We derive a combinatorial sufficient condition for a partial correlation hypersurface in the parameter space of a directed Gaussian graphical model to be nonsingular, and speculate on whether this condition can be used in algorithms for learning the graph. Since the condition is fulfilled in the case of a complete DAG on any number of vertices, the result implies an affirmative answer to a question raised by Lin–Uhler–Sturmfels–Bühlmann.
Revised:
Accepted:
Published online:
DOI: 10.5802/alco.44
Keywords: partial correlation, Gaussian graphical models, trek separation
Draisma, Jan 1
@article{ALCO_2019__2_3_439_0, author = {Draisma, Jan}, title = {Partial correlation hypersurfaces in {Gaussian} graphical models}, journal = {Algebraic Combinatorics}, pages = {439--446}, publisher = {MathOA foundation}, volume = {2}, number = {3}, year = {2019}, doi = {10.5802/alco.44}, mrnumber = {3968746}, zbl = {07066883}, language = {en}, url = {https://alco.centre-mersenne.org/articles/10.5802/alco.44/} }
TY - JOUR AU - Draisma, Jan TI - Partial correlation hypersurfaces in Gaussian graphical models JO - Algebraic Combinatorics PY - 2019 SP - 439 EP - 446 VL - 2 IS - 3 PB - MathOA foundation UR - https://alco.centre-mersenne.org/articles/10.5802/alco.44/ DO - 10.5802/alco.44 LA - en ID - ALCO_2019__2_3_439_0 ER -
Draisma, Jan. Partial correlation hypersurfaces in Gaussian graphical models. Algebraic Combinatorics, Volume 2 (2019) no. 3, pp. 439-446. doi : 10.5802/alco.44. https://alco.centre-mersenne.org/articles/10.5802/alco.44/
[1] Positivity for Gaussian graphical models, Adv. Appl. Math., Volume 50 (2013) no. 5, pp. 661-674 | DOI | MR | Zbl
[2] Lectures on algebraic statistics, Oberwolfach Seminars, 39, Birkhäuser, 2009, viii+271 pages | MR | Zbl
[3] Binomial determinants, paths, and hook length formulae, Adv. Math., Volume 58 (1985), pp. 300-321 | DOI | MR | Zbl
[4] Hypersurfaces and their singularities in partial correlation testing, Found. Comput. Math., Volume 14 (2014) no. 5, pp. 1079-1116 | MR | Zbl
[5] Causation, prediction, and search. With additional material by David Heckerman, Christopher Meek, Gregory F. Cooper and Thomas Richardson., MIT Press, 2001, xxii+496 pages | Zbl
[6] Trek separation for Gaussian graphical models, Ann. Stat., Volume 38 (2010) no. 3, pp. 1665-1685 | DOI | MR | Zbl
[7] The method of path coefficients, Ann. Math. Stat., Volume 5 (1934), pp. 161-215 | DOI | Zbl
Cited by Sources: