Nothing
R/genCpp_operatorLists.R
In nimble: MCMC, Particle Filtering, and Programmable Hierarchical Modeling
Defines functions
makeCallList
## This takes a character vector as the first argument and length-1
## character vector as the second argument. It returns a list with
## the first vector as names and the second argument as the value of
## each element. E.g. makeCallList(c('A','B'), 'foo') is equivalent
## to list(A = 'foo', B = 'foo')
makeCallList <- function(opList, call) {
ans <- rep(list(call), length(opList))
names(ans) <- opList
ans
}
ifOrWhile <- c('if','while')
## Following are a set of operators organized into categories that
## various processing steps use
binaryMidLogicalOperatorsLogical <- c('&','|')
binaryMidLogicalOperatorsComparison <- c('==','!=','<=','>=','<','>')
binaryMidLogicalOperators <- c(binaryMidLogicalOperatorsLogical,
binaryMidLogicalOperatorsComparison)
binaryMidDoubleOperators <- c('/', '^')
binaryMidPromoteNoLogicalOperators <- c('*','%%')
binaryMidOperators <- c(binaryMidDoubleOperators,
binaryMidPromoteNoLogicalOperators)
binaryLeftDoubleOperators <- c('pow','nimMod', 'pow_int')
binaryLeftPromoteOperators <- c('pmin','pmax','pairmin','pairmax')
binaryLeftLogicalOperators <- c( 'nimEquals')
binaryLeftOperators <- c(binaryLeftDoubleOperators,
binaryLeftPromoteOperators,
binaryLeftLogicalOperators)
binaryOperators <- c(binaryMidOperators,
binaryLeftOperators)
binaryOrUnaryOperators <- c('+','-')
unaryPromoteNoLogicalOperators <- c('abs','cube')
unaryIntegerOperators <- c('nimStep')
unaryLogicalOperators <- c('!', 'nimIsNA', 'nimIsNaN')
unaryDoubleOperators <- c('exp',
'log',
'logit',
'ilogit',
'probit',
'iprobit',
'sqrt', ## these do not go directly into cppOutputCalls. They should be direct C++ names or go through eigProxyCalls or eigProxyCallsExternalUnary
'gammafn',
'lgammafn', ## these also do not go direclty into eigenizeCalls but rather should be entered directly there for eigenize_cWiseUnaryEither, eigenize_cWiseUnaryArray or eigenize_cWiseUnaryMatrix
## 'lgamma1p',
'log1p',
'lfactorial',
'factorial',
'cloglog',
'icloglog',
'nimRound',
'ftrunc',
'ceil',
'floor',
'cos',
'sin',
'tan',
'acos',
'asin',
'atan',
'cosh',
'sinh',
'tanh',
'acosh',
'asinh',
'atanh')
unaryOperators <- c(unaryPromoteNoLogicalOperators,
unaryIntegerOperators,
unaryDoubleOperators,
unaryLogicalOperators)
unaryOrNonaryOperators <- list()
assignmentOperators <- c('<-','<<-','=')
reductionUnaryDoubleOperatorsEither <- c('mean', 'prod','squaredNorm')
reductionUnaryPromoteOperatorsEither <- c('min','max', 'sum')
reductionUnaryLogicalOperatorsEither <- c('any','all')
reductionUnaryOperatorsEither <- c(reductionUnaryDoubleOperatorsEither,
reductionUnaryPromoteOperatorsEither,
reductionUnaryLogicalOperatorsEither) # removed norm as not consistent between R and C
reductionUnaryOperatorsArray <- c('sd','var')
reductionUnaryOperators <- c(reductionUnaryOperatorsEither,
reductionUnaryOperatorsArray)
matrixSquareReductionOperators <- c('det','logdet')
reductionBinaryOperatorsEither <- c('inprod')
reductionBinaryOperators <- reductionBinaryOperatorsEither
coreRnonSeqBlockCalls <- c('nimNonseqIndexedd',
'nimNonseqIndexedi',
'nimNonseqIndexedb')
coreRmanipulationCalls <- c('nimC',
'nimRepd',
'nimRepi',
'nimRepb',
'nimSeqByD',
'nimSeqLenD',
'nimSeqByLenD',
'nimSeqByI',
'nimSeqLenI',
'nimSeqByLenI',
'nimDiagonalD',
'nimDiagonalI',
'nimDiagonalB',
'nimNewMatrixD',
'nimNewMatrixI',
'nimNewMatrixB')
nonNativeEigenCalls <- c('logdet',
'sd',
'var',
'inprod',
coreRmanipulationCalls,
coreRnonSeqBlockCalls)
matrixMultOperators <- c('%*%')
matrixFlipOperators <- c('t')
matrixSquareOperators <- c('chol','inverse')
nimbleListReturningOperators <- c('nimEigen',
'nimSvd',
'getDerivs_wrapper',
'nimDerivs_dummy') ## These use sizeNimbleListReturningFunction. Note that nimOptim and nimDerivs_calculate are handled separately.
