Peter G. Jensen, Kim G. Larsen, Marius Mikučionis*

*Corresponding author for this work

Research output: Contribution to book/anthology/report/conference proceedingBook chapterResearchpeer-review


In this paper we model and solve the popular game Wordle using Uppaal Stratego. We model three different game-modes in terms of POMDPs, with more than 12,000 controllable actions. These constitute by far the largest models ever presented to Uppaal Stratego. Our experimental evaluation is encouraging: e.g. in the hard game-mode the partitioning-refinement learning method of Uppaal Stratego reduces the expected number of guesses from a baseline of 7.67 to 4.40 using 1 million training episodes. To better understand the convergence properties of our learning method we also study reduced versions of Wordle.

Original languageEnglish
Title of host publicationA Journey from Process Algebra via Timed Automata to Model Learning : Essays Dedicated to Frits Vaandrager on the Occasion of His 60th Birthday
EditorsNils Jansen, Mariëlle Stoelinga, Petra van den Bos
Number of pages23
Publication date2022
ISBN (Print)978-3-031-15628-1
ISBN (Electronic)978-3-031-15629-8
Publication statusPublished - 2022
SeriesLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13560 LNCS

Bibliographical note

Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.


Dive into the research topics of 'PlayingWORDLE with UPPAAL STRATEGO'. Together they form a unique fingerprint.

Cite this