 0) & > & lsum(8) = lsum(8) + 0.5*solnData(idens,i,j,k) * & > & solnData(igpot,i,j,k)*dvol > > -------------------- > > while I would expect to see the potential solnData(igpot,i,j,k) > not multiplied by half the density 0.5*solnData(idens,i,j,k), > but by half the density difference with respect to the average > density. > In the input file flash.par the reference density is given > but I am not sure how to pass this quantity properly in > wr_integrals.F90. > > Will the Jeans problem be included in FLASH 3.0? > > > Thanks for any tips, > > Dan > From Daniel.Pfenniger at obs.unige.ch Thu Jan 17 05:49:30 2008 From: Daniel.Pfenniger at obs.unige.ch (Daniel Pfenniger) Date: Thu, 17 Jan 2008 12:49:30 +0100 Subject: [FLASH-USERS] Total energy in periodic self-gravitating fluids In-Reply-To: <37163.128.135.152.57.1200426635.squirrel@flash.uchicago.edu> References: <478D0801.4040403@obs.unige.ch> <37163.128.135.152.57.1200426635.squirrel@flash.uchicago.edu> Message-ID: <478F40CA.2000006@obs.unige.ch> Dear John, zuhone at flash.uchicago.edu wrote: > Dan, > > > I think that you are correct, it should be the one-half the density > difference. In wr_integrals the "reference_density" parameter would > be included by adding the following line to wr_integrals.F90: > > call get_parm_from_context('reference_density', reference_density) > > and by making sure you have the real variable reference_density defined > somewhere. I recommend putting your own copy of wr_integrals.F90 into your > setup directory and doing this. Thanks. This indeed improves the energy conservation check. Here are the differences I included in the wr_integrals.F90 file: --------------------------------------------------------------------------------------- $ diff FLASH2.5/setups/jeans/wr_integrals.F90 FLASH2.5/source/io/wr_integrals.F90 187d186 < real, save :: reference_density = 0. 212,215d210 < if (igpot >= 0) then < call get_parm_from_context('reference_density', reference_density) < endif < 276,277c271,272 < & lsum(8) = lsum(8) + 0.5*(solnData(idens,i,j,k)-reference_density) * & < & solnData(igpot,i,j,k)*dvol --- > & lsum(8) = lsum(8) + 0.5*solnData(idens,i,j,k) * & > & solnData(igpot,i,j,k)*dvol --------------------------------------------------------------------------------------- Now the indicated total energy in the Jeans test is very well conserved during the regular phases of evolution, and shows up to when the calculations may be trusted (despite lowering the Courant condition by an order of magnitude the total energy may be increased by an order of magnitude when singularities grow). I will redo the Jeans test in unstable conditions when FLASH 3.0 will become available. Best regards, Dan From brockp at umich.edu Fri Jan 18 08:56:03 2008 From: brockp at umich.edu (Brock Palen) Date: Fri, 18 Jan 2008 09:56:03 -0500 Subject: [FLASH-USERS] disable check pointing Message-ID: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> For parallel runs even if I set trstrt=0 two checkpoint files were written. [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will send 10 blocks per message. [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint name=Ni_Tem_hdf5_chk_0000 [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint name=Ni_Tem_hdf5_chk_0000 blocks=5449 [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile name=Ni_Tem_hdf5_plt_cnt_0000 [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile name=Ni_Tem_hdf5_plt_cnt_0000 [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint name=Ni_Tem_hdf5_chk_0001 [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint name=Ni_Tem_hdf5_chk_0001 blocks=5449 No other checkpoint files have been written so far, though several plot files have been written. Is this expected behavior? Flash2.5 + pgi +openmpi-1.2.3 hdf5/serial Brock Palen Center for Advanced Computing brockp at umich.edu (734)936-1985 From zuhone at flash.uchicago.edu Fri Jan 18 12:17:19 2008 From: zuhone at flash.uchicago.edu (John ZuHone) Date: Fri, 18 Jan 2008 12:17:19 -0600 Subject: [FLASH-USERS] disable check pointing In-Reply-To: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> Message-ID: Brock, I think that if you want no checkpoints at all you should set trstrt < 0, and make sure that nrstrt < 0 (integer) and zrstrt < 0 as well. Best, John ZuHone On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: > For parallel runs even if I set > > trstrt=0 > > two checkpoint files were written. > [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will > send 10 blocks per message. > [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint > name=Ni_Tem_hdf5_chk_0000 > [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint > name=Ni_Tem_hdf5_chk_0000 blocks=5449 > [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile > name=Ni_Tem_hdf5_plt_cnt_0000 > [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile > name=Ni_Tem_hdf5_plt_cnt_0000 > [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... > [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 > [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint > name=Ni_Tem_hdf5_chk_0001 > [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint > name=Ni_Tem_hdf5_chk_0001 blocks=5449 > > No other checkpoint files have been written so far, though several > plot files have been written. Is this expected behavior? > > Flash2.5 + pgi +openmpi-1.2.3 > hdf5/serial > > > Brock Palen > Center for Advanced Computing > brockp at umich.edu > (734)936-1985 > From brockp at umich.edu Fri Jan 18 12:51:06 2008 From: brockp at umich.edu (Brock Palen) Date: Fri, 18 Jan 2008 13:51:06 -0500 Subject: [FLASH-USERS] disable check pointing In-Reply-To: References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> Message-ID: <2E167B6B-820E-44F5-9862-AB2DB84D8219@umich.edu> Ok I will try setting them all to -1, The run just crashed trying to write checkpoint_002 It segfault'ed when the file size was 2.2 GB. We were using HDF5 serial, on luster. Thanks, Brock Palen Center for Advanced Computing brockp at umich.edu (734)936-1985 On Jan 18, 2008, at 1:17 PM, John ZuHone wrote: > Brock, > > I think that if you want no checkpoints at all you should set > trstrt < 0, and make sure that nrstrt < 0 (integer) and zrstrt < 0 > as well. > > Best, > > John ZuHone > > On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: > >> For parallel runs even if I set >> >> trstrt=0 >> >> two checkpoint files were written. >> [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will >> send 10 blocks per message. >> [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint >> name=Ni_Tem_hdf5_chk_0000 >> [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint >> name=Ni_Tem_hdf5_chk_0000 blocks=5449 >> [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile >> name=Ni_Tem_hdf5_plt_cnt_0000 >> [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile >> name=Ni_Tem_hdf5_plt_cnt_0000 >> [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... >> [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 >> [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint >> name=Ni_Tem_hdf5_chk_0001 >> [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint >> name=Ni_Tem_hdf5_chk_0001 blocks=5449 >> >> No other checkpoint files have been written so far, though >> several plot files have been written. Is this expected behavior? >> >> Flash2.5 + pgi +openmpi-1.2.3 >> hdf5/serial >> >> >> Brock Palen >> Center for Advanced Computing >> brockp at umich.edu >> (734)936-1985 >> > > > From miller at astro.umd.edu Fri Jan 18 13:01:03 2008 From: miller at astro.umd.edu (Cole Miller) Date: Fri, 18 Jan 2008 14:01:03 -0500 (EST) Subject: [FLASH-USERS] Setup of hydrostatic equilibrium in FLASH 3.0beta? In-Reply-To: References: Message-ID: Hi all, Has anyone set up a good hydrostatic equilibrium routine for FLASH 3.0beta? I've got something that basically works (in the context of an X-ray burst problem), but I'm concerned about its robustness and there is a mysterious factor of 0.25 that I have to put in, so I'd appreciate greater understanding. As far as I can tell, the setup breaks into two tasks: initializing the interior cells, which seems straightforward, and setting the boundary conditions at the edge, which is where potential issues are arising. 