matrixSolveOperators <- c('solve','forwardsolve','backsolve')
passThroughOperators <- c('return')
returnTypeCodes <- list(
double = 1L,
integer = 2L,
logical = 3L,
promote = 4L,
promoteNoLogical = 5L)
returnTypeHandling <- with(
returnTypeCodes,
c(list('(' = promote),
makeCallList(binaryMidLogicalOperators, logical),
makeCallList(binaryMidDoubleOperators, double),
makeCallList(binaryMidPromoteNoLogicalOperators, promoteNoLogical),
makeCallList(binaryLeftDoubleOperators, double),
makeCallList(binaryLeftPromoteOperators, promoteNoLogical),
makeCallList(binaryLeftLogicalOperators, logical),
makeCallList(binaryOrUnaryOperators, promoteNoLogical),
makeCallList(unaryPromoteNoLogicalOperators, promoteNoLogical),
makeCallList(unaryLogicalOperators, logical),
makeCallList(unaryIntegerOperators, integer),
makeCallList(unaryDoubleOperators, double),
makeCallList(reductionUnaryDoubleOperatorsEither, double),
makeCallList(reductionUnaryPromoteOperatorsEither, promoteNoLogical),
makeCallList(reductionUnaryLogicalOperatorsEither, logical),
makeCallList(reductionUnaryOperatorsArray, double),
makeCallList(matrixSquareReductionOperators, double),
makeCallList(reductionBinaryOperatorsEither, promoteNoLogical),
makeCallList(c(matrixMultOperators, matrixSquareOperators, matrixSolveOperators), double)))
## deliberately omitted (so they just return same type as input):
## matrixFlipOperators ('t')
midOperators <- as.list(
paste0(' ',
c(binaryMidOperators,
binaryMidLogicalOperators,
binaryOrUnaryOperators,
assignmentOperators),
' ')
)
names(midOperators) <- c(binaryMidOperators,
binaryMidLogicalOperators,
binaryOrUnaryOperators,
assignmentOperators)
midOperators <- c(midOperators,
list('$' = '$', '%*%' = ' %*% ', ':' = ':', '%o%' = '%o%'))
brackOperators <- list('[' = c('[',']'),
'[[' = c('[[',']]'))
## see distributions_processInputList for some relevant lists of distributions functions
callToSkipInEigenization <- c('copy',
'setValues',
'setValuesIndexRange',
'getValues',
'getValuesIndexRange',
'setSize',
'resize',
'getsize',
'size',
'resizeNoPtr',
'assert',
'return',
'blank',
'rankSample',
'nimArr_dmnorm_chol',
'nimArr_dmvt_chol',
'nimArr_dlkj_corr_cholesky',
'nimArr_dwish_chol',
'nimArr_dinvwish_chol',
'nimArr_dcar_normal',
'nimArr_dcar_proper',
'nimArr_dmulti',
'nimArr_dcat',
'nimArr_dinterval',
'nimArr_ddirch',
'nimArr_rmnorm_chol',
'nimArr_rmvt_chol',
'nimArr_rlkj_corr_cholesky',
'nimArr_rwish_chol',
'nimArr_rinvwish_chol',
'nimArr_rcar_normal',
'nimArr_rcar_proper',
'nimArr_rmulti',
'nimArr_rcat',
'nimArr_rinterval',
'nimArr_rdirch',
'calculate',
'calculateDiff',
'simulate',
'getLogProb',
'nimEquals',
'startNimbleTimer',
'endNimbleTimer')
