oxfordska / oskar Goto Github PK

Looks like some more guard macros might be needed.

Output:

      Start  1: binary_test
 1/13 Test  #1: binary_test ......................   Passed    0.01 sec
      Start  2: settings_test
 2/13 Test  #2: settings_test ....................   Passed    0.01 sec
      Start  3: convert_test
 3/13 Test  #3: convert_test .....................***Failed    0.03 sec
      Start  4: correlate_test
 4/13 Test  #4: correlate_test ...................   Passed    0.21 sec
      Start  5: imager_test
 5/13 Test  #5: imager_test ......................   Passed    1.24 sec
      Start  6: math_test
 6/13 Test  #6: math_test ........................   Passed    0.22 sec
      Start  7: mem_test
 7/13 Test  #7: mem_test .........................***Failed    0.06 sec
      Start  8: jones_test
 8/13 Test  #8: jones_test .......................***Failed    0.11 sec
      Start  9: sky_test
 9/13 Test  #9: sky_test .........................***Exception: SegFault  0.08 sec
      Start 10: station_test
10/13 Test #10: station_test .....................***Exception: SegFault  0.02 sec
      Start 11: telescope_test
11/13 Test #11: telescope_test ...................***Failed    0.22 sec
      Start 12: utility_test
12/13 Test #12: utility_test .....................   Passed    0.13 sec
      Start 13: vis_test
13/13 Test #13: vis_test .........................***Failed    0.44 sec

Hi, I am using Python 3.8.10. I am trying to run the python code hello-world.py. For my project, I need to simulate visibilities for only one station but when I am trying to do so I am getting the following error:

Traceback (most recent call last):
  File "hello-world.py", line 62, in <module>
    output = imager.run(return_images=1)
  File "/usr/local/lib/python3.8/dist-packages/oskar/imager.py", line 824, in run
    return _imager_lib.run(self._capsule, return_images, return_grids)
RuntimeError: oskar_imager_run() failed with code -115 (data tag not found in file).

But there is no error when I use more than one station for the simulation.I am not sure why this is happening. Please can you help with this issue and also explain how to predict the visibilities of only one station with custom layout?

Thanking you in advance,
@fdulwich

Hi,

I'm having a bit of trouble installing with GPU support: I've got CUDA 11.2 installed and it looks like that newest CUDA version supported is 8.2 - so I have installed CUDA 8.0 and then used the -DCUDA_ARCH="8.0" flag, but unfortunately then get nvcc fatal : Unsupported gpu architecture 'compute_80' when I run make. I wonder if I need CUDA>8.2 for the GPU but <=8.2 for OSKAR?

I'm on a Ubuntu 18.04 machine and have a Tesla V100. I'll do a bit more digging but wondered if I'm missing something obvious with versions etc.

Many thanks!

Hi Team,

We want to report an issue with the current version of OSKAR on the M1-based Macs. OSKAR could not use the GPU onboard while runs fine with the CPUs when installed successfully. We tried both the direct installation from the DMG file and the build-from-source approach. Although building OSKAR through the native ARM compiler failed (not surprisingly), building it via the ROSETTA 2 emulator did work. But we are getting the error buildComputeProgram: cl2Metal failed clBuildProgram error (code -11) when trying to let OSKAR use GPU. Looking online, it seems other programs also have this issue on M1 macs.

Did you guys try to run OSKAR with the M1 Macs? Not sure if there is a solution to this error. It would be great to get OSKAR to use the M1 GPUs, and a lot of people may benefit from the solution.

Please see the full log below.

I am trying to set OSKAR running on a cluster managed by slurm. Using the singularity image to run a simple test it returns an error ERROR: Run failed with code 209: no kernel image is available for execution on the device.

Maybe this is a stupid guess, but I am wondering if there is any compatibility issue with the singularity image and more recent version of driver? All the nodes on this machine use the latest NVIDIA driver and CUDA. Any help or suggestion would be great. Many thanks!