1. For the interior cells, I simply wrote a little routine in Simulation_initBlock.F90 that computes the vertical density structure in hydrostatic equilibrium on a fine mesh. Then, when densities are assigned in each block, I obtain the z coordinate at each point and do a linear interpolation to find the density. The routine for HSE assumed a user input of density at the minimum z value in the simulation (assuming gravity is in the -z direction), combined with a temperature (assumed constant throughout), then computed the vertical density structure at the next z value up using a search by bisection for a density that would yield a pressure gradient force that counteracted gravity. This involved calls to the Eos routine, e.g., eosData(EOS_TEMP) = sim_tempAmbient dzz = (1.0/(Nhse-1.0))*(sim_zmax-sim_zmin) do k = 1, Nhse zz = sim_zmin + (k-1)*dzz if (k .EQ. 1) then density(k) = sim_rhoAmbient eosData(EOS_DENS) = density(k) call Eos(MODE_DENS_TEMP,1,eosData,massFraction) pressure(k)=eosData(EOS_PRES) else dlogmin=0.0 ! Minimum ln density in cgs dlogmax=log(density(k-1)) ! Max. This is dens of lower cell do ibisec=0,30 dlogmid=0.5*(dlogmin+dlogmax) density(k)=exp(dlogmid) eosData(EOS_DENS) = density(k) call Eos(MODE_DENS_TEMP,1,eosData,massFraction) pressure(k)=eosData(EOS_PRES) !! Now we divide the pressure gradient by the average density. If the !! pressure gradient is too small, then the density in the current !! zone has to be less gradient=(pressure(k-1)-pressure(k))/dzz gradient=gradient/(0.5*(density(k)+density(k-1))) if (gradient<(-1.00*sim_gconst)) then dlogmax=dlogmid else dlogmin=dlogmid endif enddo Here Nhse is a large value (1001 in my case) to make later interpolation more accurate. sim_tempAmbient, sim_rhoAmbient, and sim_gconst are input in flash.par. This seems to work fine. 2. However, when I ran this, although the inner regions seemed in good equilibrium, the bottom part of the simulation is not; there is a lot of bouncing and thus generation of heat. Assuming this is because of the lower boundary condition, I started looking at Grid_applyBCEdge.F90. In principle, I'd like to do the same thing here as in Simulation_initBlock.F90. That is, for each of the guard cells, I want to be able to figure out the vertical coordinate and then do something similar to what I did above. For example, for the guard cells off the low end of the grid, the *upper* boundary condition would be a density of sim_rhoAmbient and a temperature of sim_tempAmbient. I ran into two problems, possibly related to each other: (a) I was not able to find a specific routine that would tell me the coordinates of the center of a guard cell. The only routine that I saw that gives cell coordinates is Grid_getCellCoords, and one can use this to get the coordinates of guard cells. This, however, requires a blockId as input. I've been using blockId=1 and it seems to work for the simple cases I've tried, but I'm worried that it will fail after refinement. Is Grid_applyBCEdge.F90 called only once, or for every block that borders an edge? Is there a separate routine that gives the coordinates of guard cells? (b) When I made this choice, I set up a routine similar to the one I reproduced above, but in Grid_applyBCEdge.F90. It gives me good equilibrium conditions, but only if I adopt a strange ad hoc factor. What I seem to need to do is that instead of if (gradient<(-1.00*sim_gconst)) then I have to use if (gradient<(-0.25*sim_gconst)) then This factor is independent of the vertical resolution for the cases I've tried (of course, the equilibrium is better for better vertical resolution). As I say, doing all this does give very nice hydrostatic equilibrium, but until I understand these issues I'm leery of using this for any serious runs. Any idea what's happening? Does anyone have a routine for FLASH 3.0beta? I see that FLASH2.5 had a hydrostatic equilibrium routine, but I didn't see anything there that addressed guard cells. Thanks! Cole From mzingale at scotty.ess.sunysb.edu Fri Jan 18 13:04:58 2008 From: mzingale at scotty.ess.sunysb.edu (Mike Zingale) Date: Fri, 18 Jan 2008 14:04:58 -0500 (EST) Subject: [FLASH-USERS] disable check pointing In-Reply-To: <2E167B6B-820E-44F5-9862-AB2DB84D8219@umich.edu> References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> <2E167B6B-820E-44F5-9862-AB2DB84D8219@umich.edu> Message-ID: you can use the io/null module to completely disable I/O (or atleast, you used to be able to). Specify this in your Modules file. On Fri, 18 Jan 2008, Brock Palen wrote: > Ok I will try setting them all to -1, > > The run just crashed trying to write checkpoint_002 > It segfault'ed when the file size was 2.2 GB. We were using HDF5 serial, on > luster. > > Thanks, > > Brock Palen > Center for Advanced Computing > brockp at umich.edu > (734)936-1985 > > > On Jan 18, 2008, at 1:17 PM, John ZuHone wrote: > >> Brock, >> >> I think that if you want no checkpoints at all you should set trstrt >> < 0, and make sure that nrstrt < 0 (integer) and zrstrt < 0 as well. >> >> Best, >> >> John ZuHone >> >> On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: >> >>> For parallel runs even if I set >>> >>> trstrt=0 >>> >>> two checkpoint files were written. >>> [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will send >>> 10 blocks per message. >>> [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint >>> name=Ni_Tem_hdf5_chk_0000 >>> [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint >>> name=Ni_Tem_hdf5_chk_0000 blocks=5449 >>> [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile >>> name=Ni_Tem_hdf5_plt_cnt_0000 >>> [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile >>> name=Ni_Tem_hdf5_plt_cnt_0000 >>> [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... >>> [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 >>> [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint >>> name=Ni_Tem_hdf5_chk_0001 >>> [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint >>> name=Ni_Tem_hdf5_chk_0001 blocks=5449 >>> >>> No other checkpoint files have been written so far, though several plot >>> files have been written. Is this expected behavior? >>> >>> Flash2.5 + pgi +openmpi-1.2.3 >>> hdf5/serial >>> >>> >>> Brock Palen >>> Center for Advanced Computing >>> brockp at umich.edu >>> (734)936-1985 >>> >> >> > ----------------------------------------------------------------------------- Michael Zingale (mzingale at mail.astro.sunysb.edu) Assistant Professor Dept. of Physics and Astronomy office: ESS 440 SUNY Stony Brook phone: 631-632-8225 Stony Brook, NY 11794-3800 web: http://www.astro.sunysb.edu/mzingale ----------------------------------------------------------------------------- From brockp at umich.edu Fri Jan 18 13:38:17 2008 From: brockp at umich.edu (Brock Palen) Date: Fri, 18 Jan 2008 14:38:17 -0500 Subject: [FLASH-USERS] disable check pointing In-Reply-To: References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> <2E167B6B-820E-44F5-9862-AB2DB84D8219@umich.edu> Message-ID: <5C2B5B7A-3874-4BA7-8147-7989A9D571DC@umich.edu> Will this kill the plot files though? Brock Palen Center for Advanced Computing brockp at umich.edu (734)936-1985 On Jan 18, 2008, at 2:04 PM, Mike Zingale wrote: > you can use the io/null module to completely disable I/O (or > atleast, you used to be able to). Specify this in your Modules file. > > > On Fri, 18 Jan 2008, Brock Palen wrote: > >> Ok I will try setting them all to -1, >> >> The run just crashed trying to write checkpoint_002 >> It segfault'ed when the file size was 2.2 GB. We were using HDF5 >> serial, on luster. >> >> Thanks, >> >> Brock Palen >> Center for Advanced Computing >> brockp at umich.edu >> (734)936-1985 >> >> >> On Jan 18, 2008, at 1:17 PM, John ZuHone wrote: >> >>> Brock, >>> >>> I think that if you want no checkpoints at all you should set >>> trstrt < 0, and make sure that nrstrt < 0 (integer) and zrstrt < >>> 0 as well. >>> Best, >>> John ZuHone >>> On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: >>>> For parallel runs even if I set >>>> trstrt=0 >>>> two checkpoint files were written. >>>> [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will >>>> send 10 blocks per message. >>>> [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint >>>> name=Ni_Tem_hdf5_chk_0000 >>>> [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint >>>> name=Ni_Tem_hdf5_chk_0000 blocks=5449 >>>> [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile >>>> name=Ni_Tem_hdf5_plt_cnt_0000 >>>> [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile >>>> name=Ni_Tem_hdf5_plt_cnt_0000 >>>> [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... >>>> [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 >>>> [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint >>>> name=Ni_Tem_hdf5_chk_0001 >>>> [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint >>>> name=Ni_Tem_hdf5_chk_0001 blocks=5449 >>>> No other checkpoint files have been written so far, though >>>> several plot files have been written. Is this expected behavior? >>>> Flash2.5 + pgi +openmpi-1.2.3 >>>> hdf5/serial >>>> Brock Palen >>>> Center for Advanced Computing >>>> brockp at umich.edu >>>> (734)936-1985 >> > > > ---------------------------------------------------------------------- > ------- > Michael Zingale (mzingale at mail.astro.sunysb.edu) > Assistant Professor > > Dept. of Physics and Astronomy office: ESS 440 > SUNY Stony Brook phone: 631-632-8225 > Stony Brook, NY 11794-3800 web: http://www.astro.sunysb.edu/ > mzingale > ---------------------------------------------------------------------- > ------- > > > From mzingale at scotty.ess.sunysb.edu Fri Jan 18 13:40:59 2008 From: mzingale at scotty.ess.sunysb.edu (Mike Zingale) Date: Fri, 18 Jan 2008 14:40:59 -0500 (EST) Subject: [FLASH-USERS] disable check pointing In-Reply-To: <5C2B5B7A-3874-4BA7-8147-7989A9D571DC@umich.edu> References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> <2E167B6B-820E-44F5-9862-AB2DB84D8219@umich.edu> <5C2B5B7A-3874-4BA7-8147-7989A9D571DC@umich.edu> Message-ID: yes. On Fri, 18 Jan 2008, Brock Palen wrote: > Will this kill the plot files though? > > Brock Palen > Center for Advanced Computing > brockp at umich.edu > (734)936-1985 > > > On Jan 18, 2008, at 2:04 PM, Mike Zingale wrote: > >> you can use the io/null module to completely disable I/O (or atleast, you >> used to be able to). Specify this in your Modules file. >> >> >> On Fri, 18 Jan 2008, Brock Palen wrote: >> >>> Ok I will try setting them all to -1, >>> >>> The run just crashed trying to write checkpoint_002 >>> It segfault'ed when the file size was 2.2 GB. We were using HDF5 serial, >>> on luster. >>> >>> Thanks, >>> >>> Brock Palen >>> Center for Advanced Computing >>> brockp at umich.edu >>> (734)936-1985 >>> >>> >>> On Jan 18, 2008, at 1:17 PM, John ZuHone wrote: >>> >>>> Brock, >>>> >>>> I think that if you want no checkpoints at all you should set trstrt >>>> < 0, and make sure that nrstrt < 0 (integer) and zrstrt < 0 as well. >>>> Best, >>>> John ZuHone >>>> On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: >>>>> For parallel runs even if I set >>>>> trstrt=0 >>>>> two checkpoint files were written. >>>>> [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will send 10 >>>>> blocks per message. >>>>> [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint >>>>> name=Ni_Tem_hdf5_chk_0000 >>>>> [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint >>>>> name=Ni_Tem_hdf5_chk_0000 blocks=5449 >>>>> [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile >>>>> name=Ni_Tem_hdf5_plt_cnt_0000 >>>>> [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile >>>>> name=Ni_Tem_hdf5_plt_cnt_0000 >>>>> [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... >>>>> [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 >>>>> [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint >>>>> name=Ni_Tem_hdf5_chk_0001 >>>>> [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint >>>>> name=Ni_Tem_hdf5_chk_0001 blocks=5449 >>>>> No other checkpoint files have been written so far, though several plot >>>>> files have been written. Is this expected behavior? >>>>> Flash2.5 + pgi +openmpi-1.2.3 >>>>> hdf5/serial >>>>> Brock Palen >>>>> Center for Advanced Computing >>>>> brockp at umich.edu >>>>> (734)936-1985 >>> >> >> >> ----------------------------------------------------------------------------- >> Michael Zingale (mzingale at mail.astro.sunysb.edu) >> Assistant Professor >> >> Dept. of Physics and Astronomy office: ESS 440 >> SUNY Stony Brook phone: 631-632-8225 >> Stony Brook, NY 11794-3800 web: >> http://www.astro.sunysb.edu/mzingale >> ----------------------------------------------------------------------------- >> >> > ----------------------------------------------------------------------------- Michael Zingale (mzingale at mail.astro.sunysb.edu) Assistant Professor Dept. of Physics and Astronomy office: ESS 440 SUNY Stony Brook phone: 631-632-8225 Stony Brook, NY 11794-3800 web: http://www.astro.sunysb.edu/mzingale ----------------------------------------------------------------------------- From brockp at umich.edu Mon Jan 21 14:09:26 2008 From: brockp at umich.edu (Brock Palen) Date: Mon, 21 Jan 2008 15:09:26 -0500 Subject: [FLASH-USERS] disable check pointing In-Reply-To: References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> Message-ID: It still tried to write a checkpoint even with all those values set to -1. when doing a 3d problem. I will try with a mpi library compiled with mcmodel=medium, I don't know why I would need this. It failed at the same point when the second checkpoint was 2.2 gb. So it fails consistently. Brock Palen Center for Advanced Computing brockp at umich.edu (734)936-1985 On Jan 18, 2008, at 1:17 PM, John ZuHone wrote: > Brock, > > I think that if you want no checkpoints at all you should set > trstrt < 0, and make sure that nrstrt < 0 (integer) and zrstrt < 0 > as well. > > Best, > > John ZuHone > > On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: > >> For parallel runs even if I set >> >> trstrt=0 >> >> two checkpoint files were written. >> [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will >> send 10 blocks per message. >> [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint >> name=Ni_Tem_hdf5_chk_0000 >> [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint >> name=Ni_Tem_hdf5_chk_0000 blocks=5449 >> [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile >> name=Ni_Tem_hdf5_plt_cnt_0000 >> [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile >> name=Ni_Tem_hdf5_plt_cnt_0000 >> [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... >> [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 >> [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint >> name=Ni_Tem_hdf5_chk_0001 >> [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint >> name=Ni_Tem_hdf5_chk_0001 blocks=5449 >> >> No other checkpoint files have been written so far, though >> several plot files have been written. Is this expected behavior? >> >> Flash2.5 + pgi +openmpi-1.2.3 >> hdf5/serial >> >> >> Brock Palen >> Center for Advanced Computing >> brockp at umich.edu >> (734)936-1985 >> > > > From brockp at umich.edu Fri Jan 25 16:04:49 2008 From: brockp at umich.edu (Brock Palen) Date: Fri, 25 Jan 2008 17:04:49 -0500 Subject: [FLASH-USERS] flash network effects Message-ID: <45F91DE9-51BC-4904-81F7-B131DE708725@umich.edu> What really affects how many nodes a flash run can use? Also the notion that flash 'scales to hundreds of cpus' does that only apply to fast low latency networks. Going from just 1 to 2 nodes (8 to 16 cpus)i see system time jump to 50% of cpu time. Do uses use gig-e at all? This is all we have avilable right now. Any tricks for getting flash to run well on gig-e ? Brock Palen Center for Advanced Computing brockp at umich.edu (734)936-1985 From tomek at scs.fsu.edu Fri Jan 25 22:45:56 2008 From: tomek at scs.fsu.edu (Tomasz Plewa) Date: Fri, 25 Jan 2008 23:45:56 -0500 Subject: [FLASH-USERS] flash network effects In-Reply-To: <45F91DE9-51BC-4904-81F7-B131DE708725@umich.edu> References: <45F91DE9-51BC-4904-81F7-B131DE708725@umich.edu> Message-ID: <20080125234556.bil0m2e3kk888koc@webmail.scs.fsu.edu> Brock - Many codes will scale well provide the interconnect is sufficiently fast or, conversely, that processing elements are relatively slow. The slow network asks for limiting the off-processor (off-node) communication and maximizing the amount of local calculations. Plus the excellent scaling for FLASH can be obtained only in a limited set of physics-devoid cases (see below). No cure is readily available, but there are a few aspects worth considering: (1) Is your problem well-balanced? The more physics you include, the more variation in processor usage will likely be observed. (2) Is your algorithm simple enough? Yeah, hydro scales well. But hydro plus multigrid does not. (3) Is the problem size large enough or communication plays significant role already for small pools of PEs? (4) Have you tried any weak scaling tests for your particular problem? So does your problem scale "in theory"? (1 - 2 - 4 - 8 processors first...) And there are likely a few more... Greetings - Tomek -- School of Computational Science, DSL 443 ph: 850.644.1959 Florida State University fax: 850.644.0098 Tallahassee, FL 32306 web: people.scs.fsu.edu/~tomek/ -- Brock Palen wrote: > What really affects how many nodes a flash run can use? > Also the notion that flash 'scales to hundreds of cpus' does that only > apply to fast low latency networks. Going from just 1 to 2 nodes (8 to > 16 cpus)i see system time jump to 50% of cpu time. > > Do uses use gig-e at all? This is all we have avilable right now. Any > tricks for getting flash to run well on gig-e ? > > > Brock Palen > Center for Advanced Computing > brockp at umich.edu > (734)936-1985 > > > > -- > This message has been scanned for viruses and > dangerous content by MailScanner, and is > believed to be clean. ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. From latife at astro.rug.nl Tue Jan 29 05:05:22 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Tue, 29 Jan 2008 12:05:22 +0100 Subject: [FLASH-USERS] Problem in installing Flash View Message-ID: <479F0872.20808@astro.rug.nl> Hi Dear all, I am trying to install flash view on my pc. i have installed all libraries required for it but when i "make" .......... it gives the following error. In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: error: invalid conversion from ?hssize_t*? to ?const hsize_t*? /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: error: initializing argument 3 of ?herr_t H5Sselect_hyperslab(hid_t, H5S_seloper_t, const hsize_t*, const hsize_t*, const hsize_t*, const hsize_t*)? any body has experienced this problem? what could be the possible solution? kind regards M.A.Latife From hudson at mcs.anl.gov Tue Jan 29 08:06:59 2008 From: hudson at mcs.anl.gov (Randy Hudson) Date: Tue, 29 Jan 2008 08:06:59 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <479F0872.20808@astro.rug.nl> References: <479F0872.20808@astro.rug.nl> Message-ID: <479F3303.5080504@mcs.anl.gov> Which version of HDF5 are you using? M.A. Latife wrote: > Hi > Dear all, > I am trying to install flash view on my pc. i have installed all > libraries required for it but when i "make" .......... it gives the > following error. > In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: > /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: > error: invalid conversion from ?hssize_t*? to ?const hsize_t*? > /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: > error: initializing argument 3 of ?herr_t H5Sselect_hyperslab(hid_t, > H5S_seloper_t, const hsize_t*, const hsize_t*, const hsize_t*, const > hsize_t*)? > any body has experienced this problem? > what could be the possible solution? > kind regards > M.A.Latife > > -- Randy. From brockp at umich.edu Tue Jan 29 09:25:27 2008 From: brockp at umich.edu (Brock Palen) Date: Tue, 29 Jan 2008 10:25:27 -0500 Subject: [FLASH-USERS] disable check pointing In-Reply-To: References: <99DC0CC5-4B2A-43DA-B23B-237DF1A79275@umich.edu> Message-ID: Turns out to disable checkpoint files from being written there is a parameter wall_clock_checkpoint which defaults in both flash2.5 and flash3 to 12 hours. I set this to -1 in my flash.par rebuilt, and all appears well. I am now up to 29 plot's without a single checkpoint, when before I would always get killed around 10 plots. Of-course I would like checkpointing, so I would like to find the segfault and fix it ether in ROMIO, hdf5, or OpenMPI or flash. We were using hdf5-serial. I have not tried parallel. Note to anyone reading be sure to _always_ compile with optimizations for your compiler. (I use -fast under pgi-7.0, fastsse I have not verified correctness). A flash run compiled with -g and no overriding optimizations, is ~120 step's behind one compiled with -fast on the same hardware/problem/ libraries/etc. for the same wallclock. THanks guys I think we are all set to do research now, If there Is anything you want me to do to track down the segfault let me know. Brock Palen Center for Advanced Computing brockp at umich.edu (734)936-1985 On Jan 18, 2008, at 1:17 PM, John ZuHone wrote: > Brock, > > I think that if you want no checkpoints at all you should set > trstrt < 0, and make sure that nrstrt < 0 (integer) and zrstrt < 0 > as well. > > Best, > > John ZuHone > > On Jan 18, 2008, at 8:56 AM, Brock Palen wrote: > >> For parallel runs even if I set >> >> trstrt=0 >> >> two checkpoint files were written. >> [ 01-17-2008 23:51.27 ] message: [CHECKPOINT_WR] NOTE: will >> send 10 blocks per message. >> [ 01-17-2008 23:51.29 ] file_wr_open: type=checkpoint >> name=Ni_Tem_hdf5_chk_0000 >> [ 01-17-2008 23:52.30 ] file_wr_close: type=checkpoint >> name=Ni_Tem_hdf5_chk_0000 blocks=5449 >> [ 01-17-2008 23:52.32 ] file_wr_open: type=plotfile >> name=Ni_Tem_hdf5_plt_cnt_0000 >> [ 01-17-2008 23:52.36 ] file_wr_close: type=plotfile >> name=Ni_Tem_hdf5_plt_cnt_0000 >> [ 01-17-2008 23:52.36 ] [FLASH]: Enter evolution loop... >> [ 01-17-2008 23:52.36 ] step: n=1 t=0.000000E+00 dt=1.000000E-16 >> [ 01-17-2008 23:59.26 ] file_wr_open: type=checkpoint >> name=Ni_Tem_hdf5_chk_0001 >> [ 01-18-2008 00:00.13 ] file_wr_close: type=checkpoint >> name=Ni_Tem_hdf5_chk_0001 blocks=5449 >> >> No other checkpoint files have been written so far, though >> several plot files have been written. Is this expected behavior? >> >> Flash2.5 + pgi +openmpi-1.2.3 >> hdf5/serial >> >> >> Brock Palen >> Center for Advanced Computing >> brockp at umich.edu >> (734)936-1985 >> > > > From latife at astro.rug.nl Tue Jan 29 10:06:28 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Tue, 29 Jan 2008 17:06:28 +0100 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <479F3303.5080504@mcs.anl.gov> References: <479F0872.20808@astro.rug.nl> <479F3303.5080504@mcs.anl.gov> Message-ID: <479F4F04.7000800@astro.rug.nl> Hi i am using hdf5 version 1.6.6 cheers M.A.Latife Randy Hudson wrote: > > > Which version of HDF5 are you using? > > > M.A. Latife wrote: >> Hi >> Dear all, >> I am trying to install flash view on my pc. i have installed all >> libraries required for it but when i "make" .......... it gives the >> following error. >> In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: >> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: >> error: invalid conversion from ?hssize_t*? to ?const hsize_t*? >> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: >> error: initializing argument 3 of ?herr_t H5Sselect_hyperslab(hid_t, >> H5S_seloper_t, const hsize_t*, const hsize_t*, const hsize_t*, const >> hsize_t*)? >> any body has experienced this problem? >> what could be the possible solution? >> kind regards >> M.A.Latife >> >> > From klaus at flash.uchicago.edu Tue Jan 29 10:13:04 2008 From: klaus at flash.uchicago.edu (Klaus Weide) Date: Tue, 29 Jan 2008 10:13:04 -0600 (CST) Subject: [FLASH-USERS] Setup of hydrostatic equilibrium in FLASH 3.0beta? In-Reply-To: References: Message-ID: Cole, Sorry for responding late; we have been busy with preparing for the FLASH3.0 release. On Fri, 18 Jan 2008, Cole Miller wrote: > Has anyone set up a good hydrostatic equilibrium routine for FLASH > 3.0beta? I've got something that basically works (in the context of > an X-ray burst problem), but I'm concerned about its robustness and > there is a mysterious factor of 0.25 that I have to put in, so I'd > appreciate greater understanding. As far as I can tell, the setup > breaks into two tasks: initializing the interior cells, which seems > straightforward, and setting the boundary conditions at the edge, > which is where potential issues are arising. > > 1. For the interior cells, I simply wrote a little routine in > Simulation_initBlock.F90 that computes the vertical density structure > in hydrostatic equilibrium on a fine mesh. Then, when densities > are assigned in each block, I obtain the z coordinate at each point > and do a linear interpolation to find the density. The routine > for