The jvm can have only one data schema. It is not possible to have multiples and switch between them.
This question was posed back in 2015 and here is the engineers response.
"As shipped, this is possible only if you put each application on a separate JVM -- and if your operators NEVER switch access groups / application. Remember - all applications share asignments, attachments, locks, queues; operators; access groups and other data-*.
What is purpose of schema segregation? It is possible to hack 'virtual schemas' in the database where PRPC 'sees' one data* schema but app specific tables are in a unique schema for ownership and storage purposes."
I'm not sure if anything in Pega's architecture has changed to make this possible - but the above analysis was done in regards to Pega 7.1.7.