## used for cppOutputs
## things here should have the inverse listing in the eigenizeTranslate list
eigProxyTranslate <- c(eigTranspose = 'transpose',
eigCos = 'cos',
eigSin = 'sin',
eigTan = 'tan',
eigAcos = 'acos',
eigAsin = 'asin',
eigExp = 'exp',
eigPow = 'pow',
eigLog = 'log',
eigCube = 'cube',
eiginprod = 'eigenInprod', ## need lowercase i for nonNativeEigCalls files
cwiseProduct = 'cwiseProduct',
cwiseQuotient = 'cwiseQuotient',
eigArray = 'array',
eigMatrix = 'matrix',
eigInverse = 'inverse',
setAll = 'setConstant',
eigEval = 'eval',
eigDiagonal = 'diagonal',
eigenBlock = 'block') ## created in makeEigenBlockExprFromBrackets called from sizeIndexingBracket
newEPT <- reductionUnaryOperators
names(newEPT) <- paste0('eig', reductionUnaryOperators)
eigProxyTranslate <- c(eigProxyTranslate, newEPT)
eigProxyTranslate[['eigmax']] <- 'maxCoeff'
eigProxyTranslate[['eigmin']] <- 'minCoeff'
eigProxyTranslate[['eigpmax']] <- 'max'
eigProxyTranslate[['eigpmin']] <- 'min'
newEPT <- reductionBinaryOperators
names(newEPT) <- paste0('eig', reductionBinaryOperators)
eigProxyTranslate <- c(eigProxyTranslate,
newEPT)
newEPT <- c(coreRmanipulationCalls,
coreRnonSeqBlockCalls)
names(newEPT) <- paste0('eig',
c(coreRmanipulationCalls,
coreRnonSeqBlockCalls)
)
eigProxyTranslate <- c(eigProxyTranslate, newEPT)
newEPT <- matrixSquareReductionOperators
names(newEPT) <- paste0('eig', matrixSquareReductionOperators)
eigProxyTranslate <- c(eigProxyTranslate, newEPT)
eigProxyTranslate[['eigdet']] <- 'determinant'
## nonNativeEigenProxyCalls should each appear in one of the other
## operatorLists feeding into eigProxyTranslate Then they are removed
## from eigProxyCalls so that in cppOutputCalls the
## nonNativeEigenProxyCalls can be generated differently (as a
## function, not a member function)
nonNativeEigenProxyCalls <- paste0('eig', nonNativeEigenCalls)
eigProxyCalls <- setdiff(names(eigProxyTranslate), nonNativeEigenProxyCalls)
## things here should have the inverse entry in the eigenizeTranslate
## list. Those are created automatically in genCpp_eigenization.R The
## C++ name here is (unfortunately) also used supposed to match the
## DSL keyword to be included in eigenizeCalls correctly
eigProxyTranslateExternalUnary <- list(
## (C++ name, arg type, return type, DSL name if different from C++ name) for std::ptr_fun<argtype, returntype>(fun name)
eigAtan = c('atan', 'double', 'double'),
eigCosh = c('cosh', 'double', 'double'),
eigSinh = c('sinh', 'double', 'double'),
eigTanh = c('tanh', 'double', 'double'),
eigAcosh = c('acosh', 'double', 'double'),
eigAsinh = c('asinh', 'double', 'double'),
eigAtanh = c('atanh', 'double', 'double'),
eigLogit = c('logit', 'double', 'double'),
eigIlogit = c('ilogit', 'double', 'double'),
eigProbit = c('probit', 'double', 'double'),
