Abstract
The asymptotic behavior of the iterative proportional fitting procedure (IPF procedure) is analyzed comprehensively. Given a nonnegative matrix as well as row and column marginals the IPF procedure generates a sequence of matrices, called the IPF sequence, by alternately fitting rows and columns to match their respective marginals. We prove that the IPF sequence has at most two accumulation points. They originate as the limits of the even-step subsequence, and of the odd-step subsequence. The well-known IPF convergence criteria are then retrieved easily. Our proof is based on Csiszár’s and Tusnády’s (Stat Decis Suppl Issue 1:205–237, 1984) results on the interplay of the I-divergence geometry and alternating minimization procedures.
Similar content being viewed by others
References
Bauschke H, Combettes P (2011) Convex analysis and monotone operator theory in Hilbert spaces. Springer, New York
Bregman L (1967) Proof of the convergence of Sheleikhovskii’s method for a problem with transportation constraints. USSR Comput Math Math Phys 7:191–204
Brown J, Chase P, Pittenger A (1993) Order independence and factor convergence in iterative scaling. Linear Algebra Appl 190:1–38
Cramer E (2000) Probability measures with given marginals and conditionals: I-projections and conditional iterative proportional fitting. Stat Decis 18:311–329
Csiszár I (1964) Über topologische und metrische Eigenschaften der relativen Information der Ordnung \(\alpha \). In: Transactions of the third Prague conference on information theory, statistical decision functions and random processes. Publishing House of the Czechoslovak Academy of Sciences, Prague, pp 63–73
Csiszár I (1975) I-divergence geometry of probability distributions and minimization problems. Ann Probab 3:146–158
Csiszár I, Tusnády G (1984) Information geometry and alternating minimization procedures. Stat Decis Suppl Issue 1:205–237
Deming W, Stephan F (1940) On a least squares adjustment of a sampled frequency table when the expected marginal totals are known. Ann Math Stat 11:427–444
Jiroušek R, Vomlel J (1994) Inconsistent knowledge integration in a probabilistic model. In: Proceedings of the workshop mathematical models for handling partial knowledge in A.I., Plenum Publ Corp, Erice, Sicily, pp 263–270
Knight P (2008) The Sinkhorn-Knopp algorithm: convergence and applications. SIAM J Matrix Anal Appl 30:261–275
Kruithof J (1937) Telefoonverkeersrekening. De Ingenieur 52:E15–E25
Kullback S (1968) Probability densities with given marginals. Ann Math Stat 39(4):1236–1243
Lauritzen S (1996) Graphical models. Clarendon Press, Oxford
McCord M, Mishalani R, Goel P, Strohl B (2010) Iterative proportional fitting procedure to determine bus route passenger origin-destination flows. Trans Res Record J Trans Res Board 2145:59–65
Pinsker M (1964) Information and information stability of random variables and processes. Holden-Day, San Francisco
Pukelsheim F (2012) Biproportional matrix scaling and the iterative proportional fitting procedure, Ann Oper Res (submitted)
Reid M, Williamson R (2009) Generalised Pinsker inequalities. arXiv:09061244v1
Rockafellar R (1972) Convex analysis. Princeton University Press, Princeton
Rüschendorf L (1995) Convergence of the iterative proportional fitting procedure. Ann Stat 23:1160–1174
Vomlel J (2004) Integrating inconsistent data in a probabilistic model. J Appl Non Class Log 14:365–386
Zhang S, Peng Y, Wang X (2008) An efficient method for probabilistic knowledge integration. In: 2008 20th IEEE international conference on tools with artificial intelligence, vol 2. IEEE, pp 179–182
Acknowledgments
We are very grateful to our advisor Friedrich Pukelsheim for continuing support. We thank our friend and colleague Kai-Friederike Oelbermann from our Augsburg group for her valuable remarks when discussing the work. We are thankful to Fero Matúš and Ludger Rüschendorf for their assistance with the pertinent literature. We thank an anonymous referee for valuable remarks on an earlier version of this paper, which in particular helped us condense Sects. 2 and 3.
Author information
Authors and Affiliations
Corresponding author
Additional information
This research has been supported by the Elite Network of Bavaria through its graduate program TopMath and by the TUM Graduate School through its Thematic Graduate Center TopMath.
Rights and permissions
About this article
Cite this article
Gietl, C., Reffel, F.P. Accumulation points of the iterative proportional fitting procedure. Metrika 76, 783–798 (2013). https://doi.org/10.1007/s00184-012-0415-7
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00184-012-0415-7
Keywords
- Iterative proportional fitting
- Accumulation points
- I-divergence
- I-projection
- Alternating minimization
- Distributions with given marginals