The error log is:

 |                                                                   
=|== OSKAR-2.7.6 starting at 2022-03-04, 03:35:29 (GMT).
 |                                                                   
 | + Current dir is /share/nas/zchen/skalow
 | + Logging to file oskar_2022-03-04_033529.log
 |                                                                   
=|== Loaded settings file 'beam.ini'
 |                                                                   
 | + Simulator settings
 |   - Use double precision              : false
 |   - Use GPUs                          : true
 |   - CUDA device IDs to use            : all
 |   - Number of compute devices         : auto
 |   - Max. number of sources per chunk  : 16384
 | + Observation settings
 |   - Phase centre RA [deg]             : 0.0
 |   - Phase centre Dec [deg]            : -27.0
 |   - Start frequency [Hz]              : 203069585.01
 |   - Number of frequency channels      : 1
 |   - Frequency increment [Hz]          : 781250.0
 |   - Start time (UTC)                  : 01-01-2000 07:33:04.864
 |   - Observation length [sec, or hh:mm:ss]: 14400.0
 |   - Number of time steps              : 1
 | + Telescope model settings
 |   - Input directory                   : SKA_core_area_Rev3.tm/
 |   - Normalise beams at phase centre   : true
 |   - Allow station beam duplication    : false
 |   - Polarisation mode                 : Scalar
 |   - Station type                      : Aperture array
 | + Beam pattern settings
 |   - All stations                      : false
 |   - Station ID(s)                     : 0
 |   - Coordinate frame                  : Equatorial
 |   - Beam image
 |     * Image dimensions [pixels]       : 512
 |     * Field of view [deg]             : 20.0
 |   - Output root path name             : Beam/
 |   - Per-station outputs
 |     * FITS image
 |       + Amplitude pattern             : true
 |                                                                   
=|== Telescope model summary
 |                                                                   
 | + Longitude [deg]                     : 116.764
 | + Latitude [deg]                      : -26.825
 | + Altitude [m]                        : 0
 | + Num. stations                       : 224
 | + Max station size                    : 256
 | + Max station depth                   : 1
 | + Identical stations                  : false
 |                                                                   
=|== Initial memory usage
 |                                                                   
 | + CUDA device 0 [NVIDIA A100-PCIE-40GB] memory is 1.0% (418/40536 MiB) used.
 | + CUDA device 1 [NVIDIA A100-PCIE-40GB] memory is 1.0% (418/40536 MiB) used.
 |                                                                   
=|== Starting simulation...
 |                                                                   
 |   - Chunk  2/16, Time 1/1, Channel 1/1 [Device 1]
E|                                                                   
E|== ERROR: Run failed with code 209: no kernel image is available for execution on the device.
E|                                                                   
 |                                                                   
=|== OSKAR-2.7.6 ending at 2022-03-04, 03:35:30 (GMT).
 |                                                                   

My slurm script to run OSKAR is:

#!/bin/bash
#SBATCH --constraint=A100
#SBATCH --time=01:00:00

runoskar() {
    singularity exec --bind /share/nas/zchen/skalow/ --nv /share/nas/zchen/OSKAR/OSKAR-2.7.6-Python3.sif $1 $2
}

runoskarpy() {
    singularity exec --bind /share/nas/zchen/skalow/ --nv /share/nas/zchen/OSKAR/OSKAR-2.7.6-Python3.sif python3 $1
}


echo ">>> start"

echo ">>> run oskar beam"
runoskar oskar_sim_beam_pattern beam.ini >& log/beam.log

echo ">>> finish"

Hi,

I have encountered a problem while creating a Measurement Set (MS). OSKAR by default sets the telescope name to "OSKAR 2.8.3" in the MS and there is no setting available where one can put a custom name instead of the default ( P.S : atleast I did not find ). Please can you guide me on how to enter a custom name ( say, "XYZ") to the telescope name instead of the default while creating the MS?

Thank you very much in advance.

With kind regards,
Sonia

We are using OSKAR to inject custom beams using CST file and "aperture array" as station type in the settings. I am reproducing visibility simulation using station type 'Gaussian' and "aperture array" by passing a Gaussian beam from a CST file.
We do 2 simulations, one with "Gaussian" station type and the second with "aperture array" with CST file for Gaussian beam. Both simulation is done with tracking mode. I am using oskar_sim_beam_pattern for the beam simulation and generating beam image in the equatorial frame. The station type 'Gaussian' works fine.
The size beam is also implemented is fine in the aperture array. However, the pointing direction of the beam is somehow not centred in the given ra,dec position. The direction of beam center is somehow depended on the time of the day and changes in ra mostly. For the station type 'Gaussian' is working perfectly with the correct beam center. The VLA beam also works fine.

Can input sky model for OSKAR 2.7.1 be two files together-an ascii and a fits file? And how can they be set in the simulator?

Using:

• cmake 3.7.1
• macOS 10.12

the following error prevents the build from proceeding. Casacore is an optional dependency so this behaviour is unintended.

-- A library with LAPACK API found.
CMake Error at cmake/modules/FindCasaCore.cmake:64 (get_filename_component):
  get_filename_component called with incorrect number of arguments
Call Stack (most recent call first):
  cmake/oskar_dependencies.cmake:9 (find_package)
  CMakeLists.txt:42 (include)


-- Could NOT find CASACORE (missing:  CASACORE_INCLUDE_DIR)

Is there a way to input diffuse radiation? The GSM data file produced from the software obtained from http://space.mit.edu/angelica/gsm/ cannot be obtained since it is showing

The content you were looking for could not be found

Hi Team!

I want to start by thanking your efforts on the OSKAR simulator. Especially the OpenCL support, it was great! However, I'm getting a strange issue with OSKAR (2.7.6 via dmg installation) on macOS with OpenCL.

I am using an iMac (27 inches, 2020, macOS Catalina 10.15.7) with AMD Radeon Pro 5700 XT graphics. The simulation did seem to run well with no error or anything, but the cleaned image from the ms files is empty. Detailed info can be found at here.

I checked the output visibility ms file and find out that the visibility has an all zero data column. Please see the image attached below.

I have used other macs to run the same task with no issue at all. Everything is quite normal. The machine info, OSKAR initial file, and logs can be found at here and here.

I have tried to pinpoint the issue, but haven't got anyway for the last few days. Could you please help me out with this issue?

I'm getting weird linking errors when I have (seemingly un-related) ~/miniconda3 directory.
An existing ~/miniconda2 seemed ok.

After cmake, the make commmand then results in:

Scanning dependencies of target ms_test
[ 79%] Building CXX object oskar/ms/test/CMakeFiles/ms_test.dir/Test_MeasurementSet.cpp.o
[ 79%] Linking CXX executable ms_test
//home/teuben/miniconda2/lib/libmkl_intel_lp64.so: undefined reference to mkl_lapack_ao_ssyrdb' //home/teuben/miniconda2/lib/libmkl_intel_lp64.so: undefined reference to mkl_lapack_ao_dsyrdb'
collect2: error: ld returned 1 exit status
oskar/ms/test/CMakeFiles/ms_test.dir/build.make:108: recipe for target 'oskar/ms/test/ms_test' failed
make[2]: *** [oskar/ms/test/ms_test] Error 1
CMakeFiles/Makefile2:719: recipe for target 'oskar/ms/test/CMakeFiles/ms_test.dir/all' failed
make[1]: *** [oskar/ms/test/CMakeFiles/ms_test.dir/all] Error 2
Makefile:160: recipe for target 'all' failed
make: *** [all] Error 2

I am trying to run OSKAR 2.7.1 using a sky catalog that contains around 1450 sources, and I have arranged the source catalog exactly as the sample "sky.osm" file included in the example simulations. However for every run I am doing it is showing "WARNING: Gaussian ellipse solution failed for 1429 sources. These will be simulated as point sources", and the imaging procedure is subsequently failing to show any of the extended structures in the field. I tried by reducing the number of sources to just 10 sources with the same result.
Can you kindly suggest some cause for this? And any changes that I need to do in my simulation?
Thanks...

