Journal of the Korean Astronomical Society https://doi.org/10.5303/JKAS.2018.00.0.1

00: 1∼ 99, 2018 July pISSN: 1225-4614 · eISSN: 2288-890X

c©2018. The Korean Astronomical Society. All rights reserved. http://jkas.kas.org

VLBI NETWORK SIMULATOR: AN INTEGRATED SIMULATION TOOL FORRADIO ASTRONOMERS

Zhen Zhao1, Tao An1,2, and Baoqiang Lao1

1Shanghai Astronomical Observatory, Chinese Academy of Sciences, 200030 Shanghai, China;2Key Laboratory of Radio Astronomy, Chinese Academy of Sciences, 210008 Nanjing, China;

Received —; accepted —

Abstract: In this paper we introduce a software package, the Very Long Baseline Interferometry Net-work SIMulator (VNSIM), which provides an integrated platform assisting radio astronomers to designthe Very Long Baseline Interferometry (VLBI) experiments and evaluate the network performance witha user-friendly interface. Though VNSIM is motivated to be designed for the East Asia VLBI Network,it can also be expandable to other VLBI networks and generic interferometers. The software package notonly integrates the functionality of plotting (u, v) coverage, scheduling the observation, and displaying thedirty and CLEAN images, but also further extends to add new features such as the sensitivity calcula-tion of a certain network and multiple-satellite space VLBI simulations which are useful for future spaceVLBI mission. In addition, VNSIM provides flexible interactions on both command line and graphicaluser interface and offers friendly support for log report and database management. VNSIM also supportsmultiprocessing accelerations, enabling to handle large survey data. To facilitate the future developmentand update, each simulation function is encapsulated in different Python module allowing for indepen-dently invoking and testing. In order to verify the performance of VNSIM, we have carried out varioussimulations and compared the results with other simulation tools. All tests show good consistency.

Key words: journals: simulation: VLBI, Interferometry, image simulation

1. INTRODUCTION

East Asia Very Long Baseline Interferometry (VLBI)Network (EAVN, An et al. 2018), consisting of 21 ra-dio telescopes from China, Japan and South Korea, hasbeen astronomically operational since mid -2018. Thediverse sub-array configurations and frequency setupsof the EAVN make it cover a wide range of researchareas including astronomical masers (e.g. hydroxyl,methanol, water and SiO masers in Galactic objects andextragalactic megamasers as well), jets of active galacticnuclei, pulsars and transients (e.g. supernovae, gamma-ray bursts), space exploration and tracking, astrometryand geodesy. The EAVN is expected to promote theregional collaborations in East Asia. Such academiccollaborations will offer great opportunities to expandthe discovery fields in astronomy and space science, andform a successful model of international academic col-laborations with a sustainable operation scheme.

Along with the formal operation of the EAVN, aset of user assistance software, having functions of eval-uating the network performance, would be helpful forusers to prepare proposals and necessary for expand-ing the user community. The diversity and versatilityof the EAVN configurations demand such a tool to besufficiently flexible and expandable.

Currently, there are a number of auxiliary toolsused in the VLBI community for different simulationpurposes. SCHED1 designed by the National Radio As-

Corresponding author: T. An, antao@shao.ac.cn1http://www.aoc.nrao.edu/software/sched/

tronomy Observatory (NRAO) of the US, is commonlyused for scheduling VLBI observations. It supportsplotting the (u, v) coverage of a given array and timerange, the corresponding dirty beam, and displayingthe telescope elevation angle change with time which as-sists scientists to choose proper telescopes. SCHED canalso show multiple source scans in time sequence, whichfacilitates the arrangement of time blocks to optimizethe (u, v) coverage within the allocated time period.Since SCHED is configured through importing a pre-defined ’key’ file, any parameter change or setup adjust-ment requires restarting the program. Difmap (Shep-herd 1997) is largely used to analyze VLBI data, asa part of the Caltech VLBI Analysis Programs (Read-head & Wilkinson 1978; Readhead et al. 1980). Itis characteristic of interactive operations with editing,hybrid imaging, self-calibration and automated pipelinecapacities. Difmap provides both scripting languageand commands for operations. Although these softwarepackages have been and are still widely used in VLBIcommunity, the architecture and algorithms of SCHEDand Difmap are slightly outdated. The aperture syn-thesis simulator (APSYNSIM) (Martı-Vidal 2017) isa Python-based software package, providing an interac-tive tool to visualize the aperture synthesis and performeducational level simulations which are very useful fornon-radio astronomers.

Besides these auxiliary tools for ground-basedVLBI networks, some other software packages were de-veloped with additional functions specially to adapt to

1

arX

iv:1

808.

0672

6v1

[as

tro-

ph.I

M]

21

Aug

201

8

2 Zhao, An, & Lao

space VLBI. These include: the Space VLBI AssistanceSoftware (SPAS) developed by the Satellite GeodeticObservatory of the Institute of Geodesy (Frey et al.1998) and Fakesat (Smith et al. 2000) for VLBI SpaceObservatory Programme (Hirabayashi et al. 2000) pro-posal preparation; Astronomical Radio InterferometerSimulator (ARIS) developed by the Japan AerospaceExploration Agency (Asaki et al. 2009, 2007) forJapan’s second-generation space VLBI project VSOP2,which adds some new functions to assess the impactsof a variety of error sources on the image quality; theFakeRat software package developed by Astro SpaceCenter of the Lebedev Physical Institute (Zhuravlev2015) and used for the Russia RadioAstron (Kardashevet al. 2013) space VLBI mission. Since these worksare dedicate for specific space missions, thus their gen-eral versatility is relatively limited. Shanghai Astro-nomical Observatory of China once proposed the spacemillimeter-wavelength VLBI array (SMVA) programmewhich for the first time involves two space satellites on-board 10-m radio telescopes operated at the highest fre-quency of 43GHz (Hong et al. 2014). In order to sup-port this SMVA, a simulation software (An et al, 2016)was designed to provide fundamental space-ground andspace-space VLBI (u, v) coverage simulations.

The major functions of these existing softwarepackages which are adapted for specific applicationshave been discussed and compared in An et al, (2016).For these tools to work for a new VLBI network such asthe EAVN, major modifications have to be made. More-over, a module-independent, highly scalable, flexible,and user-friendly solution, combining the advantages ofeach individual software and data-modifying flexibility,is highly desirable. For this purpose, we developed anew software package, VLBI Network SIMulator (VN-SIM), which integrates most commonly-used simulationprograms and offers assistance for radio astronomers.It is a cross-platform Python-based software packagewith high scalability and reusability. Each functionhas been separately designed and independently imple-mented for the sake of future extension. In addition tocommon simulation functions, VNSIM also augmentssome extended functions such as displaying all-year-round (u, v) plots to track the space VLBI (u, v) cov-erage changes due to the satellite orbit precession andpresenting (u, v) plots of multiple sources along withtheir dirty maps so as to evaluate the imaging perfor-mance. The data of stations and sources are dynam-ically managed by the SQLite database, and the pa-rameter configuration can either be saved and loadeddirectly or be set up through the interactive graphicalinterfaces. The data processing with multiple process-ing accelerations is also adopted, significantly enhanc-ing the execution performance of large survey data. Thesimulations of EAVN are performed in a straightforwardway. Although VNSIM is initially designed for EAVN,it is naturally adapted to other VLBI networks or othergeneric interferometers.

The remaining part of this paper is organized asfollows. Section 2 describes the overall designing con-

cept of VNSIM. Details of the main functions are pre-sented in Section 3. In Section 4, we demonstrate someexamples of the major functionality of VNSIM and alsocompare the experimental results with other tools. Asummary is given in Section 5.

2. THE VNSIMThe VNSIM is programmed in Python language andcan be implemented on all common platforms, Win-dows, Linux and Mac OS. The source code is availableon Github2. In this section we describe the overall de-sign and framework of the VNSIM.

2.1. Overall DesignTo fully match the requirements of simulating EAVNtasks, we mainly propose the following design consider-ations:

• Highly scalable and reusable. Each function clusteris independently encapsulated into a single Pythonfile to facilitate the testing and future extensions.In this sense, the design obeys the Model ViewController (MVC) architectural pattern. We dividethe development into three interconnected parts,i.e., an SQLite database for data management, agraphical interface designed by Tkinter for user in-teractions, and functional clusters for parametercalculations.

