Hi,
I've tried to use a censored normal distribution as prior to a latent variable and i get the following error:
BugsCmds.SetFilePath( 'C:/Users/.../inits1.txt' ); BugsCmds.LoadInitsGuard; BugsCmds.specificationDialog.chain := 1; BugsCmds.LoadInits
error for node T of type dnorm node censored node can not be used as prior

I ran multibugs from R using the R2MultiBUGS package. I've also ran the same model with openbugs from R without trouble.

A short overview of the hierarchical model :
T0 ~ dnorm(0,1.0E-6)C(0,)
for( i in 1 : n ) {
T[i] ~ dnorm(T0,tau.T0)C(0,)
}

Thanks

This created some red herrings for me until I realized what was happening. On Windows 11, pinning a shortcut on the taskbar while MultiBUGS is open only creates a shortcut to OpenBUGS.

The new MS MPI v10.0 can support multiple processors now, if version <10.0, then all MultiBUGS processes always crowd in the same processor group (i.e. all the processes run on the same physical CPU, as I tested on a two-way server)

See: https://docs.microsoft.com/en-us/message-passing-interface/microsoft-mpi-release-notes

MS-MPI v10.0 includes the following improvements and fixes. Download MS-MPI v10.0 from the Microsoft Download Center.
Support for utilizing multiple processor groups

First of all, thank you all for your fantastic work on MultiBUGS. Amazing.
We have been running simulations, and from time to time a message pops up, like "TRAP 77", "TRAP 15", etc. I have used pkill to kill OpenBUGS

pkill OpenBUGS

However, while the process OpenBUGS is successfully killed and disappears from the screen, the trap message remains on the screen. Because we have been using loop, with time the screen gets full of trap error messages. Is there any way to close those traps messages specifically?

All the best,

Tiago

Hello,

I met an error message "unable to link BugsWorker executable" when click the distribute button. Could you please help me to solve this issue?

Many thanks.

Best,
Sesil

I've been using MATLAB to create my BUGS scripts as well as my input files (data and initialized values) and have it call OpenBUGS to run my model. Unfortunately, I've been unsuccessful with doing this using MultiBUGS. The issues seems to be with the additional line in the script "modelDistribute(4)". MultiBUGS stops here and the log files shows:

...
BugsCmds.UpdateGaurd;
BugsCmds.specificationDialog.numProc :=4;
BugsMaster.Install('MPImsimp');
BugsCmds.Distribute('MPImsimp')

port name:
ncalrpc:AE550-2831[794cef1a-041a-4c3f-9fa3-dfe9f5e6f8e3]
ncacn_ip_tcp:ae550-2831[57750] 1
BugsWorker.exe written 605696 293392

mpi command:
mpiexec -n 4 "C:\MultiBUGS\BugsWorker"

However, if I set the number of cores to be 1 in MATLAB then there isn't an issue. Also, if I run the script I created with Matlab directly in MultiBUGS (without using matlab to call MultiBUGS) the model also runs without an issue. I was wondering if there is a reason that MultiBUGS doesn't want to run when using MATLAB to call it, specificaly when the number of cores is set to more than one?

Thanks,
Alessio

I am encountering a very perplexing problem. I am running BUGS code in OpenBUGS as well as MultiBUGS. The reason I started using the latter is because I think my problem will likely need to be scaled up considerably but I am still in the diagnostic simulation phase.

Surprisingly, my program works in OpenBUGS but does not work in MultiBUGS. My program right now is a simple Poisson GLM with multivariate normal priors (using a small sample size (91) and minimal MCMC iterations. A dialogue box pops up saying Trap 066 as well as other dialogue boxes saying undefined real result. I understand that can happen but I also can't believe that this model can't run because it actually works in OpenBUGS.

So, what changes do I need to make to get it to run?

Below is the invocation of MultiBUGS using R. I'm also attaching the screenshot and the BUGS code. And to reiterate, the program works in OpenBUGS.

data<-list(n=dim(Disc)[1],m=dim(Disc)[2],x1=Lag1,y=Disc,W=W)
inits <- function(){list(alpha=0.5,beta1=0.5,mu0=0.5,tau0=1,rho=0,mu11=1,tau01=1,tau11=1)}

parameters<-c("alpha","beta1","mu0","mu11","rho","tau0")
sim1 <- bugs(data, inits, parameters, model.file="PoissonReg.txt",n.chains = 1, n.iter = 1000, n.burn=20, n.workers = 7, MultiBUGS.pgm='D:/MultiBUGS//MultiBUGS.exe',working.directory =NULL,debug=TRUE)
#sim1 <- bugs(data, inits, parameters, model.file="PoissonReg.txt",n.chains = 1, n.iter = 100, n.burn=1, working.directory = NULL,debug=TRUE)

PoissonReg.txt

I get some Black Box error that causes the program to close once I close the error dialog. For some reason this occurs after about 50,000 iterations. I've attached the model with data.
VB_growth_model_crash50k.zip

R2MultiBUGS keeps failing, and I've noticed that there are may be differences between the traditional BUGS script and the MultiBUGS script, such as the call like modelDistribute(), is there a manual for MultiBUGS script? Thanks.

Hi there,

I've had to re-install MultiBUGS on my machine after having the drive replaced. However, I now get an error when I attempt to run the MultiBUGS exe file - "Windows cannot find 'mpiexec'....". I've installed the MS-MPI file as per the MultiBUGS installation instructions. No problems in the install that I can see, but I can find no trace of 'mpiexec'. Could there have been a change with the latest version of MS-MPI that's causing the issue? Previous successful install on this machine was circa 3 months ago.

Greatly appreciate any assistance. MultiBUGS is a fantastic resource!

Cheers,

Geoff

I get the error that the distribution cannot be truncated when checking the model. For example, this line:
sigma.phi ~ dt(0,pr.tauphi[1],pr.tauphi[2]) T(0,)

OpenBugs does not throw this error.

When I was building MultiBUGS from source, I've noticed that the built package was still in separate directories, and I am not very familiar with the Black-box builder thing.

After the compilation, the 'MultiBUGS.exe' in the source directory still refuses to run on a double-click, and the MultiBUGS can run successfully using the shortcut created following the 'README.md'.