eigIprobit = c('iprobit', 'double', 'double'),
eigGammafn = c('gammafn', 'double', 'double'),
eigLgammafn = c('lgammafn', 'double', 'double'),
eigLog1p = c('log1p', 'double', 'double'),
eigLfactorial = c('lfactorial', 'double', 'double'),
eigFactorial = c('factorial', 'double', 'double'),
eigCloglog = c('cloglog', 'double', 'double'),
eigIcloglog = c('icloglog', 'double', 'double'),
eigNimRound = c('nimRound', 'double', 'double'),
eigFtrunc = c('ftrunc', 'double', 'double'),
eigCeil = c('ceil', 'double', 'double'),
eigFloor = c('floor', 'double', 'double'),
eigNimStep = c('nimStep', 'double', 'int'),
'eig!' = c('nimNot','bool','bool', '!'),
eigNimIsNA = c('nimIsNA', 'double', 'bool'),
eigNimIsNaN = c('nimIsNaN', 'double', 'bool')
)
eigProxyCallsExternalUnary <- names(eigProxyTranslateExternalUnary)
eigCalls <- c('matrix', 'array')
cppCasts = list(as.numeric = 'double', as.integer = 'int')
##http://en.cppreference.com/w/cpp/language/operator_precedence
## Used to decide when to put parentheses around LHS or RHS based on operator precendence.
operatorRank <- c(
list('<-' = 100, '^' = 4, '::' = 3),
makeCallList(c('*','/','%*%', '%%'), 5),
makeCallList(c('+', '-'), 6),
makeCallList(c('>','<','<=', '>='), 7),
makeCallList(c('==','!='), 8),
list('!' = 10), ## follows R's precedence order, not C's
list('&' = 13,
'|' = 14,
'&&' = 13,
'||' = 14)
)
nimDerivsPrependTypeOperators <- c("dnorm", "dpois", "dgamma", "dinvgamma", "dsqrtinvgamma",
"dexp_nimble", "dexp", "ddexp", "dlnorm", "dweibull",
"dbinom", "dbeta", "dchisq", "dlogis", "dt",
"dt_nonstandard", "nimArr_dmulti", "nimArr_dcat", "dnbinom", "dunif", "pairmax", "pairmin",
"nimArr_ddirch", "nimArr_dmvt_chol", "nimArr_dmnorm_chol",
"nimArr_dwish_chol", "nimArr_dinvwish_chol", "nimArr_dlkj_corr_cholesky",
"nimArr_dcar_normal", "nimArr_dcar_proper",
"nimStep", 'ilogit', 'icloglog', 'iprobit', 'probit', 'cloglog',
"nimEquals", "lgammafn", "gammafn", "lfactorial", "factorial",
"logit", "floor", "ceil", "nimRound", "ftrunc",
"cube", "inprod",
"nimStep", "dflat", "dhalfflat")
## Reflects distribution funs that support recycling rule with AD -- see nimDerivs_TMB.h.
recyclingRuleOperatorsAD <- c(
'dbinom', 'dexp_nimble', 'dnbinom', 'dpois', 'dchisq', 'dbeta', 'dnorm', 'dgamma',
'dinvgamma', 'ddexp', 'dlnorm', 'dlogis', 'dunif', 'dweibull',
## The following operators should work but are currently broken for various
## reasons which are explained in test-ADfunctions.R.
'dexp', 'dsqrtinvgamma', 'dt_nonstandard', 'dt', 'pow_int'
)
recyclingRuleOperatorsAD <- paste0(
recyclingRuleOperatorsAD,
'_RR_impl<MatrixXd>::',
recyclingRuleOperatorsAD,
'_RecyclingRule'
)
Try the nimble package in your browser
Any scripts or data that you put into this service are public.
