亲爱的所有 TMB 热心用户或在 R 中使用 C++ 的用户将帮助我。
作为一名 TMB 婴儿用户，我对来自 TMB 的错误消息有疑问。
我发现我的代码中有错误。（第2个错误已修改）
结果来自 gdbsource()
1：Incomplete final line found on我的cpp文件（其实这是一个警告信息，而不是错误。我想知道为什么会这样。）
2：编译后出现以下错误消息：
Error in ev(obj$env$par): Wrong range component.
Error in ev(obj$env$par) : Wrong range component.
check1: 1
0.208386
9.70444e-005
6.95024e-005
8.33432e-005
7.90787e-005
5.96018e-005
7.0229e-005
9.99291e-005
0.000112216
0.000101964
9.13024e-005
8.54733e-005
8.77122e-005
8.77124e-005
8.36922e-005
6.80879e-005
0.000135715
0.00014771
6.97471e-005
5.73876e-005
5.2996e-005
7.63701e-005
check2: 0.367879
check3: 20.0855
check4: 0.00273944
check5: 0.0450492
check6: 0.0301974
Optimizing tape... Done
Error in ev(obj$env$par) : Wrong range component.
In addition: Warning messages:
1: In nlminb(model$par, model$fn, model$gr) : NA/NaN function evaluation
2: In he(par) : restarting interrupted promise evaluation
outer mgc: NaN
Error in nlminb(model$par, model$fn, model$gr) :
gradient function must return a numeric vector of length 5
Execution halted
[Inferior 1 (process 8152) exited with code 01]
C:\Users\POPDYN~1\AppData\Local\Temp\RtmpkL2d5A\file1f1048dc1743:4: Error in sourced command file:
No stack.
(gdb)
您对第二个错误有任何想法吗？
感谢您的努力、反馈、提示和提前帮助！
我附上了我的 cpp、R 代码和数据。
我想在此页面上发布更多内容。我的代码中的主要问题被认为是定义参数或向量。
在评论之后，我运行了包含“cout”部分的代码。和结果
CPP代码（修订）
template<class Type>
Type objective_function<Type>::operator() ()
{
//data
DATA_VECTOR(C);
DATA_VECTOR(I);
int n = C.size();
//free parameters
PARAMETER(logR);
PARAMETER(logK);
PARAMETER(logQ);
PARAMETER(logsdproc); //log(sd) in the process error;
PARAMETER(logSigma);
PARAMETER_VECTOR(P);
Type r = exp(logR);
Type k = exp(logK);
Type q = exp(logQ);
Type sdproc = exp(logsdproc);
Type sigma = exp(logSigma);
//derived parameters
vector<Type> Ihat(n);
Type f = 0.0;
Type fpen = 0.0;
Type tmpP;
P(0)=1.0;
for(int t=0; t<(n-1); t++) {
//P(t)=B(t)/k;
tmpP = P(t) + r*P(t)*(1-P(t))-C(t)/k;
P(t+1) = posfun(tmpP, Type(0.01), fpen);
f += fpen;
f -= dnorm(log(P(t+1)), log(tmpP), sdproc, true);
};
for(int t=0; t<n; t++) {
Ihat(t)=q*P(t)*k;
};
f -= sum(dnorm(log(I), log(Ihat), sigma, true));
REPORT(P);
REPORT(Ihat); // plot
REPORT(fpen);
std::cout << " check1: " << P << std::endl; //thank you Wave!
std::cout << " check2: " << r << std::endl;
std::cout << " check3: " << k << std::endl;
std::cout << " check4: " << q << std::endl;
std::cout << " check5: " << sdproc << std::endl;
std::cout << " check6: " << sigma << std::endl;
return f;
}
albacore <- read.table("albacore.csv", header=TRUE, sep=",")
albacore
names(albacore) <- c("t", "C", "I")
n=c(dim(albacore)[1]) #the number of Bs
parameters <- list(logR=-1.0, logK=3.0, logQ=-5.9, logsdproc=-3.1, logSigma=-3.5, P=rep(0.3,n));
parameters
require(TMB)
compile("scalbav2.cpp", "-O1 -g", DLLFLAGS="")
dyn.load(dynlib("scalbav2"))
library(TMB)
gdbsource("scalbav2.R", interactive=TRUE)
################################################################################
model<- MakeADFun(albacore, parameters, random="P", DLL="scalbav2")
model$par
length(parameters$P)
fit <- nlminb(model$par, model$fn, model$gr)
rep <- sdreport(model)
print(summary(rep))
数据
year catch cpue
1967 15.9 61.89
1968 25.7 78.98
1969 28.5 55.59
1970 23.7 44.61
1971 25.0 56.89
1972 33.3 38.27
1973 28.2 33.84
1974 19.7 36.13
1975 17.5 41.95
1976 19.3 36.63
1977 21.6 36.33
1978 23.1 38.82
1979 22.5 34.32
1980 22.5 37.64
1981 23.6 34.01
1982 29.1 32.16
1983 14.4 26.88
1984 13.2 36.61
1985 28.4 30.07
1986 34.6 30.75
1987 37.5 23.36
1988 25.9 22.36
1989 25.3 21.91
我还附上了 sessioninfo：
R version 3.5.0 (2018-04-23)
Platform: i386-w64-mingw32/i386 (32-bit)
Running under: Windows >= 8 (build 9200)
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] TMB_1.7.13
loaded via a namespace (and not attached):
[1] compiler_3.5.0 Matrix_1.2-14 tools_3.5.0 grid_3.5.0
[5] lattice_0.20-35