• Dynamic database management. The informa-tion of VLBI stations and astronomical sources isstored and managed by an SQLite database, whichis implemented through a graphical user interface(GUI), enabling to add, delete and modify exist-ing data records in a convenient way. It also allowsfor appending self-defined fake stations for the sim-ulation purpose. The related database currentlyconsists of four tables: source table, VLBI stationtable, satellite table and telemetry station table.The first two are used for ground- and space-basedVLBI network simulation, and the last two are onlyneeded for space VLBI mission. The data can ei-ther be inserted through the GUI record by recordor be directly added as a whole through importingexternal files in .txt or .csv formats.

• User-friendly interface. Both graphic and com-mand line interfaces are supported in VNSIM fordifferent-level users with different usage habits. Ineither mode, the parameter configurations can besaved and loaded conveniently, and all the result-ing images can be saved in a variety of image for-mats. Besides, indicative logging information anddialogues are designed to help users to track therunning status of VNSIM.

• Running performance. Unlike SCHED, VNSIM iscompatible with generic interferometers, such asVery Large Array (VLA) and the Square Kilome-tre Array (SKA). VNSIM contains some new ad-vanced functions which are suitable for (u, v) cov-erage simulations of large scale sky survey of these

2https://github.com/ZhenZHAO/EAVNSIM

VLBI Network SIMulator 3

GraphicInterface

Menu & Help

Cluster 1:UV functions

Cluster 3:Observation

Survey

Cluster 2:Imaging &

Clean

Cluster 5:ParameterCalculation

GeneralSettings

FunctionalInterface

Cluster 6:Database

Operations

AllYearUV

All SkyUV

BasicUV

MultiSource

UV

SourceModel

Dirtymap

Dirtybeam

CleanImage

Stationselection

Imagingsettings

ParameterSettings

Fits filesize

Timesmear

noise

BWsmear

Update

View

Picture/ResultPanel

Logging &Status

Az-El

SkySurvey

UV

Radplots

Cluster 4:Real DataProcess

SettingPanel

Figure 1. The software framework of VNSIM. This diagram shows the example of the graphical user interface. Detaileddescription of the main functions are given in Section 3. VNSIM currently comprises six function clusters including 14sub-functions in total.

4 Zhao, An, & Lao

Figure 2. The main window of VNSIM. It consists of threefunctions zones: configuration zone, log zone, display zone.The plot in the top-right panel shows the EAVN (u, v) cov-erage of 0134+329. More (u, v) coverages can be invokedby changing the source in the configuration panel (ObserveSettings). The bottom-right panel shows the simulations ofthe EAVN (u, v) coverages of sources at different RA andDec coordinates. The red circle marks the position of theSun in the plane of the sky at the observation epoch.

many-element interferometers, for long-term eval-uation of the (u, v) coverage of main targets inthe space VLBI. Undoubtedly these functions in-volve huge calculations and produce large-size im-ages, typically consuming considerable computingresources. To address these problems, we considerthe compute-intensive accelerations via multipleprocessing. In addition, multiple threads are cre-ated to maintain the graphical interface and func-tional calculation separately.

• Functional designs. To deliver a complete solu-tion, VNSIM aims to integrate the most commonsimulation functions, including the (u, v) coverageplots of a single source and multiple sources, all-sky(u, v) plots of a large sample, and all-year-round(u, v) coverage changes of a single source with theprecession of satellite orbits; image simulations ofdirty beam, dirty map and CLEAN map; schedul-ing setups showing azimuth and elevation changesof sources in VLBI stations with observing time;the visibility amplitude of the simulated data as afunction of the projected (u, v) distance. Besides,VNSIM also enables us to estimate the VLBI net-work performance, such as the sensitivity estimatewith the given time range and data rate.

2.2. Software FrameworkAs shown in Figure 1, the framework of VNSIM mainlyconsists of six function clusters, parameter configura-tion and other common GUI components such as statusbar, menu bar, and logging information. The six clus-ters answer for four fundamental simulation functions,including (u, v) coverage plotting, imaging, observation

survey and simulation of visibility data (clusters 1-4),and two additional functions with parameter evaluationand database management (clusters 5-6). As mentionedbefore, the programming of these function clusters aremutually independent, so each of them can be directlyinvoked with appropriate configurations and run sepa-rately.

