A scalable and accurate framework for self-calibrating null depth retrieval using neural posterior estimation

Open Access

Table 2

Results of null depth retrieval on synthetic data (μ_OPD and σ_OPD in nanometer).

Parameter	Truth	NPE retrieval	NSC retrieval
N_a	0.0063	${0.0061}_{- 0.0003}^{+ 0.0002}$ $0.0061^{+0.0002}_{-0.0003}$	${0.0065}_{- 0.0009}^{+ 0.0008}$ $0.0065^{+0.0008}_{-0.0009}$
μ_OPD	300	${308.82}_{- 11.89}^{+ 11.50}$ $308.82^{+11.50}_{-11.89}$	${296.56}_{- 31.01}^{+ 27.96}$ $296.56^{+27.96}_{-31.01}$
σ_OPD	250	${241.75}_{- 7.58}^{+ 8.16}$ $241.75^{+8.16}_{-7.58}$	${240.46}_{- 18.10}^{+ 21.12}$ $240.46^{+21.12}_{-18.10}$

Notes. Medians of posterior distributions are used as point estimates. Superscripts and subscripts give the 16^th and 84^th percentiles (i.e., the 68% credible interval).

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.