OpenMx Example Scripts

Here are example scripts in OpenMx for a variety of models. Most models will utilize twin samples.

Common Path Model in One Twin Sample

Linear Growth Model in One Twin Sample

OpenMx Optimizer Codes

• A value of −1 means that the optimizer found itself stuck in a location where the objective function could not be calculated, and could not find a way out. This most often happens if the starting values make the calculation impossible. Estimates from runs with output$status1 equal to -1 can never be trusted, and the job must be re-run with different starting values, or a corrected model specification.

• 0 indicates a successful optimization--no error returned.

• 1 means that an optimal solution was found (like status 0), but that the sequence of iterates did not converge. In practice, this is the same as status 0, but involves slightly different criteria and is distinguished for technical reasons. This code is labeled (Mx status GREEN).

• 2 or 3 means that the box constraints or non-linear constraints, respectively, could not be satisfied.

• 4 means that the iteration limit was reached with no solution found. You can use mxOptions() to set a higher iteration limit, or just run mxRun() using the output—it will restart from the most recent set of estimates, and make another run of the same number of iterations.

• 5 means that the Hessian at the solution is not convex. There is likely a better solution, but the optimizer is stuck in a region of confusing geometry (like a saddle point).

To put another way thanks to Rob Kirkpatrick: The Hessian matrix (the matrix of second partial derivatives of the fitfunction w/r/t the free parameters) is not positive-definite at the solution. To put it another way, the fitfunction is not locally convex at the solution. This is a sign that the solution does not meet second-order conditions for being a local minimum of the fitfunction.

• 6 means that gradient is not close enough to zero but the optimizer could find no way to improve the estimate. A variety of situations can lead to this outcome. Check whether the model is identified (mxCheckIdentification). It is also possible that gradients are just slightly larger than the arbitrary threshold used by OpenMx. The estimates resulting from this run MIGHT not be optimal estimates, and it would be risky to treat them as solutions without further investigation, so this code is labeled (Mx status RED). Sometimes re-running the model from its solution [firstRun <- mxRun(myModel) followed by secondRun <- mxRun(firstRun)] is sufficient to make code Red disappear.

To put another way thanks to Mike Neale: A value of 6 means that optimality conditions could not be reached, and the optimizer could find no way to improve the estimate. It often implies either a mistake in the model specification or starting values in an intractable range. The estimates resulting from this run are often NOT optimal estimates, and should not be treated as solutions, so this code is labeled (Mx status RED). The job should be re-run, either from the final solution, or from new starting values, or both. If multiple different starting values converge on the same solution (same parameter estimates and fit function) then confidence that the minimum has been found increases.

• 7 means that the analytic derivatives are incorrect. This should be reported to developers.

• 9 means that an invalid parameter was passed to the optimizer. This should be reported to developers.

• 10 means that the parameter vector is outside the feasible set. This most often happens when the starting values are infeasible. Given feasible starting values, optimizer should under no circumstances escape the feasible set.