As per Figure 2, the main area of the VNSIM GUIis divided into three parts: result panel, configurationpanel and logging panel. Four main simulation func-tions locate at different tabs on the result panel, whiletwo additional functions can be accessed through thetool menu. The configure panel comprises of five dif-ferent kinds of configuring items, and only one of themis visible by default with others folded up. The log-ging panel shows the currently running information inreal-time to help track the VNSIM status. The wholeinterface is resizable and can be adjusted according tothe display window on different computers.

3. MAJOR FUNCTIONS

VNSIM currently comprises six function clusters includ-ing 14 sub-functions in total. Those functions are de-scribed in detail below.

3.1. Simulation Functions3.1.1. (u, v) coverage plotting

The projection of the separation between any twoantennas in a VLBI network (so-called baseline, B)perpendicular to the direction of the obesrved radiosource can be decomposed into east–west and north–south components, represented by u and v, respectively.Given the position (xλ, yλ, zλ) of baseline vector Bλ in(X,Y, Z) coordinate system, the components (u, v, w)are obtained by (Thompson et al. 2007),uvw

=

sinHs cosHs 0− sin δs cosHs sin δs sinHs cos δscos δs cosHs − cos δs sinHs sin δs

xλyλzλ

,

where Hs and δs are the hour angle and declinationof the source position, and (xλ, yλ, zλ) are in unit ofthe observing wavelength. Each baseline in continuousobservations of a radio source will project a portion ofan arc of an ellipse on the (u, v) plane, while creates asingle curve line in the (u, v) coverage. Multiple stationsof VLBI networks form a set of baselines. They togethercreate the usually seen plot in the (u, v) spacing.

In addition to the above simplest (u, v) coverageinvolving a single source in a certain time period, threeadditional functions of (u, v) coverage are implementedin VNSIM: (i) all-year-round (u, v) plotting can createtwelve (u, v) coverage plots generated through simulat-ing the first-day observation of each month, which is de-signed to observe the coverage evolution of space VLBIdue to the satellite orbit precession; (ii) by evenly divid-ing the whole sky into 5×6 blocks, all-sky (u, v) plottingfunction roughly shows the survey ability of selectedstation combinations at a given observing time. This is

VLBI Network SIMulator 5

CLEANAlgorithm

Input parameters: Threshold, Gain, Iteration numberInput images: Dirty beam, Dirty mapOUTPUT: Density image, Residual image

maximum of absolute deviationof residual image < Threshold

Search for peakin Residual image

Initialize Residual image with Dirty mapInitialize Density image with all zeros

Current Iterations > Iteration number

translate Dirty Beamcenter to location of

peak

subtract Gain timesDirty Beam fromResidual Image

enter componentlocation and amplitude

into Density file

NO

NO

EXITAlgorithm

YES

YES

Figure 3. Flowchart of the CLEAN loop.

useful for scheduling large-sample survey. In the exam-ple shown in Figure 2 bottom-right panel, the δ = 0

source is marginally seen by the EAVN; (iii) Similarly,the function of multiple-source (u, v) plotting can dis-play the (u, v) coverage of selected multiple sources toassess the performance of a given VLBI network. Thisis invoked by clicking the button MultiSource UV.

3.1.2. ImagingThe intensity distribution Iν of a radio source at ob-serving frequency ν can be estimated by the two-dimensional Fourier transformation of the spatial co-herence function Vν (Thompson 1999),

Iν(l,m) =

∫∫Vν(u, v)e2πi(ul+vm)dudv, (1)

where (l,m is the sky position of the sources. However,in practice Vν is not a continuous function, but is sam-pled at particular positions on the (u, v) plane. Then,

we actually have

IDν (l,m) =

∫∫Vν(u, v)S(u, v)W (u, v)e2πi(ul+vm)dudv,

(2)where IDν (l,m) is the so-called dirty image; S(u, v) andW (u, v) represent the adopted sampling and weightingfunctions, respectively. Besides, the observed intensitydistribution is often expressed as a convolution of intrin-sic dirty image and the synthesized beam or the pointspread function of the interferometer Bθ,

IDν (l,m) = Iν ∗Bθ, (3)

where Bθ is given by

Bθ(l,m) =

∫∫S(u, v)W (u, v)e2πi(ul+vm)dudv. (4)

Given a selected source model (point source model,Gaussian model, etc) and proper observation configu-rations in simulations, VNSIM can generate the corre-sponding (u, v) coverage plot, dirty beam Bθ, and dirtyimage IDν . Referring to (Hogbom 1974), we also in-tegrate the CLEAN algorithm in VNSIM to performquick inspection of the image quality. As shown in Fig-ure 3, the deconvoluted (CLEAN) algorithm can passtwo controlling parameters, i.e., the Threshold and It-eration Number which control how to stop the iterationprocedure. Eventually, adding the group of CLEANcomponents back into the final residual image producesthe simulated CLEAN image. As mentioned before,this function is useful for evaluating the performanceof a new VLBI network, or select in optimized networkconfiguration before site selection. In future, more re-alistic noise model will be considered in order to makethe simulation close to the practical condition.

3.1.3. Observation

In the third cluster (Figure 1), observation survey, weimplement two kinds of simulations. The first one is toplot the azimuth and elevation angles at selected sta-tions as a function of time within the observation du-ration. It can clearly manifest the best observing timeduration of selected stations for a specific target source.The other is on the sky survey function, which aims tocheck whether different parts of the sky are visible bya VLBI network or not at a certain epoch. Specifically,it visualizes the number of visible observing stations indifferent color patch at each divided sky position andalso the position of the Sun and Moon during the ob-serving time.

3.1.4. Real Data Process

VNSIM is not only a simulation and demonstrationtool, but also allows for processing real observing data.For example, by extracting the station position andobserved visibility data, VNSIM can draw the plotsof (u, v) coverage of the real observing data and thecorresponding dirty beam. As a useful demonstration,

6 Zhao, An, & Lao

Figure 4. The GUIs of parameter evaluation of VNSIM(top panel) and the EVN Calculator (bottom panel). Thedatabase of VNSIM is dynamically managed, easy to mod-ify.

the figure of visibility amplitude – projected (u, v) dis-tanceıs also provided for astronomers to observe the vis-ibility changes, gain a rough knowledge of the sourcestructure (resolved or unresolved), and diagnose baddata points.

3.2. Additional Functions3.2.1. Parameter Evaluation

Referring to the EVN Calculator3, VNSIM providesthe following parameter calculation: the image ther-mal noise Ω, bandwidth-smearing-limited field of viewFbw, time-smearing-limited field of view Ftime, and anestimate of the FITS file size C.

First of all, the image thermal noise (in unit of Jybeam−1) can be calculated by

Ω =1

η

Se√rTobs

2

, (5)

where r and Tobs represent the data rate in bit per sec-ond (bps) and on-source time in minute, respectively. η

3http://www.evlbi.org/cgi-bin/EVNcalc

is a constant. Se is given by

Se = (

√√√√1

2

N∑i=1

N∑j=1,j 6=i

1

SEFDi × SEFDj)−1, (6)

where SEFDi denotes the system equivalent flux den-sity of telescope i in Jy.

The field of view is limited by the bandwidthsmearing effect and is expressed by

Fbw = κ1Nch

LblBWsub, (7)

where Nch, Lbl, and BWsub denote the number of chan-nels, maximum baseline length, and the bandwidth ofsub-band, respectively. κ1 is a constant, which usuallyequals to 49500. Fbw is in unit of arcsecond.

Similarly, the field of view limited by time-smearing effect is given by

Ftime = κ2LwaveLbl Tint

, (8)

where Lwave, Lbl and Tint represent the wavelength,maximum baseline length, and integration time of cor-relation, respectively. Constant κ2 usually equals to18560. Note that these values are calculated by tak-ing into account for 10% loss in the response of a pointsource.

To evaluate the size of resulting FITS file, the fol-lowing equation is applied.

C = σN2staNpolNsubNch131072 × 3600

TobsTint

(9)

where σ represents a constant. Nsta, Npol, and Nsub,Nch denote the number of stations, polarization, sub-bands and channels per sub-band, respectively. Tobsand Tint are the total observation time and the integra-tion time of correlation, respectively.