nimble documentation built on Sept. 11, 2024, 7:10 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions makeCallList
## This takes a character vector as the first argument and length-1
## character vector as the second argument. It returns a list with
## the first vector as names and the second argument as the value of
## each element. E.g. makeCallList(c('A','B'), 'foo') is equivalent
## to list(A = 'foo', B = 'foo')
makeCallList <- function(opList, call) {
ans <- rep(list(call), length(opList))
names(ans) <- opList
ans
}
ifOrWhile <- c('if','while')
## Following are a set of operators organized into categories that
## various processing steps use
binaryMidLogicalOperatorsLogical <- c('&','|')
binaryMidLogicalOperatorsComparison <- c('==','!=','<=','>=','<','>')
binaryMidLogicalOperators <- c(binaryMidLogicalOperatorsLogical,
binaryMidLogicalOperatorsComparison)
binaryMidDoubleOperators <- c('/', '^')
binaryMidPromoteNoLogicalOperators <- c('*','%%')
binaryMidOperators <- c(binaryMidDoubleOperators,
binaryMidPromoteNoLogicalOperators)
binaryLeftDoubleOperators <- c('pow','nimMod', 'pow_int')
binaryLeftPromoteOperators <- c('pmin','pmax','pairmin','pairmax')
binaryLeftLogicalOperators <- c( 'nimEquals')
binaryLeftOperators <- c(binaryLeftDoubleOperators,
binaryLeftPromoteOperators,
binaryLeftLogicalOperators)
binaryOperators <- c(binaryMidOperators,
binaryLeftOperators)
binaryOrUnaryOperators <- c('+','-')
unaryPromoteNoLogicalOperators <- c('abs','cube')
unaryIntegerOperators <- c('nimStep')
unaryLogicalOperators <- c('!', 'nimIsNA', 'nimIsNaN')
unaryDoubleOperators <- c('exp',
'log',
'logit',
'ilogit',
'probit',
'iprobit',
'sqrt', ## these do not go directly into cppOutputCalls. They should be direct C++ names or go through eigProxyCalls or eigProxyCallsExternalUnary
'gammafn',
'lgammafn', ## these also do not go direclty into eigenizeCalls but rather should be entered directly there for eigenize_cWiseUnaryEither, eigenize_cWiseUnaryArray or eigenize_cWiseUnaryMatrix
## 'lgamma1p',
'log1p',
'lfactorial',
'factorial',
'cloglog',
'icloglog',
'nimRound',
'ftrunc',
'ceil',
'floor',
'cos',
'sin',
'tan',
'acos',
'asin',
'atan',
'cosh',
'sinh',
'tanh',
'acosh',
'asinh',
'atanh')
unaryOperators <- c(unaryPromoteNoLogicalOperators,
unaryIntegerOperators,
unaryDoubleOperators,
unaryLogicalOperators)
unaryOrNonaryOperators <- list()
assignmentOperators <- c('<-','<<-','=')
reductionUnaryDoubleOperatorsEither <- c('mean', 'prod','squaredNorm')
reductionUnaryPromoteOperatorsEither <- c('min','max', 'sum')
reductionUnaryLogicalOperatorsEither <- c('any','all')
reductionUnaryOperatorsEither <- c(reductionUnaryDoubleOperatorsEither,
reductionUnaryPromoteOperatorsEither,
reductionUnaryLogicalOperatorsEither) # removed norm as not consistent between R and C
reductionUnaryOperatorsArray <- c('sd','var')
reductionUnaryOperators <- c(reductionUnaryOperatorsEither,
reductionUnaryOperatorsArray)
matrixSquareReductionOperators <- c('det','logdet')
reductionBinaryOperatorsEither <- c('inprod')
reductionBinaryOperators <- reductionBinaryOperatorsEither
coreRnonSeqBlockCalls <- c('nimNonseqIndexedd',
'nimNonseqIndexedi',
'nimNonseqIndexedb')
coreRmanipulationCalls <- c('nimC',
'nimRepd',
'nimRepi',
'nimRepb',
'nimSeqByD',
'nimSeqLenD',
'nimSeqByLenD',
'nimSeqByI',
'nimSeqLenI',
'nimSeqByLenI',
'nimDiagonalD',
'nimDiagonalI',
'nimDiagonalB',
'nimNewMatrixD',
'nimNewMatrixI',
'nimNewMatrixB')
nonNativeEigenCalls <- c('logdet',
'sd',
'var',
'inprod',
coreRmanipulationCalls,
coreRnonSeqBlockCalls)
matrixMultOperators <- c('%*%')
matrixFlipOperators <- c('t')
matrixSquareOperators <- c('chol','inverse')
nimbleListReturningOperators <- c('nimEigen',
'nimSvd',
'getDerivs_wrapper',
'nimDerivs_dummy') ## These use sizeNimbleListReturningFunction. Note that nimOptim and nimDerivs_calculate are handled separately.