A 'Linking.odc' was located in the 'Developer' sub-directory, but it appears to me that this file is not dedicated to MultiBUGS (as 'OpenBUGS.exe' is still in the file).

I'm wondering how can I construct a release-like MultiBUGS package (just like the download on the MultiBUGS website) so I can deploy MultiBUGS on a remote server more easily?
Thx.

Somehow I just lost all my results, and it occurs to me that I might have set too many variables monitored (some of them are time-varying large matrices).

All the 'model' 'update' 'sampler' menu all refused to work properly and gave 'trap' issues, and It is surprising to see that the iteration is beyond the preset updates.

Any suggestions for monitoring complex models? As I can see the MultiBUGS is still a 32-bit program, providing warnings (for too many monitors set) would be nice, because losing all the results after a long run is very frustrating.

I've been working with OpenBUGS for awhile now and looking to move towards MultiBUGS in hopes of speeding up my model runs. I usually run OpenBUGS using MATLAB and tried to alter my code to work with MultiBUGS by specify the number of core I want to use in my script file but I was getting the "NIL dereference (read)" error when doing this. I thought I would try to run MultiBUGS following the step-by-step example outlined in "Massively parallel MCMC for Bayesian hierarchical models." using the seeds data but I get the same error when I click on "distribute". I was wondering if there is something more that needs to be done when moving from OpenBUGS to MultiBUGS than just specifying "modelDistribute(2)" in the script file or if its there is something else that I might be doing wrong to get this error.

Thank you in advance for your help.

Alessio

Any experiences with using non 32-bit MPIs?

I'm on a compute cluster that uses MVAPICH2 (MPICH ABI compatible)

https://www.mpich.org/abi/

There are a few bugs in handling of spaces in the paths to scripts when using

MultiBUGS.exe /PAR "path/my folder/to/script.txt"

Some of these are fixed by avoiding MultiBUGS.exe wrapper, and using the direct command

mpiexec -n 1 OpenBUGS.exe /PAR "path/my folder/to/script.txt"

The script being on a different volume sometimes seems to add a complication. Need a full set of test cases.

I am trying to speed up my run times, and so have downloaded MultiBUGS to make use of the modelDistribute() command.

When I try to distribute the model over any number of cores, a black box error "undefined real result" shows up. I suspect this is an issue with the modelDistribute() command, as when I run the problem on only 1 core (i.e. not distributed), it always runs to completion and produces reasonable results.

I have tried adjusting the refresh parameter in the modelUpdate() command (per #33) but the problem persists.

Any advice on how to resolve this? Thanks!

Hi.

I tried to download MultiBUGS via Chrome but it blocked the file saying "MultiBUGS-2.0.zip is dangerous". I turned off the protection of the browser but then the file got removed automatically. Any suggestions? Thanks!

A colleague and I are trying to install multibugs but we hit the antivirus problem, one with Windows Defender and the other with McAfee. Are there any workarounds?

Hi,
The model in question is a very simple beta regression with fixed effects.
When using 1 core, the results obtained are coherent and in line with what i expect.
I do 10000 iterations, and I observe convergence after ~1000 iterations with 2 chains. data is 10000 lines.
The problem appears when I do the exact same operation using 2 or 4 cores. The programm update correctly up to 10000 and is way faster than when using 1 core, but when i go to check the posteriors, I obtain completely different and absurd results, with variances of 30 on all my parameters.

To Resume, it's obvious there is something wrong with calculations when using multiple cores, but i have no idea what. Feel free to ask for details.

Any help appreciated, Thanks

Avira says MultiBUGS has the pattern of HEUR/APC (cloud). I reported it to Avira and added an exception. Avira exceptions are here in case others have the issue: https://support.avira.com/hc/en-us/articles/360000834465-How-do-I-setup-Exceptions-for-Files-Folders-and-Processes-?utm_source=CS&utm_medium=KB

We receive the following error while we try to run run-MultiBUGS:

[siavoa01@cn-0003 ~]$ run-MultiBUGS
HostFiles: Value of BB_PRIMARY_DIR isn't directory, using cwd

• BlackBox
• HostGnome: can not load libgtk-x11-2.0.so.0

• Error 01
• Init: code file not found

Primary job terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.

mpiexec detected that one or more processes exited with non-zero status, thus causing
the job to be terminated. The first process to do so was:

Process name: [[17442,1],0]
Exit code: 1

the gtk2+/2.24.32 is installed and the library is there. I would appreciate you helping to address this issue.

I've installed MS-MPI and tried to run one chain distributed to two cores. It's distributed but I got the following black box error when the model was updating. Any idea how to debug it? Thanks!

Version: MultiBUGS v1.0

Running a test model to test model externalization and internalization, the externalization seemed to work fine but the internalization process keeps throwing exceptions (the model seems to be loaded):

NIL dereference (read)

