Changing one cluster into 2 clusters without changing initial parameters

I have a code that runs through a matrix of 0's and 1's. Each 1 is represented as a vector in a field. The code selects clusters of vectors, and I am rejecting vectors that are outside of a standard deviation from the center of mass of each cluster. However, I have encountered a problem. Some clusters look like they should be 2 clusters, even though my code has selected them as one. When this happens, the center of mass is in between the two clusters, and most of the vectors are then rejected. Any suggestions on how I can handle this problem without changing the initial parameters for how the clusters are selected?

Answer 1

[Walter Roberson]

You have an algorithm we have never seen the code for and whose workings you have only fractionally described. Your algorithm does not give you the results you want. You ask if we can tell you how to get the results you want, but without changing any of the inputs to the algorithm (not that we know what the inputs are anyhow.)

Seems to me that the appropriate answer here is "NO, we have no suggestions other than 'Don't Do That!'"