I've tried using the imager with the w-projection algorithm. I've observed a core dump in the function "oskar_convert_station_uvw_to_baseline_uvw" located in oskar/convert/src/oskar_convert_station_uvw_to_baseline_uvw.c

After investigating a bit, it appears that the convert function is called from line 233 of file "private_imager_read_coords.c" without a proper allocation for arrays uu, vv, ww.
I've added those lines at line 226 of this file :

        /* Need to allocate uu, vv and ww */        
	oskar_mem_realloc(uu, num_baselines*num_times, status);
	oskar_mem_realloc(vv, num_baselines*num_times, status);
	oskar_mem_realloc(ww, num_baselines*num_times, status);

It seems to work better now.
I'm not sure whether this is the correct place to allocate. Other calls to the convert function always have the uu, vv and ww arrays already allocated.

Regards,
Simon Rokicki

Hi,

I have met a problem when I try to write data into a CASA Measurement Set.

I wrote a python code based on the example on https://fdulwich.github.io/oskarpy-doc/measurement_set.html. But I find that the outputted Measurement Set has different UVW and DATA values in some baselines, comparing with the input. To be more precise, please see the attached document.
Bug report.pdf

I use .mat files as inputs to input UVW and DATA, and the output is a .ms Measurement Set.

Could anyone please give me some advice about this issue?

Thank you very much for your time and help.

Regards,

Xiao

When exporting to MeasurementSet, the antenna positions seem to be given in the ITRF frame relative to the array center. This is incorrect (or at least unusual), the column ANTENNA::POSITION should contain ITRF locations of the station centers in an absolute way. Could be fixed by just adding the coordinates of the array center.

We aim to supply custom beams to the fit_element app. For this purpose, we want to create custom CST files. Do you have an example CST file with some documentation? We know that the cst file requires a specific format based on amplitude and phases in theta and phi coordinates. We have an electromagnetic antenna beam patterns in the farfield approximation (XY). What is the convection you use for defining theta-phi coordinates for cst file?
@rohitcbscient

With a rather straightforward settings file, I get the following error when trying to use oskar_sim_interferometer.

 |   - Time  16/360, Chunk 1/1, Channel  9/10 [Device 39, 0 sources]
 |   - Time  16/360, Chunk 1/1, Channel 10/10 [Device 39, 0 sources]
 | + Block  1/45 (  2%) complete. Simulation time elapsed: 0.030 s
terminate called after throwing an instance of 'casa::TableInvColumnDesc'
  what():  Invalid description of column FLAG: column already exists

Simulation into an OSKAR visibility file runs through to the end.

Hello,
While running a simulation with oskarpy using a python script, the simulation is taking only one sky model. The inputs are given as :
sim.set_sky_model('point_source.osm')
sim.set_sky_model('diffuse_source.fits')
It is taking only the 2nd one, i.e. sim.set_sky_model('diffuse_source.fits'). I tried switching places to see if there is something wrong with the input file, but as it turned out it is only taking whatever is the last sim.set_sky_mode. What is the way in which I can input both?

In the new version (2.7.0) the sample data for 2.6.1 is not working, it is running into error in the telescope directory. The exact error is "ERROR: Run failed with code -28: a configuration file is missing from the telescope model directory tree."
It is with the zipped data available with the 2.6.1 release.
What file is missing in the telescope folder of the previous version? And since the official site is not working where can I get the new sample data?
Thanks in advance.

Is it possible to give as the simulator input a fits image with coordinates specified in comoving Mpc instead of angular coordinates? The image has shape (200,200,300) with the 3rd axis as the frequency axis, and the first 2 as cMpc. The input as a text file (sky.osm) requires RA and Dec to be specified, but in the case of this image, they are not specified.

