Error in if (abs(ll.old - ll.new) < convergence) run <- FALSE

[周露阳]

My code for LMS are as follows:

+++++++++++++++++

library("nlsem")

model <- specify_sem(num.x = 20, num.y = 3, num.xi = 7, num.eta = 1,

                     xi = "x1-x3, x4-x6, x7-x9, x10-x12, x13-x15, x16-x18, x19-x20", 

                     eta = "y1-y3", interaction = "xi1:xi2")

load("part2b.RData")

set.seed(11111)

start <- runif(count_free_parameters(model))

res <- em(model, part2b, start, qml = F, convergence=0.01, verbose=TRUE, max.iter=99999, neg.hessian = TRUE)

summary(res)

coef(res)

logLik(res)

AIC(res)

BIC(res)

++++++++++++++++++

After being run, the results are:

+++++++++++++++++++

......

> res <- em(model, part2b, start, qml = F, convergence=0.01, verbose=TRUE, max.iter=99999, neg.hessian = TRUE)

-----------------------------------

Starting EM-algorithm for singleClass 

Convergence:  0.01 

-----------------------------------

-----------------------------------

Iteration 1 

Doing expectation-step 

Doing maximization-step 

Results of maximization 

Loglikelihood: -Inf

Convergence message: 0

Number of iterations: 1

----------------------------------- 

Iteration 2 

Doing expectation-step 

Doing maximization-step 

Results of maximization 

Loglikelihood: -Inf

Convergence message: 0

Number of iterations: 1

----------------------------------- 

Error in if (abs(ll.old - ll.new) < convergence) run <- FALSE : 

  missing value where TRUE/FALSE needed

In addition: Warning message:

In value[[3L]](cond) :

  Starting parameters for Phi are not positive definite. Identity matrix was used instead.

> 

> summary(res)

......

+++++++++++++++++++++++++++++++

How can I do on my codes following the warnings?

When running within Mplus, everything is OK.

[周露阳]

Bellow are the codes in lavaan grammar:

++++++++++++++++++

library("nlsem")

modLMS.model <- "

                 xi1 =~ x1 + x2 + x3

                 xi2 =~ x4 + x5 + x6

                 xi3 =~ x7 + x8 + x9 

                 xi4 =~ x10 + x11 + x12

                 xi5 =~ x13 + x14 + x15

                 xi6 =~ x16 + x17 + x18

                 xi7 =~ x19 + x20

                 eta =~ y1 + y2 + y3

                 eta ~ xi1 + xi2 + xi3 + xi4 + xi5 + xi6 + xi7 + xi1:xi2

               "

model <- lav2nlsem(modLMS.model)

load("part2b.RData")

set.seed(1111)

start <- runif(count_free_parameters(model))

res <- em(model, part2b, start, qml = F, convergence=0.01, verbose=TRUE, max.iter=99999, neg.hessian = TRUE, logger=TRUE, optimizer="nlminb")

summary(res)

coef(res)

logLik(res)

AIC(res)

BIC

dev.new()

plot(res)

+++++++++++++++++++++

But I faced with the same problem as shown below:

++++++++++++++++++++

> res <- em(model, part2b, start, qml = F, convergence=0.01, verbose=TRUE, max.iter=99999, neg.hessian = TRUE, logger=TRUE, optimizer="nlminb")

++++++++++++++++