HSE assumed a user input of density at the minimum z value > in the simulation (assuming gravity is in the -z direction), > combined with a temperature (assumed constant throughout), then > computed the vertical density structure at the next z value up > using a search by bisection for a density that would yield a > pressure gradient force that counteracted gravity. This involved > calls to the Eos routine, e.g., > > eosData(EOS_TEMP) = sim_tempAmbient > dzz = (1.0/(Nhse-1.0))*(sim_zmax-sim_zmin) > do k = 1, Nhse > zz = sim_zmin + (k-1)*dzz > if (k .EQ. 1) then > density(k) = sim_rhoAmbient > eosData(EOS_DENS) = density(k) > call Eos(MODE_DENS_TEMP,1,eosData,massFraction) > pressure(k)=eosData(EOS_PRES) > else > dlogmin=0.0 ! Minimum ln density in cgs > dlogmax=log(density(k-1)) ! Max. This is dens of lower cell > do ibisec=0,30 > dlogmid=0.5*(dlogmin+dlogmax) > density(k)=exp(dlogmid) > eosData(EOS_DENS) = density(k) > call Eos(MODE_DENS_TEMP,1,eosData,massFraction) > pressure(k)=eosData(EOS_PRES) > !! Now we divide the pressure gradient by the average density. If the > !! pressure gradient is too small, then the density in the current > !! zone has to be less > gradient=(pressure(k-1)-pressure(k))/dzz > gradient=gradient/(0.5*(density(k)+density(k-1))) > if (gradient<(-1.00*sim_gconst)) then > dlogmax=dlogmid > else > dlogmin=dlogmid > endif > enddo Note that you may be able to change this from a bisection method to Newton-Raphson, by taking advantage of the derivatives (like eosData(EOS_DPD) ) that your Eos call should already compute. > Here Nhse is a large value (1001 in my case) to make later > interpolation more accurate. sim_tempAmbient, sim_rhoAmbient, > and sim_gconst are input in flash.par. This seems to work fine. > > > 2. However, when I ran this, although the inner regions seemed in good > equilibrium, the bottom part of the simulation is not; there is a lot > of bouncing and thus generation of heat. Assuming this is because of > the lower boundary condition, I started looking at > Grid_applyBCEdge.F90. > > In principle, I'd like to do the same thing here as in > Simulation_initBlock.F90. That is, for each of the guard cells, > I want to be able to figure out the vertical coordinate and then > do something similar to what I did above. For example, for the > guard cells off the low end of the grid, the *upper* > boundary condition would be a density of sim_rhoAmbient and a > temperature of sim_tempAmbient. > > I ran into two problems, possibly related to each other: > > (a) I was not able to find a specific routine that would tell me the > coordinates of the center of a guard cell. The only routine that > I saw that gives cell coordinates is Grid_getCellCoords, and one > can use this to get the coordinates of guard cells. This, > however, requires a blockId as input. I've been using > blockId=1 and it seems to work for the simple cases I've > tried, but I'm worried that it will fail after refinement. > Is Grid_applyBCEdge.F90 called only once, or for every block > that borders an edge? Is there a separate routine that gives > the coordinates of guard cells? You are right, there is no clean way to get at the cell coordinates within BC handling in FLASH 3.0beta. We are going to provide ways to do this in FLASH3.0, to be released soon. Even if you could get the "blockId" in Grid_applyBCEdge.F90 there would be possible complications, because Paramesh may sometimes call this routine on behave of a data block that belongs to a different PE and thus does not, strictly speaking, have a valid blockId on the local PE. So you would also need a modified Grid_getCellCoords that can deal with this kind of situation. (These difficulties are being addressed in FLASH3.0.) Your trick of always using blockId=1 should of course only work if the cells you are trying to fill are cells of block #1. Or, if you just need to get the cell spacing (in which case you should just use Grid_getDeltas), it should work iff the block you are working on has the same resolution as block #1 -- which normally is not the case for any other block, since #1 is the root block and a parent or ancestor of all others (unless you use an NblockX / NblockY / NblockZ > 1). > (b) When I made this choice, I set up a routine similar to the one > I reproduced above, but in Grid_applyBCEdge.F90. It gives me > good equilibrium conditions, but only if I adopt a strange > ad hoc factor. What I seem to need to do is that instead of > if (gradient<(-1.00*sim_gconst)) then > I have to use > if (gradient<(-0.25*sim_gconst)) then > > This factor is independent of the vertical resolution for the > cases I've tried (of course, the equilibrium is better for better > vertical resolution). But did the block whose guard cells you are fillling have the same resolution as your root block, block #1? If not, maybe that is how the factor comes in: you resolution is off by a factor of 4? To understand this better, we would have to know more about your setup. Especially the number of blocks, lrefine_{min,max}, etc. I hope this helps, Klaus From hudson at mcs.anl.gov Tue Jan 29 10:17:38 2008 From: hudson at mcs.anl.gov (Randy Hudson) Date: Tue, 29 Jan 2008 10:17:38 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <479F0872.20808@astro.rug.nl> References: <479F0872.20808@astro.rug.nl> Message-ID: <479F51A2.5010900@mcs.anl.gov> I recommend trying VisIt. It can do much more than FlashView. You can download visit1.7.1.exe from https://wci.llnl.gov/codes/visit/executables.html. (There are also executables for several other platforms.) If you decide to try that and have trouble, or if FlashView does something for you VisIt doesn't do, let me know. VisIt documentation is at https://wci.llnl.gov/codes/visit/manuals.html. M.A. Latife wrote: > Hi > Dear all, > I am trying to install flash view on my pc. i have installed all > libraries required for it but when i "make" .......... it gives the > following error. > In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: > /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: > error: invalid conversion from ?hssize_t*? to ?const hsize_t*? > /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C:358: > error: initializing argument 3 of ?herr_t H5Sselect_hyperslab(hid_t, > H5S_seloper_t, const hsize_t*, const hsize_t*, const hsize_t*, const > hsize_t*)? > any body has experienced this problem? > what could be the possible solution? > kind regards > M.A.Latife > > -- Randy. From smc at astro.as.utexas.edu Tue Jan 29 10:59:30 2008 From: smc at astro.as.utexas.edu (Sean Couch) Date: Tue, 29 Jan 2008 10:59:30 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <479F51A2.5010900@mcs.anl.gov> References: <479F0872.20808@astro.rug.nl> <479F51A2.5010900@mcs.anl.gov> Message-ID: <15DD01AB-CFFA-4B3F-BA52-34D04AD97DC5@astro.as.utexas.edu> Hi Randy, I have been trying to get VisIt to work on my institutions Linux cluster for a while now and have been running into problems. Does any one have experience setting up VisIt to work in parallel on Linux clusters using the LSF qsub system? We have been successful in compiling and running VisIt in parallel but the interaction with the batch job system is shaky. The only way I can successfully start multiple parallel compute engines is to (accidentally) use processors on the login node, where the mdserver is running. This is, of course, not a good solution, so the system administrators have told me... Any help would be greatly appreciated. Sean Couch --------------------------------------------- Sean M. Couch Graduate Research Assistant Department of Astronomy The University of Texas at Austin 1 University Station C1400 RLM 15.202A (512) 471-8316 smc at astro.as.utexas.edu www.as.utexas.edu/~smc On Jan 29, 2008, at 10:17 AM, Randy Hudson wrote: > > > I recommend trying VisIt. It can do much more than FlashView. > > You can download visit1.7.1.exe from https://wci.llnl.gov/codes/visit/executables.html > . (There are also executables for several other platforms.) > > If you decide to try that and have trouble, or if FlashView does > something for you VisIt doesn't do, let me know. > > VisIt documentation is at https://wci.llnl.gov/codes/visit/manuals.html > . > > > M.A. Latife wrote: >> Hi >> Dear all, >> I am trying to install flash view on my pc. i have installed all >> libraries required for it but when i "make" .......... it gives the >> following error. >> In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: >> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: >> 358: error: invalid conversion from ?hssize_t*? to ?const hsize_t*? >> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: >> 358: error: initializing argument 3 of ?herr_t >> H5Sselect_hyperslab(hid_t, H5S_seloper_t, const hsize_t*, const >> hsize_t*, const hsize_t*, const hsize_t*)? >> any body has experienced this problem? >> what could be the possible solution? >> kind regards >> M.A.Latife > > -- > > Randy. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://flash.uchicago.edu/pipermail/flash-users/attachments/20080129/4153654f/attachment.html From tomek at scs.fsu.edu Tue Jan 29 12:05:09 2008 From: tomek at scs.fsu.edu (Tomasz Plewa) Date: Tue, 29 Jan 2008 13:05:09 -0500 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <15DD01AB-CFFA-4B3F-BA52-34D04AD97DC5@astro.as.utexas.edu> References: <479F0872.20808@astro.rug.nl> <479F51A2.5010900@mcs.anl.gov> <15DD01AB-CFFA-4B3F-BA52-34D04AD97DC5@astro.as.utexas.edu> Message-ID: <20080129130509.ds2p69la0wgg0w00@webmail.scs.fsu.edu> Sean - Are you sure you are going to need the parallel engine? It may not worth the trouble (for now). I can do most of my visualizations without it. Tomek -- School of Computational Science, DSL 443 ph: 850.644.1959 Florida State University fax: 850.644.0098 Tallahassee, FL 32306 web: people.scs.fsu.edu/~tomek/ -- Sean Couch wrote: > Hi Randy, > > I have been trying to get VisIt to work on my institutions Linux > cluster for a while now and have been running into problems. Does any > one have experience setting up VisIt to work in parallel on Linux > clusters using the LSF qsub system? > > We have been successful in compiling and running VisIt in parallel but > the interaction with the batch job system is shaky. The only way I > can successfully start multiple parallel compute engines is to > (accidentally) use processors on the login node, where the mdserver is > running. This is, of course, not a good solution, so the system > administrators have told me... > > Any help would be greatly appreciated. > > Sean Couch > > --------------------------------------------- > Sean M. Couch > Graduate Research Assistant > Department of Astronomy > The University of Texas at Austin > 1 University Station C1400 > RLM 15.202A > (512) 471-8316 > smc at astro.as.utexas.edu > www.as.utexas.edu/~smc > > On Jan 29, 2008, at 10:17 AM, Randy Hudson wrote: > >> >> >> I recommend trying VisIt. It can do much more than FlashView. >> >> You can download visit1.7.1.exe from >> https://wci.llnl.gov/codes/visit/executables.html . (There are >> also executables for several other platforms.) >> >> If you decide to try that and have trouble, or if FlashView does >> something for you VisIt doesn't do, let me know. >> >> VisIt documentation is at https://wci.llnl.gov/codes/visit/manuals.html . >> >> >> M.A. Latife wrote: >>> Hi >>> Dear all, >>> I am trying to install flash view on my pc. i have installed all >>> libraries required for it but when i "make" .......... it gives >>> the following error. >>> In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: >>> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: 358: >>> error: invalid conversion from ?hssize_t*? to ?const hsize_t*? >>> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: 358: >>> error: initializing argument 3 of ?herr_t >>> H5Sselect_hyperslab(hid_t, H5S_seloper_t, const hsize_t*, const >>> hsize_t*, const hsize_t*, const hsize_t*)? >>> any body has experienced this problem? >>> what could be the possible solution? >>> kind regards >>> M.A.Latife >> >> -- >> >> Randy. > > > -- > This message has been scanned for viruses and > dangerous content by MailScanner, and is > believed to be clean. ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. From jbgallag at flash.uchicago.edu Tue Jan 29 13:45:28 2008 From: jbgallag at flash.uchicago.edu (Brad Gallagher) Date: Tue, 29 Jan 2008 13:45:28 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <479F0872.20808@astro.rug.nl> References: <479F0872.20808@astro.rug.nl> Message-ID: M.A. , The "hssize_t" type was deprecated sometime in the hdf5 1.6.x series, so one way to fix this would be to edit flashhdf5.C and change all types of "hssize_t" to "hsize_t" , this should allow FlashView to compile against a 1.6.x hdf5 library. I'll look into getting this changed in the release distribution. Thanks for pointing this out and please let us know if you still have problems. --Brad On Jan 29, 2008, at 5:05 AM, M.A. Latife wrote: > Hi > Dear all, > I am trying to install flash view on my pc. i have installed all > libraries required for it but when i "make" .......... it gives the > following error. > In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: > /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: > 358: error: invalid conversion from ?hssize_t*? to ?const hsize_t*? > /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: > 358: error: initializing argument 3 of ?herr_t > H5Sselect_hyperslab(hid_t, H5S_seloper_t, const hsize_t*, const > hsize_t*, const hsize_t*, const hsize_t*)? > any body has experienced this problem? > what could be the possible solution? > kind regards > M.A.Latife From jbgallag at flash.uchicago.edu Tue Jan 29 13:47:27 2008 From: jbgallag at flash.uchicago.edu (Brad Gallagher) Date: Tue, 29 Jan 2008 13:47:27 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: References: <479F0872.20808@astro.rug.nl> Message-ID: <96B3172C-2468-4DB2-A4D4-668C7C4E0CFC@flash.uchicago.edu> M.A., I would also second Randy's advice however, Visit has a lot more functionality. --Brad On Jan 29, 2008, at 1:45 PM, Brad Gallagher wrote: > M.A. , > > The "hssize_t" type was deprecated sometime in the hdf5 1.6.x > series, so one way to fix this would be to edit flashhdf5.C and > change all types of "hssize_t" to "hsize_t" , this should allow > FlashView to compile against a 1.6.x hdf5 library. I'll look into > getting this changed in the release distribution. Thanks for > pointing this out and please let us know if you still have problems. > > --Brad > On Jan 29, 2008, at 5:05 AM, M.A. Latife wrote: > >> Hi >> Dear all, >> I am trying to install flash view on my pc. i have installed all >> libraries required for it but when i "make" .......... it gives the >> following error. >> In member function ?int FlashHDFFile::GetGlobalIds(int, int*)?: >> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: >> 358: error: invalid conversion from ?hssize_t*? to ?const hsize_t*? >> /Users/users/latife/Data/flashview/FlashView/FlashHDF5/flashhdf5.C: >> 358: error: initializing argument 3 of ?herr_t >> H5Sselect_hyperslab(hid_t, H5S_seloper_t, const hsize_t*, const >> hsize_t*, const hsize_t*, const hsize_t*)? >> any body has experienced this problem? >> what could be the possible solution? >> kind regards >> M.A.Latife > From hudson at mcs.anl.gov Tue Jan 29 14:04:01 2008 From: hudson at mcs.anl.gov (Randy Hudson) Date: Tue, 29 Jan 2008 14:04:01 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: References: <479F0872.20808@astro.rug.nl> Message-ID: <479F86B1.6010708@mcs.anl.gov> I'm currently testing flashview and preparing to commit the latest version, which uses visprep instead of flashhdf. Brad Gallagher wrote: > M.A. , > > The "hssize_t" type was deprecated sometime in the hdf5 1.6.x series, so > one way to fix this would be to edit flashhdf5.C and change all types of > "hssize_t" to "hsize_t" , this should allow FlashView to compile against > a 1.6.x hdf5 library. I'll look into getting this changed in the release > distribution. Thanks for pointing this out and please let us know if you > still have problems. > > --Brad -- Randy. From latife at astro.rug.nl Wed Jan 30 12:11:11 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Wed, 30 Jan 2008 19:11:11 +0100 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <479F86B1.6010708@mcs.anl.gov> References: <479F0872.20808@astro.rug.nl> <479F86B1.6010708@mcs.anl.gov> Message-ID: <47A0BDBF.8030105@astro.rug.nl> Hi Dear all, i am using visit 1.7.1 but problem is that my machine gets stuck when i am using visit. i have tested it on other machines as well, same problem occurs. when only visit is running and all other applications are closed then it is fine but when i start other applications then machine(which i use) gets stuck. Is it good for line plots? I am interested in visualizing one dimensional data (flash output Data for 1-d spherical geometry) According to your advice i am going to use VisIt. i have still this problem in installing Flash view. from /Users/users/latife/Data/flashview/FlashView/GUI.h:5, from /Users/users/latife/Data/flashview/FlashView/App.h:6, from /Users/users/latife/Data/flashview/FlashView/Main.C:4: /Users/users/latife/Data/flashview/FlashView/Visualizer.h:5:26: error: vtkInputPort.h: No such file or directory /Users/users/latife/Data/flashview/FlashView/Visualizer.h:6:27: error: vtkOutputPort.h: No such file or directory /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: ISO C++ forbids declaration of ?vtkOutputPort? with no type /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: expected ?;? before ?