matrixSolveOperators <- c('solve','forwardsolve','backsolve')
passThroughOperators <- c('return')
returnTypeCodes <- list(
double = 1L,
integer = 2L,
logical = 3L,
promote = 4L,
promoteNoLogical = 5L)
returnTypeHandling <- with(
returnTypeCodes,
c(list('(' = promote),
makeCallList(binaryMidLogicalOperators, logical),
makeCallList(binaryMidDoubleOperators, double),
makeCallList(binaryMidPromoteNoLogicalOperators, promoteNoLogical),
makeCallList(binaryLeftDoubleOperators, double),
makeCallList(binaryLeftPromoteOperators, promoteNoLogical),
makeCallList(binaryLeftLogicalOperators, logical),
makeCallList(binaryOrUnaryOperators, promoteNoLogical),
makeCallList(unaryPromoteNoLogicalOperators, promoteNoLogical),
makeCallList(unaryLogicalOperators, logical),
makeCallList(unaryIntegerOperators, integer),
makeCallList(unaryDoubleOperators, double),
makeCallList(reductionUnaryDoubleOperatorsEither, double),
makeCallList(reductionUnaryPromoteOperatorsEither, promoteNoLogical),
makeCallList(reductionUnaryLogicalOperatorsEither, logical),
makeCallList(reductionUnaryOperatorsArray, double),
makeCallList(matrixSquareReductionOperators, double),
makeCallList(reductionBinaryOperatorsEither, promoteNoLogical),
makeCallList(c(matrixMultOperators, matrixSquareOperators, matrixSolveOperators), double)))
## deliberately omitted (so they just return same type as input):
## matrixFlipOperators ('t')
midOperators <- as.list(
paste0(' ',
c(binaryMidOperators,
binaryMidLogicalOperators,
binaryOrUnaryOperators,
assignmentOperators),
' ')
)
names(midOperators) <- c(binaryMidOperators,
binaryMidLogicalOperators,
binaryOrUnaryOperators,
assignmentOperators)
midOperators <- c(midOperators,
list('$' = '$', '%*%' = ' %*% ', ':' = ':', '%o%' = '%o%'))
brackOperators <- list('[' = c('[',']'),
'[[' = c('[[',']]'))
## see distributions_processInputList for some relevant lists of distributions functions
callToSkipInEigenization <- c('copy',
'setValues',
'setValuesIndexRange',
'getValues',
'getValuesIndexRange',
'setSize',
'resize',
'getsize',
'size',
'resizeNoPtr',
'assert',
'return',
'blank',
'rankSample',
'nimArr_dmnorm_chol',
'nimArr_dmvt_chol',
'nimArr_dlkj_corr_cholesky',
'nimArr_dwish_chol',
'nimArr_dinvwish_chol',
'nimArr_dcar_normal',
'nimArr_dcar_proper',
'nimArr_dmulti',
'nimArr_dcat',
'nimArr_dinterval',
'nimArr_ddirch',
'nimArr_rmnorm_chol',
'nimArr_rmvt_chol',
'nimArr_rlkj_corr_cholesky',
'nimArr_rwish_chol',
'nimArr_rinvwish_chol',
'nimArr_rcar_normal',
'nimArr_rcar_proper',
'nimArr_rmulti',
'nimArr_rcat',
'nimArr_rinterval',
'nimArr_rdirch',
'calculate',
'calculateDiff',
'simulate',
'getLogProb',
'nimEquals',
'startNimbleTimer',
'endNimbleTimer')
