Gravitational-wave memory and pulsar timing arrays
This paper argues that the memory from mergers of supermassive black holes is in principle detectable by PTAs, however the expected rate is very small (perhaps one will need to time pulsars for several generations before this will be possible). Similar work was simultaneously and independently done by Maxim Pshirkov, D. Baskaran, and Konstantin (Kostya) Postnov and by Naoki Seto.
The methodological interest in our paper is that it presents, for the first time, the time-domain Wiener-Filter formalism as applied to PTAs. The peculiarity of this situation is that one needs to deal with the corrections to the pulsar timing residuals due to variations in the pulsar timing-model parameters.
Detecting Gravitational-Wave Memory with LIGO: Implications of GW150914
This paper argues that the memory from mergers of stellar-mass black holes is potentially detectable by LIGO. However, the memory signal from a single merger is too weak to be detectable, and one needs collective evidence from many strong mergers to prove the existence of memory. Like with PTAs, one needs a lot of time and patience. Therefore my personal bet is that LISA will be the first instrument to see Gravitational-Wave memory, despite its planned launch in the now-distant 2034.