 SummaryMonitors.StdMonitor.Internalize   [00000166H] 
	.i	INTEGER	0
	.monitor	SummaryMonitors.StdMonitor	[02B5D980H] 
	.node	GraphNodes.Node	NIL
	.present	BOOLEAN	TRUE
	.rd	Stores.Reader	fields
	.size	INTEGER	1
	.string	ARRAY 128 OF CHAR	"deviance"   ...
 SummaryIndex.Monitor.Internalize   [000005C1H] 
	.filePos	INTEGER	9089
	.i	INTEGER	2
	.m	SummaryIndex.Monitor	[02B5D8A0H] 
	.monitor	SummaryMonitors.Monitor	[02B5D980H] 
	.numMonitors	INTEGER	5
	.rd	Stores.Reader	fields
 MonitorMonitors.InternalizeMonitors   [00000381H] 
	.cursor	MonitorMonitors.MonitorList	NIL
	.i	INTEGER	0
	.install	ARRAY 256 OF CHAR	"SummaryIndex.Install"   ...
	.monitor	MonitorMonitors.Monitor	[02B5D8A0H] 
	.numTypes	INTEGER	2
	.rd	Stores.Reader	fields
	.start	INTEGER	9186
 BugsSerialize.InternalizeMonitors   [0000036EH] 
	.rd	Stores.Reader	fields
 BugsSerialize.Internalize   [00000667H] 
	.f	Files.File	[02B3CF40H] 
	.fileName	ARRAY 256 OF CHAR	"bugs_model.bug"   ...
	.offsetGraph	INTEGER	1152
	.offsetMonitor	INTEGER	6141
	.offsetMutable	INTEGER	5285
	.rd	Stores.Reader	fields
 BugsCmds.Internalize   [00002A38H] 
	.index	INTEGER	0
	.name	Dialog.String	"bugs_model"   ...
	.ok	BOOLEAN	TRUE
 Kernel.Call   [0000208FH] 
	.adr	INTEGER	1746101781
	.kind	INTEGER	-2
	.n	INTEGER	0
	.p	INTEGER	0
	.par	ARRAY 256 OF INTEGER	elements
	.r	REAL	1.087308849615713E-311
	.sig	POINTER	[68131D48H]
	.size	INTEGER	1988722704
	.sp	INTEGER	981852632
	.typ	POINTER	NIL
 Meta.Item.ParamCallVal   [00002A7EH] 
	.adr	INTEGER	1746101781
	.data	ARRAY 256 OF INTEGER	elements
	.dest	Meta.Item	fields
	.n	INTEGER	0
	.ok	BOOLEAN	TRUE
	.p	INTEGER	0
	.padr	INTEGER	1658323436
	.par	ARRAY 100 OF POINTER	elements
	.proc	Meta.Item	fields
	.ret	LONGINT	0063E89800000000H
	.sig	POINTER	[68131D48H]
	.x	Meta.Item	fields
 StdInterpreter.CallProc   [00000474H] 
	.i	Meta.Item	fields
	.mod	ARRAY 256 OF CHAR	"BugsCmds"   ...
	.ok	BOOLEAN	TRUE
	.proc	ARRAY 256 OF CHAR	"Internalize"   ...
	.t	Meta.Item	fields
 StdInterpreter.Command   [00000A77H] 
	.left	Kernel.Name	"BugsCmds"   ...
	.right	Kernel.Name	"Internalize"   ...
 StdInterpreter.CallHook.Call   [00000B43H] 
	.ch	CHAR	0X
	.errorMsg	ARRAY 2 OF CHAR	" "   ...
	.hook	StdInterpreter.CallHook	[02B9D9B0H] 
	.i	INTEGER	42
	.id	Kernel.Name	"Internalize"   ...
	.numPar	INTEGER	0
	.par	ARRAY 100 OF POINTER	elements
	.proc	ARRAY 256 OF CHAR	"StdCmds.CloseDialog; BugsCmds.In"   ...
	.res	INTEGER	0
	.type	INTEGER	3
	.x	INTEGER	0
 Dialog.Call   [0000388FH] 
	.errorMsg	ARRAY 2 OF CHAR	" "   ...
	.proc	ARRAY 256 OF CHAR	"StdCmds.CloseDialog; BugsCmds.In"   ...
	.res	INTEGER	0
 Controls.Call   [00002A4FH] 
	.c	Controls.PushButton	[02B4FDB0H] 
	.msg	Views.NotifyMsg	fields
	.ok	BOOLEAN	FALSE
	.p	Controls.Control	[02B4FDB0H] 
	.res	INTEGER	0
 Controls.Do   [00002AC0H] 
	.f	StdCFrames.PushButton	[02B3E6C0H] 
 HostCFrames.Execute   [00001B6BH] 
	.f	HostCFrames.PushButton	[02B3E6C0H] 
	.old	INTEGER	65539
 HostCFrames.HandleCommand   [00008C44H] 
	.b	BOOLEAN	FALSE
	.c	StdCFrames.Frame	[02B3E6C0H] 
	.i	INTEGER	243
	.j	INTEGER	22814960
	.lParam	INTEGER	133848
	.nc	INTEGER	0
	.ps	POINTER	NIL
	.res	INTEGER	4
	.s	ARRAY 512 OF CHAR	"큺㚸cc䔁眢"   ...
	.wParam	INTEGER	0
	.wnd	INTEGER	133854
 Kernel.Try   [000045F9H] 
	.a	INTEGER	133854
	.b	INTEGER	0
	.c	INTEGER	133848
	.h	PROCEDURE	HostCFrames.HandleCommand
 HostCFrames.CtrlHandler   [00009914H] 
	.c	StdCFrames.Frame	NIL
	.lParam	INTEGER	133848
	.message	INTEGER	273
	.p	POINTER	NIL
	.res	INTEGER	0
	.wParam	INTEGER	0
	.wnd	INTEGER	133854
<system>   (pc=7689626AH,  fp=0063F5B8H)
<system>   (pc=768871BBH,  fp=0063F69CH)