## used for cppOutputs
## things here should have the inverse listing in the eigenizeTranslate list
eigProxyTranslate <- c(eigTranspose = 'transpose',
eigCos = 'cos',
eigSin = 'sin',
eigTan = 'tan',
eigAcos = 'acos',
eigAsin = 'asin',
eigExp = 'exp',
eigPow = 'pow',
eigLog = 'log',
eigCube = 'cube',
eiginprod = 'eigenInprod', ## need lowercase i for nonNativeEigCalls files
cwiseProduct = 'cwiseProduct',
cwiseQuotient = 'cwiseQuotient',
eigArray = 'array',
eigMatrix = 'matrix',
eigInverse = 'inverse',
setAll = 'setConstant',
eigEval = 'eval',
eigDiagonal = 'diagonal',
eigenBlock = 'block') ## created in makeEigenBlockExprFromBrackets called from sizeIndexingBracket
newEPT <- reductionUnaryOperators
names(newEPT) <- paste0('eig', reductionUnaryOperators)
eigProxyTranslate <- c(eigProxyTranslate, newEPT)
eigProxyTranslate[['eigmax']] <- 'maxCoeff'
eigProxyTranslate[['eigmin']] <- 'minCoeff'
eigProxyTranslate[['eigpmax']] <- 'max'
eigProxyTranslate[['eigpmin']] <- 'min'
newEPT <- reductionBinaryOperators
names(newEPT) <- paste0('eig', reductionBinaryOperators)
eigProxyTranslate <- c(eigProxyTranslate,
newEPT)
newEPT <- c(coreRmanipulationCalls,
coreRnonSeqBlockCalls)
names(newEPT) <- paste0('eig',
c(coreRmanipulationCalls,
coreRnonSeqBlockCalls)
)
eigProxyTranslate <- c(eigProxyTranslate, newEPT)
newEPT <- matrixSquareReductionOperators
names(newEPT) <- paste0('eig', matrixSquareReductionOperators)
eigProxyTranslate <- c(eigProxyTranslate, newEPT)
eigProxyTranslate[['eigdet']] <- 'determinant'
## nonNativeEigenProxyCalls should each appear in one of the other
## operatorLists feeding into eigProxyTranslate Then they are removed
## from eigProxyCalls so that in cppOutputCalls the
## nonNativeEigenProxyCalls can be generated differently (as a
## function, not a member function)
nonNativeEigenProxyCalls <- paste0('eig', nonNativeEigenCalls)
eigProxyCalls <- setdiff(names(eigProxyTranslate), nonNativeEigenProxyCalls)
## things here should have the inverse entry in the eigenizeTranslate
## list. Those are created automatically in genCpp_eigenization.R The
## C++ name here is (unfortunately) also used supposed to match the
## DSL keyword to be included in eigenizeCalls correctly
eigProxyTranslateExternalUnary <- list(
## (C++ name, arg type, return type, DSL name if different from C++ name) for std::ptr_fun<argtype, returntype>(fun name)
eigAtan = c('atan', 'double', 'double'),
eigCosh = c('cosh', 'double', 'double'),
eigSinh = c('sinh', 'double', 'double'),
eigTanh = c('tanh', 'double', 'double'),
eigAcosh = c('acosh', 'double', 'double'),
eigAsinh = c('asinh', 'double', 'double'),
eigAtanh = c('atanh', 'double', 'double'),
eigLogit = c('logit', 'double', 'double'),
eigIlogit = c('ilogit', 'double', 'double'),
eigProbit = c('probit', 'double', 'double'),
