Abstract
Many databases have been filled with the chemical reactions found in scientific publications and the associated information (efficiency, chemical products involved...). They can be used to define functions representing costs such as the ecoligical impact of the reactions. A major challenge is to use computer driven optimization in order to improve synthesis process and to provide algorithms to help determining a minimum cost pathway (series of reactions) for the synthesis of a molecule.
As, the classical Petri nets do not allows us to consider the optimization component, a weighted model has to be defined and the complexity of the associated problems studied. In this paper we introduce the weighted Petri nets in which each transition is associated with a weight. We define the Minimum Weight Synthesis Problem: find a minimum weight series of transitions to fire to produce a given target component. It mainly differ from classical coverability as it is an optimization problem.
We prove that this problem is EXPSPACE-Complete and that there is no polynomial approximation even when both in and outdegree are fixed to two and the target state is a single component. We also consider a more constraint version of the problem limiting the number of fired transitions. We prove this problem falls into PSPACE and the parametrized versions into XP but it remains not approximable.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Reaxys. Elsevier (2016). https://www.elsevier.com/solutions/reaxys
Abdulla, P.A., Mayr, R.: Minimal cost reachability/coverability in priced timed petri nets. In: Alfaro, L. (ed.) FoSSaCS 2009. LNCS, vol. 5504, pp. 348–363. Springer, Heidelberg (2009). doi:10.1007/978-3-642-00596-1_25
Abdulla, P.A., Mayr, R.: Computing optimal coverability costs in priced timed Petri nets. In: Proceedings of the Symposium on Logic in Computer Science, pp. 399–408 (2011)
Abdulla, P.A., Mayr, R.: Priced timed Petri nets. Logical Methods Comput. Sci. 9(4), 1–51 (2013)
Andersen, J.L., Flamm, C., Merkle, D., Stadler, P.F.: Maximizing output and recognizing autocatalysis in chemical reaction networks is NP-complete. J. Syst. Chem. 3(1), 1–9 (2012). http://www.jsystchem.com/content/3/1/1
Angeli, D., De Leenheer, P., Sontag, E.: A Petri Net approach to the study of persistence in chemical reaction networks. Math. Biosci. 210(2), 598–618 (2007). http://arxiv.org/abs/q-bio/0608019
Bøgevig, A., Federsel, H.J., Huerta, F., Hutchings, M.G., Kraut, H., Langer, T., Löw, P., Oppawsky, C., Rein, T., Saller, H.: Route design in the 21st century: the IC SYNTH software tool as an idea generator for synthesis prediction. Org. Process Res. Dev. 19(2), 357–368 (2015)
Downey, R., Fellows, M.: Parameterized Complexity, vol. 3. Springer, New York (1999). https://www.dagstuhl.de/Reports/01/01311.pdf
Eigner-Pitto, V., Huerta, F.F., Hutchings, M.G., Saller, H., Loew, P.: Reaction prediction tools for both idea generation in new synthesis route planning and for de novo molecule design. In: 12th German Conference on Chemoinformatics (2016)
Finkel, A.: The minimal coverability graph for Petri nets. In: Rozenberg, G. (ed.) ICATPN 1991. LNCS, vol. 674, pp. 210–243. Springer, Heidelberg (1993). doi:10.1007/3-540-56689-9_45
Finkel, A., Leroux, J.: Recent and simple algorithms for Petri nets. Softw. Syst. Model. 14(2), 719–725 (2015)
Genrich, H., Küffner, R., Voss, K.: Executable Petri net models for the analysis of metabolic pathways. Int. J. Softw. Tools Technol. Transf. 3(4), 394–404 (2001)
Hastings, J., de Matos, P., Dekker, A., Ennis, M., Harsha, B., Kale, N., Muthukrishnan, V., Owen, G., Turner, S., Williams, M., Steinbeck, C.: The chebi reference database and ontology for biologically relevant chemistry: enhancements for 2013. Nucleic Acids Res. 41(D1), D456 (2012). http://dx.doi.org/10.1093/nar/gks1146
Heiner, M.: Preface: Petri nets for systems and synthetic biology. Natural Comput. 10(3), 987–992 (2011)
Hofestädt, R.: A Petri net application to model metabolic processes. Syst. Anal. Model. Simul. 16(2), 113–122 (1994)