The parameter evaluation interface of the VNSIMis shown in the top panel of Figure 4. The commonly-used telescopes are included in the default list. Othertelescopes or new ones can be easily added via thedatabase. The configuration parameters can be selectedfrom the drop-down menu. The on-source timetextframe accepts manual input. Clicking the button RUN,the calculated parameters are shown in the bottomlines. For comparison, we investigated the parametercalculation of the EVN Calculator. As Figure 4 shows,typical stations of EVN at K band were selected andthe four parameters were calculated and showed on theGUI. The same configurations are applied to the VN-SIM calculator too. It is obvious that the calculatedresults in VNSIM are consistent with those obtained bythe EVN Calculator. Note that the slight difference ofthe image thermal noises derived from the VNSIM andEVN Calculator is due to that SEFDs of some tele-scopes in these two databases have minor difference.

VLBI Network SIMulator 7

Figure 5. The GUI of database editor providing a high qual-ity, visual and open source tool to create, delete and editdatabase files. Users can either edit the records via theGUI, or import a telescope or source table from an externalfile.

Figure 6. The comparison of (u, v) coverages of EAVN de-rived from VNSIM (left, the present paper) and from theliterature (right, An et al. (2018)). For the comparisonpurpose, the inner 2,000 km region of the (u, v) coverageare displayed. They show good consistency.

3.2.2. Database Managements

Compared with other existing software packages, VN-SIM adopts a friendly graphic user interface, allowingfor easy access and operation. We selected SQLiteas the SQL database engine. It is self-contained,highly-reliable, open-source and also full-featured. Thegraphic editor of the database management is shown inFigure 5. The related database consists of four tablesthat are responsible for manipulating the correspond-ing data of the sources, satellites, VLBI stations andtelemetry stations, respectively. Taking the VLBI sta-tion table as an example, users can delete old records orinsert new records easily through the GUI operations.In addition to inserting new data record by record, userscan also import multiple records through loading an ex-ternal file containing well-formatted data.

4. EXAMPLE EXPERIMENTS

In this section, we take two main functions, (u, v) cov-erage and source imaging, as examples to demonstratethe operation of VNSIM.

4.1. (u, v) plotting of a single source(u, v) coverage is a basic visualization of evaluating theVLBI network performance and predicting the imagequality. We first use VNSIM to create the (u, v) cover-age plot of a single source, and compare with the exist-

Table 1Simulation Parameter Settings

Parameters Settings

General

Time 12 hScan length 5 minFrequency 22 GHz

SourceMain M87

Othersa 0202+319,0529+483,1030+415,1128+385,1418+546, etc.

StationbCVN Tianma, Urumqi,

SheshanJVN Gifu, Hitachi,

Kashima, Taka-hagi

KVN Sejong, Tamma,Ulsan, Yonsei

NRO NobeyamaVERA Iriki, Ishigaki-

jima, Mizusawa,Ogasawara

ImagingIteration 100

Gain 0.2Threshold 0.001

a VNSIM allows for processing multiple sources simultaneously.In the experimental examples, we take 10 bright and compactAGNs.b the stations are same as in An et al. (2018) for comparison.

ing software. The radio galaxy M87 is chosen as the tar-get. As VNSIM supports parallel processing of multiplesources, we also import ten more sources in the sourcelist. The multiple source (u, v) plotting is described inSection 4.2. The experiment parameters used in thissimulation are listed in Table 1. For comparison, theseparameters are same with those adopted in Figure 2 ofAn et al. (2018). The 22 GHz frequency, one of theoperational frequency bands in the first EAVN open-use session, is chosen. A total of sixteen telescopes areincluded in the simulation. The 12-hr full-track (u, v)coverages of M87 are shown in Figure 6. VNSIM ob-tained the exactly same (u, v) plot (Figure 6 left panel)with that shown in An et al. (2018). We note the (u, v)plot in An et al. (2018) was made by using the UVSIMsoftware (An et al, 2016), whose results have been com-pared with SCHED. The consistent results in Figure 6verify the accuracy of VNSIM and compatibility withother similar software tools.