<system>   (pc=7688695CH,  fp=0063F700H)
<system>   (pc=76886622H,  fp=0063F734H)
<system>   (pc=768B0731H,  fp=0063F758H)
<system>   (pc=768B042EH,  fp=0063F780H)
<system>   (pc=768AF937H,  fp=0063F820H)
<system>   (pc=768AEE51H,  fp=0063F84CH)
<system>   (pc=7689626AH,  fp=0063F878H)
<system>   (pc=768871BBH,  fp=0063F95CH)
<system>   (pc=768862EAH,  fp=0063F9D0H)
<system>   (pc=768860BFH,  fp=0063F9DCH)
 HostCFrames.HandleMouse   [0000168BH] 
	.b	INTEGER	0
	.buttons	SET	{16}
	.hc	INTEGER	1
	.m	INTEGER	513
	.res	INTEGER	0
	.tc	HostCFrames.TrapCleaner	[02B5C620H] 
	.w	INTEGER	133848
	.wnd	INTEGER	133848
	.x	INTEGER	47
	.y	INTEGER	13
 HostCFrames.PushButton.MouseDown   [00001B29H] 
	.buttons	SET	{16}
	.f	HostCFrames.PushButton	[02B3E6C0H] 
	.x	INTEGER	451875
	.y	INTEGER	128025
 Controls.CatchCtrlMsg   [000028B3H] 
	.c	Controls.Control	[02B4FDB0H] 
	.f	Views.Frame	[02B3E6C0H] 
	.focus	Views.View	NIL
	.msg	Controllers.TrackMsg	fields
 Controls.PushButton.HandleCtrlMsg2   [00002C9CH] 
	.c	Controls.PushButton	[02B4FDB0H] 
	.f	Views.Frame	[02B3E6C0H] 
	.focus	Views.View	NIL
	.msg	Controllers.TrackMsg	fields
 Controls.Control.HandleCtrlMsg   [00001F73H] 
	.c	Controls.Control	[02B4FDB0H] 
	.dcOk	BOOLEAN	TRUE
	.f	Views.Frame	[02B3E6C0H] 
	.focus	Views.View	NIL
	.msg	Controllers.TrackMsg	fields
	.p	Controls.Control	NIL
	.sp	Properties.SizeProp	NIL
 Views.ForwardCtrlMsg   [000026B0H] 
	.a	Views.View	NIL
	.f	Views.Frame	[02B3E6C0H] 
	.focus	Views.View	NIL
	.g	Views.Frame	[02B3BBF0H] 
	.h	Views.Frame	NIL
	.msg	Controllers.TrackMsg	fields
	.op	INTEGER	0
	.req	BOOLEAN	FALSE
	.v	Views.View	[02B4FDB0H] 
 Views.ForwardCtrlMsg   [0000276BH] 
	.a	Views.View	NIL
	.f	Views.Frame	[02B3BBF0H] 
	.focus	Views.View	[02B4FDB0H] 
	.g	Views.Frame	[02B6BC70H] 
	.h	Views.Frame	[02B3E6C0H] 
	.msg	Controllers.TrackMsg	fields
	.op	INTEGER	0
	.req	BOOLEAN	FALSE
	.v	Views.View	[02B774B0H] 
 Views.ForwardCtrlMsg   [0000276BH] 
	.a	Views.View	NIL
	.f	Views.Frame	[02B6BC70H] 
	.focus	Views.View	[02B774B0H] 
	.g	Views.Frame	NIL
	.h	Views.Frame	[02B3BBF0H] 
	.msg	Controllers.TrackMsg	fields
	.op	INTEGER	0
	.req	BOOLEAN	FALSE
	.v	Views.View	[02B5C590H] 
 Windows.Window.ForwardCtrlMsg   [000002AFH] 
	.msg	Controllers.TrackMsg	fields
	.w	Windows.Window	[02B47E40H] 
 HostWindows.Window.ForwardCtrlMsg   [000016A7H] 
	.d	BOOLEAN	FALSE
	.msg	Controllers.TrackMsg	fields
	.res	INTEGER	513
	.w	HostWindows.Window	[02B47E40H] 
 HostWindows.Window.MouseDown   [00001BF3H] 
	.modifiers	SET	{16}
	.ph	INTEGER	36
	.pw	INTEGER	234
	.time	INTEGER	0
	.track	Controllers.TrackMsg	fields
	.w	HostWindows.Window	[02B47E40H] 
	.x	INTEGER	61
	.y	INTEGER	21
 HostWindows.HandleMouse   [00003287H] 
	.b	SET	{16}
	.f	Views.Frame	NIL
	.g	Views.Frame	NIL
	.isDown	BOOLEAN	TRUE
	.lParam	INTEGER	1376317
	.w	HostWindows.Window	[02B47E40H] 
	.wParam	INTEGER	257
	.wnd	INTEGER	460784
	.x	INTEGER	61
	.y	INTEGER	21
 Kernel.Try   [000045F9H] 
	.a	INTEGER	460784
	.b	INTEGER	257
	.c	INTEGER	1376317
	.h	PROCEDURE	HostWindows.HandleMouse
 HostWindows.DialogHandler   [000045C3H] 
	.dx	INTEGER	0
	.dy	INTEGER	22983680
	.lParam	INTEGER	1376317
	.message	INTEGER	513
	.res	INTEGER	0
	.w	HostWindows.Window	NIL
	.wParam	INTEGER	1
	.wnd	INTEGER	460784
	.x	INTEGER	36
<system>   (pc=7689626AH,  fp=0063FCACH)
<system>   (pc=768871BBH,  fp=0063FD90H)
<system>   (pc=768862EAH,  fp=0063FE04H)
<system>   (pc=768860BFH,  fp=0063FE10H)
 HostMenus.Loop   [00003DDDH] 
	.done	BOOLEAN	FALSE
	.f	SET	{0..5}
	.n	INTEGER	0
	.res	INTEGER	0
	.w	HostWindows.Window	NIL
 Kernel.Start   [000036E3H] 
	.code	PROCEDURE	HostMenus.Loop

