We may want to explore implementing N1 deconvolution: Appendix C of 1611.01446 This involves calculations closely related to issue #3 .