4.2. (u, v) plotting of multiple sourcesOne of the new practical functions included in VNSIMis the multi-source (u, v) coverage plotting. It allowsfor calculating the (u, v) coverages of multiple sourceswith pre-defined observing configurations through one-shot running and either displays the results in the in-terface or saves in external files. Figure 7 demonstratesan example of the multi-source (u, v) plotting function.In this experiment, we calculated the (u, v) coveragesof ten radio-loud active galactic nuclei selected from

8 Zhao, An, & Lao

Figure 7. An example of (u, v) coverage plotting of multiple sources. This function is useful for large sample surveys. Thepresent figure shows 10 sources with the same VLBI network configuration in Table 1. Even more sources are supported byusing multiprocess acceleration.

Table 2Four Source Models

Relative RA Relative Dec Flux Density(mas) (mas) (Jy)

1 0.0 0.0 1.02 17.2 −22.7 0.53 −11.5 16.5 1.54 -18.5 - 9.5 1.5

Cheng et al. (2017). The observing frequency is 43GHz. The total observing period of each source in thesimulation is 12 h. The EAVN is used for the VLBI net-work configuration. As all these sources are at high dec-lination, they are visible by the EAVN in most time. Inparticular, 1823+568 and 1928+738 show almost circu-lar (u, v) coverages. Moreover, we should mention thatthe maximal source number is not limited to ten. Evenmore sources are allowed and the calculation can beaccelerated by multiprocess parallelization. This func-tionality is quite useful for large-sample survey. Besidesdisplaying the result plots in the GUI, the output canalso be exported to external ’.eps’ or other format files.

4.3. Dirty and CLEAN imagesVNSIM is not only a simple display tool of (u, v) cov-erage, but also a powerful software package enablingimage simulation that is visual evaluation of VLBI net-work imaging performance. To demonstrate this func-tionality, we created four fake sources in the experimentwhose parameters are listed in Table 2. As shown inTable 1, the same EAVN station and general settingsare also applied. The dirty beam and source modelsare displayed in the top panel of Figure 8. Source #1with a flux density of 1 Jy is a point source, locat-ing at the sky position RA= 03h19m48.160s and Dec=4130′42.10′′. Sources #2 and #3 are also point sources

and locate at the southeast and northwest directions ofSource #1. Their corresponding flux densities are 0.5and 1.5 Jy, respectively. Source #4 is an extended com-ponent, with a size of 0.5×θmin

beam and a total flux densityof 1.5 Jy, locating at the southwest of the image center.The bottom-left panel in Figure 8 depicts the gener-ated dirty image. The ripples of sidelobes of Source#4 are prominent. Sidelobes of other point sources arealso visible. The bottom-right panel shows the decon-volution image after 100 iterations of CLEANing. Thesidelobes have been substantially suppressed, and thebackground noise becomes smooth. The four sourcesare distinctive. The dynamic range increases from 11:1to 237:1. Deeper CLEANing may further decrease therms noise.

5. SUMMARY

In this paper, we have introduced an auxiliary tool aid-ing VLBI network simulations, named as VNSIM. Themotivation is to provide an integrated software packageto help radio astronomers to make observation scheduleand to gain a preliminary evaluation of the interferom-eter performance. Compared with the existing simula-tion tools, VNSIM not only integrates commonly usedfunctions but also supplements new features support-ing large surveys containing multiple sources. Anothernew feature is the space VLBI simulation which sup-plies valuable guidance to future space VLBI missions.Details of the space VLBI (u, v) coverage simulationwill be presented in a forthcoming paper. By design,VNSIM is not limited to VLBI networks, but is in gen-eral applicable to connected-element interferometers,such as VLA. Considering the usability for non-VLBIastronomers without much interferometric knowledge,VNSIM has been designed to be more friendly in user in-terfaces and more convenient in database management.All kinds of parameters can also be user specified to

VLBI Network SIMulator 9

Figure 8. Results of the image simulation experiment.The minor axis of dirty beam size is around 0.653 mas.Top-right panel shows the distribution of four simulatedsources. Bottom-left panel shows the dirty image. Thesidelobes around each source are clearly seen. Bottom-rightpanel shows the deconvoluted image after 100 iterations ofCLEANing. The strong sidelobes become less significant,and the dynamic range increases to 237:1.

investigate more simulation scenarios. The comparisonof the simulation results from VNSIM with other toolsverified the consistency between them. The current ver-sion of VNSIM provides functionality matching the as-tronomers’ basic requirements for scheduling and evalu-ating ground-based interferometric observations. Moresophisticated functions are under development. In thefuture version, we aim to provide more precise spaceVLBI simulation with realistic constraints and completescheduling plans.