An error occurred when I specified the total number of cores. The error messages are "TRAP 101 (invariant violated)". Hope the good person can answer me, this error generally refers to what？

The link:
https://blackbox.obertone.ru/download
should be replaced to:
https://blackbox.oberon.org/download

Trend micro identifies the exe as a Trojan.

Edit: this seems to be fine now.

I am fitting a group of Bernoulli models on data with 258,000 observations. Model compiles and runs fine on a subset of first 100,000 observations. With the full data, the compilation step ends without any warning or error message.
Is there any limit on the data size?
Thanks,
Paramjit Gill

After the run I found that I can't save the coda file
Am I outputting too many variables at once, and what is the limit?
The 'stat' seems to work fine

The error messages are listed as followed

TRAP 101  (invariant violated)

 HostFiles.Writer.SetPos   [00002090H] 
	.buf	HostFiles.Buffer	[5EAF2CD0H] 
	.count	INTEGER	2048
	.f	HostFiles.File	[5EAF11B0H] 
	.i	INTEGER	0
	.offset	INTEGER	0
	.org	INTEGER	2147481600
	.pos	INTEGER	2147481600
	.res	INTEGER	0
	.w	HostFiles.Writer	[0328EF40H] 
 HostFiles.Writer.WriteByte   [0000214DH] 
	.w	HostFiles.Writer	[0328EF40H] 
	.x	BYTE	49
 TextModels.WriteSChar   [00005123H] 
	.bunch	BOOLEAN	FALSE
	.ch	SHORTCHAR	"1"
	.op	TextModels.EditOp	NIL
	.p	TextModels.Piece	NIL
	.pos	INTEGER	2147481600
	.spillPos	INTEGER	2147481599
	.t	TextModels.StdModel	[029F0A30H] 
	.u	TextModels.Run	NIL
	.un	TextModels.Run	NIL
	.wr	TextModels.StdWriter	[0327D7B0H] 
 TextModels.StdWriter.WriteChar   [0000544CH] 
	.bunch	BOOLEAN	FALSE
	.ch	CHAR	"1"
	.fw	Files.Writer	NIL
	.lp	TextModels.LPiece	NIL
	.op	TextModels.EditOp	NIL
	.pos	INTEGER	1
	.spillPos	INTEGER	1729236914
	.t	TextModels.StdModel	NIL
	.u	TextModels.Run	NIL
	.un	TextModels.Run	NIL
	.wr	TextModels.StdWriter	[0327D7B0H] 
 TextMappers.Formatter.WriteChar   [0000180DH] 
	.f	TextMappers.Formatter	fields
	.x	CHAR	"1"
 TextMappers.Formatter.WriteString   [0000191FH] 
	.f	TextMappers.Formatter	fields
	.i	INTEGER	1
	.x	ARRAY 80 OF CHAR	"81701"   ...
 TextMappers.Formatter.WriteIntForm   [00001AAEH] 
	.base	INTEGER	10
	.f	TextMappers.Formatter	fields
	.fillCh	CHAR	8FX
	.minWidth	INTEGER	0
	.s	ARRAY 80 OF CHAR	"81701"   ...
	.showBase	BOOLEAN	FALSE
	.x	LONGINT	81701
 TextMappers.Formatter.WriteInt   [00001873H] 
	.f	TextMappers.Formatter	fields
	.x	LONGINT	81701
 SamplesFormatted.FormatCODA   [00000B92H] 
	.beg	INTEGER	0
	.chain	INTEGER	1
	.end	INTEGER	10000000
	.f	ARRAY 2 OF TextMappers.Formatter	elements
	.firstChain	INTEGER	1
	.i	INTEGER	3721
	.iteration	INTEGER	81701
	.j	INTEGER	0
	.k	INTEGER	6700
	.label	ARRAY 128 OF CHAR	"factor[37,16]"   ...
	.lastChain	INTEGER	1
	.line	INTEGER	171405000
	.num	INTEGER	9475
	.numChains	INTEGER	1
	.offsets	POINTER	[5FFE6A10H] 
	.sample	POINTER	[5EBCC2E0H] 
	.sampleSize	INTEGER	7500
	.thin	INTEGER	1
	.variable	ARRAY 256 OF CHAR	"factor"   ...
 SamplesFormatted.CODA   [00000CF0H] 
	.beg	INTEGER	0
	.end	INTEGER	10000000
	.f	ARRAY 2 OF TextMappers.Formatter	elements
	.firstChain	INTEGER	1
	.i	INTEGER	16
	.lastChain	INTEGER	1
	.len	INTEGER	23
	.line	INTEGER	171405000
	.monitors	POINTER	[07AF6CF0H] 
	.thin	INTEGER	1
	.variable	ARRAY 256 OF CHAR	"factor"   ...
 SamplesCmds.CODA   [00000791H] 
	.beg	INTEGER	0
	.end	INTEGER	10000000
	.f	POINTER	[029BC9A0H] 
	.firstChain	INTEGER	1
	.i	INTEGER	2
	.lastChain	INTEGER	1
	.numAsString	ARRAY 8 OF CHAR	""   ...
	.numChains	INTEGER	1
	.oldWhereOut	INTEGER	0
	.pos	INTEGER	0
	.string	Dialog.String	"*"   ...
	.text	POINTER	[03275DA0H] 
	.thin	INTEGER	1
 Kernel.Call   [0000208FH] 
	.adr	INTEGER	1694504897
	.kind	INTEGER	-2
	.n	INTEGER	0
	.p	INTEGER	0
	.par	ARRAY 256 OF INTEGER	elements
	.r	REAL	1.012470477442054E+262
	.sig	POINTER	[65000A20H]
	.size	INTEGER	1989181456
	.sp	INTEGER	-722191906
	.typ	POINTER	NIL
 Meta.Item.ParamCallVal   [00002A7EH] 
	.adr	INTEGER	1694504897
	.data	ARRAY 256 OF INTEGER	elements
	.dest	Meta.Item	fields
	.n	INTEGER	0
	.ok	BOOLEAN	TRUE
	.p	INTEGER	0
	.padr	INTEGER	7799276
	.par	ARRAY 100 OF POINTER	elements
	.proc	Meta.Item	fields
	.ret	LONGINT	0063E89800000000H
	.sig	POINTER	[65000A20H]
	.x	Meta.Item	fields
 StdInterpreter.CallProc   [00000474H] 
	.i	Meta.Item	fields
	.mod	ARRAY 256 OF CHAR	"SamplesCmds"   ...
	.ok	BOOLEAN	TRUE
	.proc	ARRAY 256 OF CHAR	"CODA"   ...
	.t	Meta.Item	fields
 StdInterpreter.Command   [00000A77H] 
	.left	Kernel.Name	"SamplesCmds"   ...
	.right	Kernel.Name	"CODA"   ...
 StdInterpreter.CallHook.Call   [00000B34H] 
	.ch	CHAR	0X
	.errorMsg	ARRAY 2 OF CHAR	" "   ...
	.hook	StdInterpreter.CallHook	[02A1D9B0H] 
	.i	INTEGER	17
	.id	Kernel.Name	"CODA"   ...
	.numPar	INTEGER	0
	.par	ARRAY 100 OF POINTER	elements
	.proc	ARRAY 256 OF CHAR	"SamplesCmds.CODA"   ...