*? token cheers M.A.Latif Randy Hudson wrote: > > I'm currently testing flashview and preparing to commit the latest > version, which uses visprep instead of flashhdf. > > Brad Gallagher wrote: >> M.A. , >> >> The "hssize_t" type was deprecated sometime in the hdf5 1.6.x series, >> so one way to fix this would be to edit flashhdf5.C and change all >> types of "hssize_t" to "hsize_t" , this should allow FlashView to >> compile against a 1.6.x hdf5 library. I'll look into getting this >> changed in the release distribution. Thanks for pointing this out and >> please let us know if you still have problems. >> >> --Brad > From tomek at scs.fsu.edu Wed Jan 30 12:13:55 2008 From: tomek at scs.fsu.edu (Tomasz Plewa) Date: Wed, 30 Jan 2008 13:13:55 -0500 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <47A0BDBF.8030105@astro.rug.nl> References: <479F0872.20808@astro.rug.nl> <479F86B1.6010708@mcs.anl.gov> <47A0BDBF.8030105@astro.rug.nl> Message-ID: <20080130131355.kc2c1zpd4w4w4kwo@webmail.scs.fsu.edu> M.A. - You may want to report VisIt freez to the visit mailing list. We have experience the same problem here running on CentOS 4.5 and 5.x machines (Intel and AMD). VisIt works all right in 1D, yes. Tomek -- School of Computational Science, DSL 443 ph: 850.644.1959 Florida State University fax: 850.644.0098 Tallahassee, FL 32306 web: people.scs.fsu.edu/~tomek/ -- M.A. Latife wrote: > Hi > Dear all, > i am using visit 1.7.1 but problem is that my machine gets stuck when i > am using visit. i have tested it on other machines as well, same > problem occurs. when only visit is running and all other applications > are closed then it is fine but when i start other applications then > machine(which i use) gets stuck. > Is it good for line plots? > > I am interested in visualizing one dimensional data (flash output Data > for 1-d spherical geometry) > According to your advice i am going to use VisIt. > i have still this problem in installing Flash view. > > from /Users/users/latife/Data/flashview/FlashView/GUI.h:5, > from /Users/users/latife/Data/flashview/FlashView/App.h:6, > from /Users/users/latife/Data/flashview/FlashView/Main.C:4: > /Users/users/latife/Data/flashview/FlashView/Visualizer.h:5:26: error: > vtkInputPort.h: No such file or directory > /Users/users/latife/Data/flashview/FlashView/Visualizer.h:6:27: error: > vtkOutputPort.h: No such file or directory > /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: > error: ISO C++ forbids declaration of ?vtkOutputPort? with no type > /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: > error: expected ?;? before ?*? token > > cheers > M.A.Latif > > Randy Hudson wrote: >> >> I'm currently testing flashview and preparing to commit the latest >> version, which uses visprep instead of flashhdf. >> >> Brad Gallagher wrote: >>> M.A. , >>> >>> The "hssize_t" type was deprecated sometime in the hdf5 1.6.x >>> series, so one way to fix this would be to edit flashhdf5.C and >>> change all types of "hssize_t" to "hsize_t" , this should allow >>> FlashView to compile against a 1.6.x hdf5 library. I'll look into >>> getting this changed in the release distribution. Thanks for >>> pointing this out and please let us know if you still have problems. >>> >>> --Brad >> > > > -- > This message has been scanned for viruses and > dangerous content by MailScanner, and is > believed to be clean. ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. From latife at astro.rug.nl Thu Jan 31 08:10:35 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Thu, 31 Jan 2008 15:10:35 +0100 Subject: [FLASH-USERS] Question about star formation using FLASH Message-ID: <47A1D6DB.4070103@astro.rug.nl> Hi all, one of my colleagues is interested in using FLASH, His question is as follows, i Hope you people can reply his question in better way. "I want to simulate star formation starting with gas particles. Is Flash able to convert gas particles (when a certain condition is met) into star particles? This because I want to see the initial mass function. Do I have to write my own module for this, or is there such an option already there? " Thanks, M.A.Latife From hudson at mcs.anl.gov Thu Jan 31 08:41:44 2008 From: hudson at mcs.anl.gov (Randy Hudson) Date: Thu, 31 Jan 2008 08:41:44 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <47A0BDBF.8030105@astro.rug.nl> References: <479F0872.20808@astro.rug.nl> <479F86B1.6010708@mcs.anl.gov> <47A0BDBF.8030105@astro.rug.nl> Message-ID: <47A1DE28.5040209@mcs.anl.gov> M, Re the first 2 errors: when I've built VTK, the installation step has failed to collect the headers all in one nice place. Could it be that vtkInput- and vtkOutputPort are somewhere in the vtk tree, and that you have to add a "-I" argument to your flashview build (or move the header files)? I think fixing the first 2 errors will eliminate the second 2. M.A. Latife wrote: > Hi > Dear all, > i am using visit 1.7.1 but problem is that my machine gets stuck when i > am using visit. i have tested it on other machines as well, same problem > occurs. when only visit is running and all other applications are closed > then it is fine but when i start other applications then machine(which i > use) gets stuck. > Is it good for line plots? > > I am interested in visualizing one dimensional data (flash output Data > for 1-d spherical geometry) > According to your advice i am going to use VisIt. > i have still this problem in installing Flash view. > > from /Users/users/latife/Data/flashview/FlashView/GUI.h:5, > from /Users/users/latife/Data/flashview/FlashView/App.h:6, > from /Users/users/latife/Data/flashview/FlashView/Main.C:4: > /Users/users/latife/Data/flashview/FlashView/Visualizer.h:5:26: error: > vtkInputPort.h: No such file or directory > /Users/users/latife/Data/flashview/FlashView/Visualizer.h:6:27: error: > vtkOutputPort.h: No such file or directory > /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: > ISO C++ forbids declaration of ?vtkOutputPort? with no type > /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: > expected ?;? before ?*? token > > cheers > M.A.Latif > -- Randy. From hudson at mcs.anl.gov Thu Jan 31 08:48:35 2008 From: hudson at mcs.anl.gov (Randy Hudson) Date: Thu, 31 Jan 2008 08:48:35 -0600 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <47A0BDBF.8030105@astro.rug.nl> References: <479F0872.20808@astro.rug.nl> <479F86B1.6010708@mcs.anl.gov> <47A0BDBF.8030105@astro.rug.nl> Message-ID: <47A1DFC3.7030305@mcs.anl.gov> 1 Under which OS's has visit gotten stuck? 2 Are there any error messages associated with the sticking? I don't remember visit ever hanging because there are other app's running. And, I second Tomasz recommendation that you email the visit help list (visit-help at llnl.gov). M.A. Latife wrote: > Hi > Dear all, > i am using visit 1.7.1 but problem is that my machine gets stuck when i > am using visit. i have tested it on other machines as well, same problem > occurs. when only visit is running and all other applications are closed > then it is fine but when i start other applications then machine(which i > use) gets stuck. > Is it good for line plots? > > I am interested in visualizing one dimensional data (flash output Data > for 1-d spherical geometry) > According to your advice i am going to use VisIt. > i have still this problem in installing Flash view. > > from /Users/users/latife/Data/flashview/FlashView/GUI.h:5, > from /Users/users/latife/Data/flashview/FlashView/App.h:6, > from /Users/users/latife/Data/flashview/FlashView/Main.C:4: > /Users/users/latife/Data/flashview/FlashView/Visualizer.h:5:26: error: > vtkInputPort.h: No such file or directory > /Users/users/latife/Data/flashview/FlashView/Visualizer.h:6:27: error: > vtkOutputPort.h: No such file or directory > /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: > ISO C++ forbids declaration of ?vtkOutputPort? with no type > /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: > expected ?;? before ?