Horvath, D., Jeandenans, C.: Molecular similarity and virtual screening. In silico methods to retrieve active analogs in the context of discovering therapeutic compounds. Actual. Chim. 9, 64–67 (2000)
Johnson, A.P., Marshall, C.: Starting material oriented retrosynthetic analysis in the LHASA program. 2. Mapping the SM and target structures. J. Chem. Inf. Comput. Sci. 32(5), 418–425 (1992)
Jones, N.D., Landweber, L.H., Edmund Lien, Y.: Complexity of some problems in Petri nets. Theoret. Comput. Sci. 4(3), 277–299 (1977)
Koch, I.: Petri nets-a mathematical formalism to analyze chemical reaction networks. Mol. Inform. 29(12), 838–843 (2010)
Lipton, R.J.: The Reachability Problem Requires Exponential Space. Technical report, Yale Research Report #63 (1976)
Nouleho, S., Barth, D., David, O., Watel, D., Weisser, M.: A new definition of molecule similarity to determine molecular construction. Poster in 12th German Conference on Chemoinformatics (2016)
Petri, C.A.: Kommunikation mit Automaten. Ph.D. Thesis, Universität Hamburg (1962). http://epub.sub.uni-hamburg.de/informatik/volltexte/2011/160/
Pietrzak, K.: On the parameterized complexity of the fixed alphabet shortest common supersequence and longest common subsequence problems. J. Comput. Syst. Sci. 67(4), 757–771 (2003)
Rackoff, C.: The covering and boundedness problems for vector addition systems. Theoret. Comput. Sci. 6(2), 223–231 (1978). http://linkinghub.elsevier.com/retrieve/pii/0304397578900361
Reddy, V.N., Mavrovouniotis, M.L., Liebman, M.N.: Petri net representations in metabolic pathways. In: Proceedings of the International Conference on Intelligent Systems for Molecular Biology, vol. 1, pp. 328–336 (1993)
Savitch, W.J.: Relationships between nondeterministic and deterministic tape complexities. J. Comput. Syst. Sci. 4, 177–192 (1970)
Watel, D., Weisser, M.A.: A note on the inapproximability of the Minimum Monotone Satisfying Assignment problem. Technical report, HAL (2016). https://hal.archives-ouvertes.fr/hal-01377704
Acknowledgment
We thank the three anonymous reviewers for their constructive comments that helped to improve the quality of this paper and of future work.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
A Proof of Theorem 12
A Proof of Theorem 12
Theorem 11. FS–MLWSP is W[1]-complete with respect to l and \(\pi \).
Proof
By Lemma 3 and Theorem 10, we have to prove that FS–MLWSP is W[1] with respect to l, \(\rho \), \(\pi \) and C. To do so, we describe a reduction to the partitioned clique problem. Let \(\mathcal {I} = ((\mathcal {P}, \mathcal {T}, \emptyset ), \omega , \mathcal {C}, l)\) be an instance of FS–MLWSP. We assume that \(\mathcal {I}\) contains a fake transition \(t_0\) such that \(t^-_0 = t^+_0 = \emptyset \) so that \(\mathcal {I}\) contains a feasible solution of size lower than l if and only if it contains a feasible solution of size exactly l. We now create an instance \(\mathcal {J} = (G, V_1, V_2, \dots , V_l, V_{l+1})\) of the partitioned clique problem such that \(\mathcal {J}\) contains a feasible solution if and only if \(\mathcal {I}\) does. An example is given in Fig. 7.
In the remaining of this proof, given a state M, the ordered set of states \(\{\alpha _1, \alpha _2, \dots , \alpha _k\}\) is called a decomposition of M if and only if \(\sum _{j = 1}^{k} \alpha _j = M\). The state \(\alpha _j\) may be empty.
Let t be a transition of \(\mathcal {I}\). For each \(i \in [\![2;l ]\!]\), we create at most \((i-1)^\rho \cdot (l-(i+1))^\pi \) nodes in \(V_i\), one per decomposition of \(t^-\) in \(i-1\) states and per decomposition of \(t^+\) in \(l-i+1\) states. If \(t^- = \emptyset \), we also create at most \(l^\pi \) nodes in \(V_1\), one per decomposition of \(t^+\) into l states. Finally, we add at most \(l^C\) nodes to \(V_{l+1}\), one for each decomposition of \(\mathcal {C}\) into l sets. For example, in Fig. 7, \(t_1^- = \emptyset \) and \(t_1^+ = x + 2y\). We add six nodes to \(V_1\) as there are six possible decompositions of \(t_i^+\) into 2 states : \(\{\emptyset , x + 2y\}, \{x, 2y\}, \{x+y, y\}\) and the symmetrical decompositions. Note that the number of nodes in G is FPT with respect to the parameters l, \(\rho \), \(\pi \) and C.