	.res	INTEGER	0
	.type	INTEGER	3
	.x	INTEGER	0
 Dialog.Call   [0000388FH] 
	.errorMsg	ARRAY 2 OF CHAR	" "   ...
	.proc	ARRAY 256 OF CHAR	"SamplesCmds.CODA"   ...
	.res	INTEGER	0
 Controls.Call   [00002A4FH] 
	.c	Controls.PushButton	[5EA8ED70H] 
	.msg	Views.NotifyMsg	fields
	.ok	BOOLEAN	FALSE
	.p	Controls.Control	[5EA8ED70H] 
	.res	INTEGER	0
 Controls.Do   [00002AC0H] 
	.f	StdCFrames.PushButton	[5EADC660H] 
 HostCFrames.Execute   [00001B6BH] 
	.f	HostCFrames.PushButton	[5EADC660H] 
	.old	INTEGER	65539
 HostCFrames.HandleCommand   [00008C44H] 
	.b	BOOLEAN	FALSE
	.c	StdCFrames.Frame	[5EADC660H] 
	.i	INTEGER	243
	.j	INTEGER	23246368
	.lParam	INTEGER	526276
	.nc	INTEGER	0
	.ps	POINTER	NIL
	.res	INTEGER	4
	.s	ARRAY 512 OF CHAR	"䝱ꊠcc䓱癔"   ...
	.wParam	INTEGER	0
	.wnd	INTEGER	1312170
 Kernel.Try   [000045F9H] 
	.a	INTEGER	1312170
	.b	INTEGER	0
	.c	INTEGER	526276
	.h	PROCEDURE	HostCFrames.HandleCommand
 HostCFrames.CtrlHandler   [00009914H] 
	.c	StdCFrames.Frame	NIL
	.lParam	INTEGER	526276
	.message	INTEGER	273
	.p	POINTER	NIL
	.res	INTEGER	0
	.wParam	INTEGER	0
	.wnd	INTEGER	1312170
<system>   (pc=7690626AH,  fp=0063F5B8H)
<system>   (pc=768F71BBH,  fp=0063F69CH)
<system>   (pc=768F695CH,  fp=0063F700H)
<system>   (pc=768F6622H,  fp=0063F734H)
<system>   (pc=76920731H,  fp=0063F758H)
<system>   (pc=7692042EH,  fp=0063F780H)
<system>   (pc=7691F937H,  fp=0063F820H)
<system>   (pc=7691EE51H,  fp=0063F84CH)
<system>   (pc=7690626AH,  fp=0063F878H)
<system>   (pc=768F71BBH,  fp=0063F95CH)
<system>   (pc=768F62EAH,  fp=0063F9D0H)
<system>   (pc=768F60BFH,  fp=0063F9DCH)
 HostCFrames.HandleMouse   [0000168BH] 
	.b	INTEGER	0
	.buttons	SET	{16}
	.hc	INTEGER	1
	.m	INTEGER	513
	.res	INTEGER	0
	.tc	HostCFrames.TrapCleaner	[029B95B0H] 
	.w	INTEGER	526276
	.wnd	INTEGER	526276
	.x	INTEGER	21
	.y	INTEGER	15
 HostCFrames.PushButton.MouseDown   [00001B29H] 
	.buttons	SET	{16}
	.f	HostCFrames.PushButton	[5EADC660H] 
	.x	INTEGER	204450
	.y	INTEGER	146625
 Controls.CatchCtrlMsg   [000028B3H] 
	.c	Controls.Control	[5EA8ED70H] 
	.f	Views.Frame	[5EADC660H] 
	.focus	Views.View	NIL
	.msg	Controllers.TrackMsg	fields
 Controls.PushButton.HandleCtrlMsg2   [00002C9CH] 
	.c	Controls.PushButton	[5EA8ED70H] 
	.f	Views.Frame	[5EADC660H] 
	.focus	Views.View	NIL
	.msg	Controllers.TrackMsg	fields
 Controls.Control.HandleCtrlMsg   [00001F73H] 
	.c	Controls.Control	[5EA8ED70H] 
	.dcOk	BOOLEAN	TRUE
	.f	Views.Frame	[5EADC660H] 
	.focus	Views.View	NIL
	.msg	Controllers.TrackMsg	fields
	.p	Controls.Control	NIL
	.sp	Properties.SizeProp	NIL
 Views.ForwardCtrlMsg   [000026B0H] 
	.a	Views.View	NIL
	.f	Views.Frame	[5EADC660H] 
	.focus	Views.View	NIL
	.g	Views.Frame	[07986740H] 
	.h	Views.Frame	NIL
	.msg	Controllers.TrackMsg	fields
	.op	INTEGER	0
	.req	BOOLEAN	FALSE
	.v	Views.View	[5EA8ED70H] 
 Views.ForwardCtrlMsg   [0000276BH] 
	.a	Views.View	NIL
	.f	Views.Frame	[07986740H] 
	.focus	Views.View	[5EA8ED70H] 
	.g	Views.Frame	[078710A0H] 
	.h	Views.Frame	[5EADC660H] 
	.msg	Controllers.TrackMsg	fields
	.op	INTEGER	0
	.req	BOOLEAN	FALSE
	.v	Views.View	[029BF640H] 
 Views.ForwardCtrlMsg   [0000276BH] 
	.a	Views.View	NIL
	.f	Views.Frame	[078710A0H] 
	.focus	Views.View	[029BF640H] 
	.g	Views.Frame	NIL
	.h	Views.Frame	[07986740H] 
	.msg	Controllers.TrackMsg	fields
	.op	INTEGER	0
	.req	BOOLEAN	FALSE
	.v	Views.View	[0779F620H] 
 Windows.Window.ForwardCtrlMsg   [000002AFH] 
	.msg	Controllers.TrackMsg	fields
	.w	Windows.Window	[5EADABD0H] 
 HostWindows.Window.ForwardCtrlMsg   [000016A7H] 
	.d	BOOLEAN	FALSE
	.msg	Controllers.TrackMsg	fields
	.res	INTEGER	513
	.w	HostWindows.Window	[5EADABD0H] 
 HostWindows.Window.MouseDown   [00001BF3H] 
	.modifiers	SET	{16}
	.ph	INTEGER	217
	.pw	INTEGER	395
	.time	INTEGER	0
	.track	Controllers.TrackMsg	fields
	.w	HostWindows.Window	[5EADABD0H] 
	.x	INTEGER	23
	.y	INTEGER	177
 HostWindows.HandleMouse   [00003287H] 
	.b	SET	{16}
	.f	Views.Frame	NIL
	.g	Views.Frame	NIL
	.isDown	BOOLEAN	TRUE
	.lParam	INTEGER	11599895
	.w	HostWindows.Window	[5EADABD0H] 
	.wParam	INTEGER	257
	.wnd	INTEGER	3933382
	.x	INTEGER	23
	.y	INTEGER	177
 Kernel.Try   [000045F9H] 
	.a	INTEGER	3933382
	.b	INTEGER	257
	.c	INTEGER	11599895
	.h	PROCEDURE	HostWindows.HandleMouse
 HostWindows.DialogHandler   [000045C3H] 
	.dx	INTEGER	-1
	.dy	INTEGER	-1
	.lParam	INTEGER	11599895
	.message	INTEGER	513
	.res	INTEGER	0
	.w	HostWindows.Window	NIL
	.wParam	INTEGER	1
	.wnd	INTEGER	3933382
	.x	INTEGER	36
<system>   (pc=7690626AH,  fp=0063FCACH)
<system>   (pc=768F71BBH,  fp=0063FD90H)
<system>   (pc=768F62EAH,  fp=0063FE04H)
<system>   (pc=768F60BFH,  fp=0063FE10H)
 HostMenus.Loop   [00003DDDH] 
	.done	BOOLEAN	FALSE
	.f	SET	{0..5}
	.n	INTEGER	1
	.res	INTEGER	0
	.w	HostWindows.Window	NIL
 Kernel.Start   [000036E3H] 
	.code	PROCEDURE	HostMenus.Loop