ACKNOWLEDGMENTS

The simulations were performed on the Data ProcessorPrototype of the China SKA Regional Center hostedby Shanghai Astronomical Observatory with fundingsupport from the Ministry of Science and Technol-ogy of China (grant No. SQ2018YFA040022) andthe Chinese Academy of Sciences (CAS, grant No.114231KYSB20170003). T.A. thanks the youth innova-tion promotion association of the CAS. The authors aregrateful to Zsolt Paragi for his help about VLBI networkparameter evaluation, and Junyi Wang and Sandor Freyfor their constructive comments on the simulations ofspace VLBI.

REFERENCES

An, T., Sohn, B. W., & Imai, H. 2018, Capabilities andprospects of the East Asia Very Long Baseline Interfer-ometry Network, Nature Astronomy, 2, 118.

An, T., Lao, B., Wang, J., et al. 2016, Space millimeter-wavelength very long baseline interferometry simulationsoftware. 2016 8th IEEE International Conference Com-munication Software and Networks, 274.

Asaki, Y., & Miyoshi, M. 2009, A future submillimeter-waveSpace VLBI, In Approaching Micro-Arcsecond Resolutionwith VSOP-2: Astrophysics and Technologies, 402, 431.

Asaki, Y., Sudou, H., Kono, Y., et al. 2007, Verificationof the effectiveness of VSOP-2 phase referencing with anewly developed simulation tool, ARIS, Publications ofthe Astronomical Society of Japan, 59, 397.

Cheng, X. P., An, T., Hong, X. Y., et al. 2017, The mostcompact bright radio-loud AGNs – II. VLBA observationsof ten sources at 43 and 86 GHz. Astrophysical JournalSupplement, 234, 17.

Frey, S., Fejes, I., Heitler, G., et al. 1998, Space VLBI Assis-tance Software – New Version, International Astronomi-cal Union Colloquium, 164, 399.

Hirabayashi, H., Hirosawa, H., Kobayashi, H., et al. 2000,The VLBI space observatory programme and the radio-astronomical satellite HALCA, Publications- Astronomi-cal Society of Japan, 52, 955.

Hogbom, J. A. 1974, Aperture synthesis with a nonregulardistribution of interferometer baselines, Astronony Astro-phys Suppl, 15, 417.

Hong, X., Shen, Z., An, T., Liu, Q. 2014, The chinese spacemillimeter-wavelength VLBI array – step toward imagingthe most compact astronomical objects, Acta Astronau-tica, 102, 217.

Kardashev, N. S., Khartov, V. V., Abramov, V. V., et al.2013, ”Radioastron”-a telescope with a size of 300 000km: main parameters and;first observational results, As-tronomy Reports, 57, 153.

Smith, J. G., et al., JPL contribution to the VSOP mission,Advances in Space Research, 26, 637-640.

Martı-Vidal, I. 2017, APSYNSIM: An Interactive Tool ToLearn Interferometry, arXiv:1706.00936.

Readhead, A. C. S., Wilkinson, P. N. 1978, The mappingof compact radio sources from vlbi data, AstrophysicalJournal, 223, 25.

Readhead, A. C. S., Walker, R. C., et al. 1980, Mappingradio sources with uncalibrated visibility data, Nature,285, 137.

Shepherd, M. C. 1997, Difmap: An interactive program forsynthesis imaging. In Astronomical Data Analysis Soft-ware and Systems VI, 125, 77.

Thompson, A. R. 1999, Fundamentals of Radio Interferom-etry, Synthesis Imaging in Radio Astronomy II, 180, 11.

Thompson, A. R., Moran, J. M., Swenson, G. W. 2007,Interferometry and synthesis in radio astronomy, JohnWiley & Sons, New York.

Zhuravlev, V. I. 2015, Fakerat software in the interna-tional interferometric radioastron project with very longground-space bases, Cosmic Research, 53, 216.