P2MP is a Hub and Spoke topology. Usually all neighbors are under the same ip network, and are able to send traffic to each other, but always through the hub router. Neighbors could auto discover each others.
Under a pure PTP, hub neighbors can only "speak" with the hub, and have no knowledge about other neighbors, and no way to discover this.
P2MP is useful if you want to automatically discover new hub neighbors. Under Frame Relay you could request to add a new dlci to your hub link, install a new router in a remote site with a new fr interface, this neighbor will automatically discover the dlci it have to use, hub router will discover a new dlci and neighbor, and neighbor would autoconfigure its ip address through dhcp from the hub router, and will automatically be connected to the hub.
Dlci are automatically discovered through lmi, and its broadcast capability allow to discover the other side ip address, use ppp over fr, etc. All these methods will give an ip address to the new router.
With few configuration you could add remote sites.
PTP, need explicit configuration of new neighbors. You need new subnets for every new router, and of course know the dlci this ptp link is going to use. More secure because if you dont configure the new ptp interface there will not be connectivity through new dlci. But you have to do all the configuration.
Of course you hub could be configured under ospf as PTMP, and your spokes as PTP, but this is not related to the physical or logical interface configuration, have only sense under ospf. Have to be carefull because each link type (broadcast, nonbroadcast, ptp, ptmp) have its own hello, retransmit, dead timers, etc. Mismatch will refuse new neighbor relationships.
Br
Alex.