Parfor loop with one iteration

My overall question is:

Is there something fundamental about parfor loops that cannot deal with a single iteration (i.e. parfor ii = 1:1)?

Here is some context:

I use a parfor to simultaneously computer multiple inverse models at once (different "chains" of the same model). The number of chains I iterate through at this first step is hard coded (i.e. parfor = 1:6). After the parfor loop completes, I apply a criteria to only keep model results with acceptably low misfit to the data. I then repeat the parfor loop to redo chains with resulting high misfit models. This is done in a while loop until all models have acceptably low misfits.

The problem I am running into, is that when a single chain is high-misfit & must repeat (i.e. parfor ii = 1:1), the parfor loop will do it's thing with no obvious issue, but some of the information produced in this iteration is identical to the initial high-misfit chain. This does not seem to be an issue if the parfor loop is redoing multiple chains (i.e. parfor ii = 1:2). I understand that parfor loops are not generally intended for a single iteration loop, but because this is within a while loop (for all chains having acceptably low misfits), inevitably sometimes the parfor loop will have a single iteration.

[Edric Ellis]

I know of no problems running a parfor loop with a range of size 1. Can you post any simple reproduction steps that demonstrate the problem? Are you using any random number generation routines in the body of your loop?

[Hannah Krueger]

Thanks for the response Edric!

The function that initiates the model does use rand in a few places, so I think that I should put rng('shuffle') at the beginning of this function.

But, Matlab isn't restarting between these different instances of the parfor loop, so why would rand be acting like it had?