MSv3 describes a way to embed the element locations in the Measurement Set. I've made a small python script that adds this metadata based on the telescope.tm input to OSKAR. Maybe it would be good to embed this functionality in OSKAR, so that the MeasurementSet is more self describing.

One issue is that the MSv3 is not finalized, nor implemented. I've adopted part of the metadata scheme to embed the element data, so maybe we can call it MSv2.9 or so. (The main thing I've left out is explicit keys; so in the scheme I've adopted it's assumed a row number in the PHASED_ARRAY table matches the row number in the ANTENNA table.

I'll share the script here when I've validated it a bit more.

Hi,
If I set 'normalise_beam_at_phase_centre : 'True' while running a simulation a simulation using a python script, I get the error given in beam_err.txt. Has the option been disabled for the new oskarpy? Also, is there a way I can make all the beams unit (i.e. there should be no contribution of beams in the simulation?).
beam_err.txt

probably c++11 related

$  make
[  0%] Building CXX object extern/gtest-1.7.0/CMakeFiles/gtest.dir/src/gtest-all.cc.o
In file included from /usr/include/c++/7/ext/string_conversions.h:41:0,
                 from /usr/include/c++/7/bits/basic_string.h:6349,
                 from /usr/include/c++/7/string:52,
                 from /usr/include/c++/7/bits/locale_classes.h:40,
                 from /usr/include/c++/7/bits/ios_base.h:41,
                 from /usr/include/c++/7/ios:42,
                 from /usr/include/c++/7/ostream:38,
                 from /packaging/kern/packaging/build/oskar/extern/gtest-1.7.0/include/gtest/gtest.h:55,
                 from /packaging/kern/packaging/build/oskar/extern/gtest-1.7.0/src/gtest-all.cc:39:
/usr/include/c++/7/cstdlib:75:15: fatal error: stdlib.h: No such file or directory
 #include_next <stdlib.h>
               ^~~~~~~~~~

Is it possible to dump the W convolution kernel when W-projection gridding with the imager as a 3D FITS file? Also some explanation of how the kernels are generated from the input visibilities would also be of great help.

output as follows:

Note: Google Test filter = SkyModel.set_source
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from SkyModel
[ RUN      ] SkyModel.set_source
/Users/bmort/code/oskar_new.git/sky/test/Test_Sky.cpp:949: Failure
Value of: status
  Actual: 0
Expected: (int)OSKAR_ERR_OUT_OF_RANGE
Which is: -11
[  FAILED  ] SkyModel.set_source (169 ms)
[----------] 1 test from SkyModel (169 ms total)

[----------] Global test environment tear-down
[==========] 1 test from 1 test case ran. (169 ms total)
[  PASSED  ] 0 tests.
[  FAILED  ] 1 test, listed below:
[  FAILED  ] SkyModel.set_source

 1 FAILED TEST

Hi,
This is not an issue, rather a query. In the telescope settings, there is an option set_station_type wherein the type of station beam can be set. I want to know that if this is set to "Array”, what exactly is the primary beam function which is used? I understand that giving "Gaussian" gives a time invariant Gaussian beam for each station, but what is the functional form for "Array"? Also, if, suppose, I want to use a primary beam modelled as, say a 3rd order polynomial, is there a way to incorporate that into the OSKAR during simulation?

Hi,

First of all, my system is a MacBook Pro (16-inch, 2021) with Apple M1 Pro, running macOS Monterey v12.1. Python version 3.8.9. I've followed all installation instructions found here (OSKAR was installed using its .dmg), including assigning relevant library paths from the installed App to the OSKAR_INC_DIR and OSKAR_LIB_DIR environmental variables (located within the /Applications/OSKAR.app/Contents/Resources directory).

Currently, I'm trying to use python bindings for the OSKAR library. Unfortunately I run in to the following issue:

simon@macbook:~ % python3 -c "import oskar;oskar.SettingsTree('oskar_sim_beam_pattern')"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "~/Library/Python/3.8/lib/python/site-packages/oskar/settings_tree.py", line 138, in __init__
    raise RuntimeError("OSKAR library not found.")
RuntimeError: OSKAR library not found.

