Error al crear estimación BSR con modelo Probit

[lmperez]

Hola MMS, estoy intentando crear una estimación bayesiana de un modelo PRobit y me sale el siguiente error:

[BSR.Parse] Parsing BSR file C:/Users/lmperez/AppData/Roaming/tol/MMS/BSR/NewEstimation.1__1.0/2014.11.19_12.07.39/Priors/All.Priors/primary.bsr with 1719 bytes
[BSR.Parse] Succesfully build
[@BlockSampler @BSR.MainLinBlock.NewEstimation.1__1.0]  created with 8 parameters from 1 to 8
[@BlockSampler @BSR.SigmaBlock.NewEstimation.1__1.0]  created with 0 parameters from 9 to 8

ERROR: [8] No es posible aplicar + a matrices virtuales de dimensiones (0x0) (28x28)

[Call stack]
  [12] NameBlock BysMcmc::Bsr::Gibbs::@BlockSigma::New (Text name, Set NoiseDistrib, NameBlock mcmcPartialHandler, Real numBlock, Real firstCol, NameBlock config)
  [11] NameBlock BysMcmc::Bsr::Gibbs::@BsrMaster::New (Set modelDef, NameBlock mcmcPartialHandler, Real numBlock, Real firstCol, NameBlock config)
  [10] NameBlock BysMcmc::Bsr::Gibbs::GetCyclerFromParsed (Set parsed_, NameBlock mcmcPartialHandler, NameBlock config, NameBlock notifier)
  [9] NameBlock BysMcmc::Bsr::Gibbs::GetCyclerFromAscii (Text ascii_, Text moduleType, NameBlock mcmcPartialHandler, NameBlock config, NameBlock notifier)
  [8] NameBlock BysMcmc::Bsr::Gibbs::GetCycler (Anything modelDef, NameBlock mcmcPartialHandler, NameBlock config, NameBlock notifier)
  [7] NameBlock estimation::strategy::strategyBSR::_BuildCycler (Anything void)
  [6] Real estimation::strategy::strategyBSR::Prepare (Real void)
  [5] Real estimation::strategy::strategyBSR::Execute (Real void)
  [4] Real space::_E::Execute (Real void)
  [3] Real MMS::Menu::MainObject3.Execute (Anything instances_, Set extraData)
  [2] Real opt::inst::invoke (Text delegateOn, Anything objOrSelection)
  [1] Real GuiTools::MenuManager::invokeEntry (Text optionName, Text delegateOn, Set objOrSelection)

Warning: [5] NameBlock sigBlk se ha construido con 1 errores.
[@BSR.ArimaBlock.NewEstimation.1__1.0]  Preparing ARIMA block
  [@BSR.ArimaBlock.NewEstimation.1__1.0]  Original data length = 28
  [@BSR.ArimaBlock.NewEstimation.1__1.0]  Differenced data length = 28
  [@BSR.ArimaBlock.NewEstimation.1__1.0]  ARMA parameters = 0
[@BlockSampler @BSR.ArimaBlock.NewEstimation.1__1.0]  created with 0 parameters from 9 to 8
[@BlockSampler @BSR.MissingBlock.NewEstimation.1__1.0]  created with 0 parameters from 9 to 8

Warning: [6] NameBlock master se ha construido con 1 errores.
[@BlockSampler NonLinearFilterBlk::Output1::ProbitFilter]  created with 0 parameters from 9 to 8
[@BlockSampler NonLinearFilterBlk::Output2::ProbitFilter]  created with 0 parameters from 9 to 8
[@MasterSampler NewEstimation.1__1.0]  created with 8 parameters from 1 to 8
[@StrategyBSR::_BuildCycler] Se comienza a estimar con los valores iniciales 

Si en lugar de crear una estimación bayesiana creo una MLE, me estima correctamente. No encuentro donde está el error en la definición del modelo. ¿Podéis ayudarme? Os envío adjunto un objeto model de prueba.

Gracias de antemano