eigIprobit = c('iprobit', 'double', 'double'),
eigGammafn = c('gammafn', 'double', 'double'),
eigLgammafn = c('lgammafn', 'double', 'double'),
eigLog1p = c('log1p', 'double', 'double'),
eigLfactorial = c('lfactorial', 'double', 'double'),
eigFactorial = c('factorial', 'double', 'double'),
eigCloglog = c('cloglog', 'double', 'double'),
eigIcloglog = c('icloglog', 'double', 'double'),
eigNimRound = c('nimRound', 'double', 'double'),
eigFtrunc = c('ftrunc', 'double', 'double'),
eigCeil = c('ceil', 'double', 'double'),
eigFloor = c('floor', 'double', 'double'),
eigNimStep = c('nimStep', 'double', 'int'),
'eig!' = c('nimNot','bool','bool', '!'),
eigNimIsNA = c('nimIsNA', 'double', 'bool'),
eigNimIsNaN = c('nimIsNaN', 'double', 'bool')
)
eigProxyCallsExternalUnary <- names(eigProxyTranslateExternalUnary)
eigCalls <- c('matrix', 'array')
cppCasts = list(as.numeric = 'double', as.integer = 'int')
##http://en.cppreference.com/w/cpp/language/operator_precedence
## Used to decide when to put parentheses around LHS or RHS based on operator precendence.
operatorRank <- c(
list('<-' = 100, '^' = 4, '::' = 3),
makeCallList(c('*','/','%*%', '%%'), 5),
makeCallList(c('+', '-'), 6),
makeCallList(c('>','<','<=', '>='), 7),
makeCallList(c('==','!='), 8),
list('!' = 10), ## follows R's precedence order, not C's
list('&' = 13,
'|' = 14,
'&&' = 13,
'||' = 14)
)
nimDerivsPrependTypeOperators <- c("dnorm", "dpois", "dgamma", "dinvgamma", "dsqrtinvgamma",
"dexp_nimble", "dexp", "ddexp", "dlnorm", "dweibull",
"dbinom", "dbeta", "dchisq", "dlogis", "dt",
"dt_nonstandard", "nimArr_dmulti", "nimArr_dcat", "dnbinom", "dunif", "pairmax", "pairmin",
"nimArr_ddirch", "nimArr_dmvt_chol", "nimArr_dmnorm_chol",
"nimArr_dwish_chol", "nimArr_dinvwish_chol", "nimArr_dlkj_corr_cholesky",
"nimArr_dcar_normal", "nimArr_dcar_proper",
"nimStep", 'ilogit', 'icloglog', 'iprobit', 'probit', 'cloglog',
"nimEquals", "lgammafn", "gammafn", "lfactorial", "factorial",
"logit", "floor", "ceil", "nimRound", "ftrunc",
"cube", "inprod",
"nimStep", "dflat", "dhalfflat")
## Reflects distribution funs that support recycling rule with AD -- see nimDerivs_TMB.h.
recyclingRuleOperatorsAD <- c(
'dbinom', 'dexp_nimble', 'dnbinom', 'dpois', 'dchisq', 'dbeta', 'dnorm', 'dgamma',
'dinvgamma', 'ddexp', 'dlnorm', 'dlogis', 'dunif', 'dweibull',
## The following operators should work but are currently broken for various
## reasons which are explained in test-ADfunctions.R.
'dexp', 'dsqrtinvgamma', 'dt_nonstandard', 'dt', 'pow_int'
)
recyclingRuleOperatorsAD <- paste0(
recyclingRuleOperatorsAD,
'_RR_impl<MatrixXd>::',
recyclingRuleOperatorsAD,
'_RecyclingRule'
)
Try the nimble package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.