Hello - there used to be a guide to coding your own univariate distributions in WBDEV (the old WinBUGS development environment). The URL seems to have gone but here is an archive (https://web.archive.org/web/20141014011916/http://winbugs-development.mrc-bsu.cam.ac.uk/) and here is an archive of some of the old documentation (https://web.archive.org/web/20160222005145fw_/http://winbugs-development.mrc-bsu.cam.ac.uk/files/wbdev_tutorial.pdf) There is an especially to-the-point example PDF called "WBDev_distributions.pdf" which comes part of the WBDEV file but it is encoded and is not easily linked to here. If needed, I could upload it.

We can of course use the "ones trick" or "zeros trick" from the generic sampling distributions help but as mentioned in the WBDEV documentation the compiled versions are faster plus there are several other advantages such as code simplicity, simulation of missing values, etc.

I have done some digging around in OpenBUGS and MultiBUGS and have located the UDev area of OpenBUGS and these carry over to MultiBUGS. I see many new distributions have been added through the ReliaBUGS portion etc. There are also some code listings in Openbugs\Graph\Rsrc\ But I cannot figure out how MultiBUGS would be "informed" of the existence of new compiled distributions. This was previously done in WBDEV by updating \WBDEV\Rsrc\Distributions.odc. There is some good OpenBUGS documentation about adding functions (UDevDocumentation.pdf).

Anyway, I thought I would mention this as a possible (low-priority) feature request. I have some WBDEV code I would like to port over as MultiBUGS is a step up from WinBUGS. Thanks.

I have developed a model which can run on OpenBUGS in batch mode, and I recently moved to MultiBUGS to handle the increasing amount of data I need to process.

My script file runs in the MultiBUGS program (when I navigate to the Model > Script command), but not when I attempt to run in batch mode.

My shell script has the line $path_application/MultiBUGS.exe /PAR "$path_script/$file" but this command never executes the contents of the script "$file". (MultiBUGS opens but hangs)

This same command works when I change the path and application to point to the OpenBUGS directory.

Any advice on how to get this batch mode working for MultiBUGS?
Thanks!

I've appeared to run into a new issue when moving my model from OpenBUGS to MultiBUGS. When I run the model I run into a "trap #100" error that appears to be associated with BlackBox, so this occurs in the part of my script where I set the number of cores. I've determined that if I remove the car.normal part of my model, this issue goes away. I was wondering if it is not possible to use MultiBUGS when the model has a CAR component to it or if there is some kind of modification that must be done to get the model to work with MultiBUGS?

Thanks for the help,
Alessio

Hi all, I am new to BUGS. I used to use WinBUGS to run my CAR model, but it costs about 5 hours, so I want to transfer to MultiBUGS to shorten the computation time. My codes can be run correctly in WinBUGS and OpenBUGS, but when I use the same BUGS code in MultiBUGS(I have adjust the order of the arguments for car.proper), a trap window pop up and the program can not go on being executed(Attached the screenshot). My BUGS code is following:
model{
#likelihood
for (i in 1:nregion){
for (j in 1:noutcome){
mat.theta[i,j] ~ dnorm(theta[i,j], invSig2)
}
}

definition of the matrix theta

for (i in 1:nregion){
for (j in 1:noutcome){
theta[i,j] <- inprod(tPsi[1:j,i], CholDis[j, 1:j])
}
}

Psi matrix(spatially dependent random effects)

for (j in 1:noutcome){
tPsi[j,1:nregion] ~ car.proper(adj[], Cw[], num[], ceros[], Mv[], 1, gamma[j])
gamma[j] ~ dunif(gamma.inf, gamma.sup)
}

for(i in 1:nregion){ceros[i] <- 0}
gamma.inf <- min.bound(Cw[], adj[], num[], Mv[])
gamma.sup <- max.bound(Cw[], adj[], num[], Mv[])

Cholesky matrix inducing dependence between outcomes

diagonal

CholDis[1,1] <- sigma[1]
for (j in 2:noutcome){
diag[j] <- pow(sigma[j],2) - inprod(CholDis[j,1:(j-1)], CholDis[j,1:(j-1)])
CholDis[j,j] <- sqrt(abs(diag[j]))
}

Lower Triangle

First column

for (i in 2:noutcome){CholDis[i,1] <- sigma[i]*roDis[i,1]}

Rest of columns

for (j in 2:(noutcome-1)){
for (i in (j+1):noutcome){
CholDis[i,j] <- (roDis[i,j]*sigma[i]*sigma[j] - inprod(CholDis[i,1:(j-1)], CholDis[j,1:(j-1)]))/CholDis[j,j]
}
}

