Volume 103, Number 1, July 2013
|Number of page(s)||6|
|Published online||23 July 2013|
Reduction of interaction delays in networks
1 Humboldt - University of Berlin, Institute of Mathematics -Unter den Linden 6, 10099 Berlin, Germany, EU
2 Université Montpellier 2, Institut de Mathématiques et de Modélisation de Montpellier (I3M) Place Eugène Bataillon, 34095 Montpellier Cedex, France, EU
Received: 19 April 2013
Accepted: 26 June 2013
Delayed interactions are a common property of coupled natural systems and therefore arise in a variety of different applications. For instance, signals in neural or laser networks propagate at finite speed giving rise to delayed connections. Such systems are often modeled by delay differential equations with discrete delays. In realistic situations, these delays are not identical on different connections. We show that by a componentwise timeshift transformation it is often possible to reduce the number of different delays and simplify the models without loss of information. We identify dynamic invariants of this transformation, determine its capabilities to reduce the number of delays and interpret these findings in terms of the topology of the underlying graph. In particular, we show that networks with identical sums of delay times along the fundamental semicycles are dynamically equivalent and we provide a normal form for these systems. We illustrate the theory using a network motif of coupled Mackey-Glass systems with 8 different time delays, which can be reduced to an equivalent motif with three delays.
PACS: 02.30.Ks – Delay and functional equations / 02.10.Ox – Combinatorics; graph theory / 87.19.lj – Neuronal network dynamics
© EPLA, 2013
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.