Looking at settings_tree.py, this error is raised if _settings_lib is None. Controlling this are the following import statements within a try/except block (lines 33-38 of settings_tree.py):

try:
    from . import _settings_lib
    from . import _apps_lib
except ImportError:
    _settings_lib = None
    _apps_lib = None

So, to investigate further, I looked at manually importing _settings_lib after changing my working directory to ~/Library/Python/3.8/lib/python/site-packages/oskar/, resulting in the following terminal output:

simon@macbook:oskar % python3 -c "import _settings_lib" 
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ImportError: dlopen(~/Library/Python/3.8/lib/python/site-packages/oskar/_settings_lib.cpython-38-darwin.so, 0x0002): symbol not found in flat namespace '__ZN5oskar12SettingsTree4freeEPS0_'

Looking at the ImportError that was raised, I found the following question on StackOverflow, the accepted answer for which states:

"...The problem was caused by mixing objects that compiled with libc++ and object that compiled with libstdc++."

Any ideas what my next steps should be? Thanks in advance!

Hi,
I was wondering if there is any way we can input a fits image cube as sky model with spectral index variation while running a simulation with oskarpy. If I am using a sky.osm file, there is a specific spectral index column where I am specifying it with values like say -0.8. But is there any way I can do it for a fits image as input sky model?

I have installed the latest version in a CentOS 7 machine. There was no error during installation, but on running ctest [--verbose] the output is :
Test project /home/newuser/OSKAR/build
Start 1: binary_test
1/13 Test #1: binary_test ......................***Failed 0.00 sec
Start 2: settings_test
2/13 Test #2: settings_test .................... Passed 0.01 sec
Start 3: convert_test
3/13 Test #3: convert_test .....................***Failed 0.07 sec
Start 4: correlate_test
4/13 Test #4: correlate_test ...................***Failed 0.23 sec
Start 5: imager_test
5/13 Test #5: imager_test ...................... Passed 0.92 sec
Start 6: jones_test
6/13 Test #6: jones_test .......................***Failed 0.16 sec
Start 7: math_test
7/13 Test #7: math_test ........................***Exception: SegFault 2.24 sec
Start 8: mem_test
8/13 Test #8: mem_test .........................***Failed 0.10 sec
Start 9: sky_test
9/13 Test #9: sky_test .........................***Exception: SegFault 0.33 sec
Start 10: station_test
10/13 Test #10: station_test .....................***Exception: SegFault 0.15 sec
Start 11: telescope_test
11/13 Test #11: telescope_test ...................***Exception: SegFault 0.16 sec
Start 12: utility_test
12/13 Test #12: utility_test .....................***Failed 0.09 sec
Start 13: vis_test
13/13 Test #13: vis_test .........................***Failed 0.05 sec

15% tests passed, 11 tests failed out of 13

Total Test time (real) = 4.53 sec

The following tests FAILED: 1 - binary_test (Failed) 3 - convert_test (Failed) 4 - correlate_test (Failed) 6 - jones_test (Failed) 7 - math_test (SEGFAULT) 8 - mem_test (Failed) 9 - sky_test (SEGFAULT) 10 - station_test (SEGFAULT) 11 - telescope_test (SEGFAULT) 12 - utility_test (Failed) 13 - vis_test (Failed) Errors while running CTest

Also, the command $ oskar yeilds bash: oskar: command not found...

Hi I am trying to port the project to the ROCm platform(AMD GPU), and found there is issue when compile the code by ROCm hipcc. HIPCC does not support compile c source file by c++ compiler. Maybe I have to modify many .c files to .cpp with gloabl functions, but will break the portability...

Do you have plan to support ROCm(https://rocmdocs.amd.com/en/latest/Programming_Guides/HIP-GUIDE.html) ?
Thanks!

Is it possible to simulate an array which has dishes (like HERA) instead of dipoles?