*? token > > cheers > M.A.Latif > > Randy Hudson wrote: >> >> I'm currently testing flashview and preparing to commit the latest >> version, which uses visprep instead of flashhdf. >> >> Brad Gallagher wrote: >>> M.A. , >>> >>> The "hssize_t" type was deprecated sometime in the hdf5 1.6.x series, >>> so one way to fix this would be to edit flashhdf5.C and change all >>> types of "hssize_t" to "hsize_t" , this should allow FlashView to >>> compile against a 1.6.x hdf5 library. I'll look into getting this >>> changed in the release distribution. Thanks for pointing this out and >>> please let us know if you still have problems. >>> >>> --Brad >> > > -- Randy. From latife at astro.rug.nl Thu Jan 31 08:56:34 2008 From: latife at astro.rug.nl (M.A. Latife) Date: Thu, 31 Jan 2008 15:56:34 +0100 Subject: [FLASH-USERS] Problem in installing Flash View In-Reply-To: <47A1DFC3.7030305@mcs.anl.gov> References: <479F0872.20808@astro.rug.nl> <479F86B1.6010708@mcs.anl.gov> <47A0BDBF.8030105@astro.rug.nl> <47A1DFC3.7030305@mcs.anl.gov> Message-ID: <47A1E1A2.7090105@astro.rug.nl> Hi Randy, 1 Both under RedHat 4 Enterprise and CentOS 5 2 No error messages I know of. I am not the only one with this problem. Tomek wrote that he experienced similar problems. See his message below. "You may want to report VisIt freez to the visit mailing list. We have experience the same problem here running on CentOS 4.5 and 5.x machines (Intel and AMD). VisIt works all right in 1D, yes. Tomek " Kind Regards, M.A. Latif Randy Hudson wrote: > > > 1 Under which OS's has visit gotten stuck? > 2 Are there any error messages associated with the sticking? > > I don't remember visit ever hanging because there are other app's > running. > > And, I second Tomasz recommendation that you email the visit help list > (visit-help at llnl.gov). > > > M.A. Latife wrote: >> Hi >> Dear all, >> i am using visit 1.7.1 but problem is that my machine gets stuck when >> i am using visit. i have tested it on other machines as well, same >> problem occurs. when only visit is running and all other applications >> are closed then it is fine but when i start other applications then >> machine(which i use) gets stuck. >> Is it good for line plots? >> >> I am interested in visualizing one dimensional data (flash output >> Data for 1-d spherical geometry) >> According to your advice i am going to use VisIt. >> i have still this problem in installing Flash view. >> >> from /Users/users/latife/Data/flashview/FlashView/GUI.h:5, >> from /Users/users/latife/Data/flashview/FlashView/App.h:6, >> from /Users/users/latife/Data/flashview/FlashView/Main.C:4: >> /Users/users/latife/Data/flashview/FlashView/Visualizer.h:5:26: >> error: vtkInputPort.h: No such file or directory >> /Users/users/latife/Data/flashview/FlashView/Visualizer.h:6:27: >> error: vtkOutputPort.h: No such file or directory >> /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: >> ISO C++ forbids declaration of ?vtkOutputPort? with no type >> /Users/users/latife/Data/flashview/FlashView/Pipebranch.h:628: error: >> expected ?;? before ?*? token >> >> cheers >> M.A.Latif >> >> Randy Hudson wrote: >>> >>> I'm currently testing flashview and preparing to commit the latest >>> version, which uses visprep instead of flashhdf. >>> >>> Brad Gallagher wrote: >>>> M.A. , >>>> >>>> The "hssize_t" type was deprecated sometime in the hdf5 1.6.x >>>> series, so one way to fix this would be to edit flashhdf5.C and >>>> change all types of "hssize_t" to "hsize_t" , this should allow >>>> FlashView to compile against a 1.6.x hdf5 library. I'll look into >>>> getting this changed in the release distribution. Thanks for >>>> pointing this out and please let us know if you still have problems. >>>> >>>> --Brad >>> >> >> > From zuhone at flash.uchicago.edu Thu Jan 31 15:22:27 2008 From: zuhone at flash.uchicago.edu (John ZuHone) Date: Thu, 31 Jan 2008 15:22:27 -0600 Subject: [FLASH-USERS] Question about star formation using FLASH In-Reply-To: <47A1D6DB.4070103@astro.rug.nl> References: <47A1D6DB.4070103@astro.rug.nl> Message-ID: <30C8040D-13BB-4956-9E84-F178C2B6C698@flash.uchicago.edu> Hello M.A., Such capabilities are not currently in FLASH2 and they will not be in the first release of FLASH3, though they *may* be in future releases. Others have managed to get this working for their own use. Strictly speaking one would not be converting gas particles (since we're working with Eulerian grid variables) but would be creating massive star particles from scratch at a particular grid location once conditions were met at that location (say, density rises to a certain value and temperature falls below a certain value). Then the star particle would be taking its velocity as well from the local velocity on the grid. In essence, this is not currently implemented but once one had a prescription for this it should not be difficult to implement. Best, John Z On Jan 31, 2008, at 8:10 AM, M.A. Latife wrote: > > Hi all, > one of my colleagues is interested in using FLASH, His question is > as follows, i Hope you people can reply his question in better way. > "I want to simulate star formation starting with gas particles. Is > Flash able to convert gas particles (when a certain condition is > met) into star particles? This because I want to see the initial > mass function. Do I have to write my own module for this, or is > there such an option already there? " > > Thanks, > M.A.Latife From rfisher at flash.uchicago.edu Thu Jan 31 16:56:07 2008 From: rfisher at flash.uchicago.edu (Robert Fisher) Date: Thu, 31 Jan 2008 16:56:07 -0600 (CST) Subject: [FLASH-USERS] Question about star formation using FLASH In-Reply-To: <30C8040D-13BB-4956-9E84-F178C2B6C698@flash.uchicago.edu> References: <47A1D6DB.4070103@astro.rug.nl> <30C8040D-13BB-4956-9E84-F178C2B6C698@flash.uchicago.edu> Message-ID: To elaborate upon what John has stated, star particles like those you have described have not been implemented into FLASH. However, FLASH does support active, gravitating particles, so the addition of such a feature would be relatively straightforward -- the algorithm has already been fully laid out, for instance in Krumholz et al (2004) -- http://adsabs.harvard.edu/abs/2004ApJ...611..399K I am certain your colleague could obtain assistance where needed from both the FLASH code group and the user community. Best wishes, Bob On Thu, 31 Jan 2008, John ZuHone wrote: > Hello M.A., > > Such capabilities are not currently in FLASH2 and they will not be > in the first release of FLASH3, though they *may* be in future releases. > Others have managed to get this working for their own use. > > Strictly speaking one would not be converting gas particles (since > we're working with Eulerian grid variables) but would be creating massive > star particles from scratch at a particular grid location once conditions > were met at that location (say, density rises to a certain value and > temperature falls below a certain value). Then the star particle would be > taking its velocity as well from the local velocity on the grid. > > In essence, this is not currently implemented but once one had a > prescription for this it should not be difficult to implement. > > Best, > > John Z > > On Jan 31, 2008, at 8:10 AM, M.A. Latife wrote: > >> >> Hi all, >> one of my colleagues is interested in using FLASH, His question is as >> follows, i Hope you people can reply his question in better way. >> "I want to simulate star formation starting with gas particles. Is Flash >> able to convert gas particles (when a certain condition is met) into star >> particles? This because I want to see the initial mass function. Do I have >> to write my own module for this, or is there such an option already there? >> " >> >> Thanks, >> M.A.Latife