Figure 8

Our inference procedure recovers the correct parameters and functions from Eq. (1). (A) The correct drift function (black dots) is recovered in the last iteration (yellow curve). First iteration results are shown for comparison (blue curves). (B) We see that the measured amplitude distribution \(Q_{C}\) is in fact a mixture of \(Q_{A}\) (yellow curve) and \(\tilde{\varXi }\ast Q_{B}\) (black dots, see Eq. (19)). (C) After deconvolving the latter, we do recover the correct \(Q_{B}\) (yellow curve). Note that in Case 2, the first and last iterations are confounded, as the true jump amplitudes are almost directly separable from those of FPs