Positive definiteness condition on the covariance matrix

every term inside the square root at every cell of the diagonal of CholDis should be positive

uno <- 1
uno ~ dbern(condition)
condition <- step(sum(subcondition[2:noutcome])-(noutcome -1))
for (j in 2:noutcome){subcondition[j] <- step(diag[j])}

Prior distributions of correlation between diseases

#(cell not used the Cholesky decompostion are fixed to 0)
for (j in 1:noutcome){roDis[1,j] <- 0}
for (i in 2:noutcome){
for (j in 1:(i-1)){roDis[i,j] ~ dunif(-1,1)}
for (j in i:noutcome){roDis[i,j] <- 0}
}

Other priors

for (j in 1:noutcome){
sigma[j] ~ dunif(0,10)
}

invSig2 ~ dgamma(2, 0.02)
sigma2 <- 1/invSig2
}

Can anyone help me with that? I really appreciate any comment and help!!
Qi

I checked that my laptop computer has 4 cores but MultiBUGS sets the number of cores = 2 at the time of model checking even if I change the option to 4.
How is number of cores set?
Thanks,
Paramjit Gill

dear Dr:
when I run my code by R2MultiBUGS, it shows error that"MultiBUGS did not run correctly. No log file was produced.". How can I do for the error? thank you!

Log info :compiling "MathRandnum"
one error detected

IMPORT
	Math, Meta, Stores := Stores64 **X Symbol file of imported module not found** ; 

where is the Stores64?
Anyway, Thanks a lot!

So I tried out the Linux binary from the tar.gz https://www.multibugs.org/download/

click Doodle > New and got

file Doodle/Rsrc/sizes not found

Turned out the file is called Sizes.odc (capital S), and renaming it to sizes.odc fixed it

Similarly the GeoBUGS folder must be renamed to GeoBugs in order to load the examples

also Maps/Rsrc/ Dialogx.odc must be renamed to dialogx.odc

This are just what I found causally, there must be more.

Also, under Tools > Maths tool, there is no folder named Gft

P.S I extracted the .deb and found the same typos.

Dear,

I have installed MultiBUGS according to the specification, and also Microsoft MPI. When I first used the option distribute, administrator rights were asked and were granted. However, in 2 cases (both multiple chains, and one hierarchical model) the sampling was not distributed over the three cores. This is strange, because when I looked at the Task Manager then three BUGSWorker.exe programs seem to be active, together with OpenBUGS.exe. I also never received an error. So I am puzzled why the cores are not activated. When I compared the processing time with MultiBUGS and with OpenBUGS, I found out they were the same.

Any help is appreciated.

Emmanuel Lesaffre

While loading a data file, I got the error message, "incomplete row of data".
On google search, I found the following remedy
www.downloadchoice.com/common-error-messages-in-bugs-and-their
Click here to fix winbugs error incomplete row of data and related errors. ... Note: This article was updated on 2019-05-09 and previously published under ..

I don't know if the website offering the solution is legit. It asks you to download a fix. Before I do so, I want to make sure that it is.
Paramjit Gill

I'm trying to use MultiBUGS through scripting but I'm having issues saving the outputs in the same way I did with OpenBUGS or WinBUGS. My script has the following lines:

samplesCoda("*",'C:/Users/Documents/bayesian_output/tmp/coda_era_b10_i20_aws_1')

samplesStats("*", 'C:/Users/Documents/bayesian_output/tmp/stats_era_b10_i20_aws_1')

samplesDensity("*",'C:/Users/Documents/bayesian_output/tmp/log_era_b10_i20_aws_1.odc')

samplesHistory("*",'C:/Users/Documents/bayesian_output/tmp/history_era_b10_i20_aws_1.txt')

modelSaveLog('C:/Users/Documents/bayesian_output/tmp/log_era_b10_i20_aws_1.txt')

But this results in a trap error when used with MultiBUGS

There are no issues when I just type :

samplesCoda("*")

samplesStats("*")

samplesDensity("*")

samplesHistory("*")

modelSaveLog('C:/Users/Documents/bayesian_output/tmp/log_era_b10_i20_aws_1.txt')

But since I will be running multiple scripts I was wondering if MultiBUGS just has a different way to automatically save the out into a specified file.

Thanks for the help,
Alessio

Hi, I don't understand the windows installation.

You wrote:

<4. Make a new shortcut to the BlackBox.exe file in the newly-installed BlackBox directory (right-click on BlackBox.exe, choose "Create shortcut"). Then, right click the newly-created shortcut, choose "Properties", and choose the "Shortcut" tab. Then add the following text to the Target box, with the paths C:\Program Files (x86)\BlackBox Component Builder 1.7\BlackBox.exe (path to BlackBox.exe) and C:\MultiBUGS-master (path to the MultiBUGS directory) changed to match wherever you installed BlackBox and MultiBUGS:>

And then you give some code that we run in something? CMD I assume? Or do we even run this code?
mpiexec -n 1 "C:\Program Files (x86)\BlackBox Component Builder 1.7\BlackBox.exe" /USE "C:\MultiBUGS-master"

I don't understand the steps too well either. You cannot add two paths to the Target box. So I cannot add:
"C:\Program Files (x86)\BlackBox Component Builder 1.7\BlackBox.exe" and "C:\MultiBUGS-master" together in the target box. Do you mean, put the first path in the target path and put the second in Start in? I tried this and it also doesn't seem to work.

Thanks

Robby

I used an R interface to run MultiBUGS. However, I got the message „BugsMaster:LinkingFailure
script command modelDistribute failed”.

I also installed two apllications - msmpisetup and msmpisdk- in my laptop.

Could you please let me know how to solve this issue?

After the burn-in period, sampling failed every time, even when running a small demo model

My computer's environment:
Win10 64bit with 2 core CPU and installed MS_MPI.
the whole multiBugs prog died When using modelDistribute(2) ,but can work by modelDistribute(1).
Anyone help? Thanks a lot!

Hi i have installed multibugs in my labtop and run my model but after around 30000 sample generating, multibugs stop and hanged. do you knew what is the problem and how to fux it. thank you for your help.