Let \(i < j \in [\![1; l ]\!]\), and \(v_i \in V_i\) and \(v_j \in V_j\) be two nodes of G. The node \(v_i\) is associated with a transition \(t_i\), a decomposition of \(t^-_i\) and a decomposition \(\{\alpha _{i+1}, \alpha _{i+2}, \dots , \alpha _{l+1}\}\) of \(t_i^+\). Similarly, the node \(v_j\) is associated with a transition \(t_j\), a decomposition \(\{\beta _{1}, \beta _{2}, \dots , \beta _{j-1}\}\) of \(t_j^-\) and a decomposition of \(t_j^+\). We add to G the edge \((v_i, v_j)\) if and only if \(\alpha _{j}\) covers \(\beta _{i}\).
Similarly, let \(i \in [\![1; l ]\!]\), and \(v_i \in V_i\) and \(v_{l+1} \in V_{l+1}\) be two nodes of G. The node \(v_i\) (respectively \(v_{l+1}\)) is associated with a transition \(t_i\), a decomposition of \(t^-_i\) and a decomposition \(\{\alpha _{i+1}, \alpha _{i+2}, \dots , \alpha _{l+1}\}\) of \(t_i^+\) (resp. a decomposition \(\{\beta _1, \beta _2, \dots , \beta _l\}\) of \(\mathcal {C}\)). We add to G the edge \((v_i, v_{l+1})\) if and only if \(\alpha _{l+1}\) covers \(\beta _i\).
Example of reduction from an instance \(\mathcal {I}\) of FS–MLWSP with \(l=2\) on the left to an instance \(\mathcal {J}\) of Partitioned Clique on the right. Due to lack of space, the fake transition is not part of \(\mathcal {I}\). Each node of the clique instance is either a transition t with decompositions of \(t^-\) (on the left) and \(t^+\) (on the right) or an decomposition of C into 2 states. The sequence \((t_1,t_2)\) is a feasible solution of \(\mathcal {I}\). The bold clique on the right is a feasible solution of \(\mathcal {J}\) associated with that sequence.
For example, in Fig. 7, the last node of \(V_1\) and the second node of \(V_2\) are respectively associated with the decomposition \(\{x+2y, \emptyset \}\) of \(t_1^+\) and the decomposition \(\{x+y\}\) of \(t_2^-\). The two nodes are linked as \(x+2y\) covers \(x+y\).
We now prove that \(\mathcal {J}\) contains a feasible solution if and only if \(\mathcal {I}\) has an enabled sequence at \(\emptyset \) with l transitions such that the resulting state covers \(\mathcal {C}\).
We first assume there is a clique of G containing a node \(v_i\) of \(V_i\) for every \(i \in [\![1;l+1 ]\!]\). Let \(T = (t_1, t_2, \dots , t_l)\) be the sequence such that \(t_i\) is the transition associated with \(v_i\) for \(i \le l\). Let \(\{\beta _{i,1}, \beta _{i,2}, \dots , \beta _{i,i-1}\}\) and \(\{\alpha _{i,i+1}, \alpha _{i,i+2}, \dots , \alpha _{i,l+1}\}\) be respectively the decompositions of \(t_i^-\) and \(t_i^+\) associated with \(v_i\) and let \(\{\mathcal {C}_{1}, \mathcal {C}_{2}, \dots , \mathcal {C}_{l}\}\) be the decomposition of \(\mathcal {C}\) associated with \(v_{l+1}\).
We now show that T is enabled at \(\emptyset \). As \(v_1 \in V_1\), then \(t_1^- = \emptyset \) and \(t_1\) is enabled at \(\emptyset \). Let \(i \in [\![1;l-1 ]\!]\). As \((v_1, v_2 \dots v_i, v_{i+1})\) is a clique of G, then, by construction, \(\alpha _{j,i+1}\) covers \(\beta _{j,i+1}\) for every \(j \in [\![1;i ]\!]\). For each place x of \(\mathcal {P}\):
Consequently, for every \(i \in [\![1;l ]\!]\), \((t_1, t_2, \dots , t_i)\) is enabled at \(\emptyset \). Finally, we can similarly show that \(\sum _{j = 1}^{l} (t_j^+(x) - t_j^-(x)) \ge \sum _{j = 1}^{l} \alpha _{j,l+1}(x)\) and, as \((v_i, v_{l+1})\) is an edge of G for \(i \le l+1\), then, by construction, \(\sum _{j = 1}^{l} \alpha _{j,l+1}(x) \ge \sum _{j = 1}^{l} \mathcal {C}_j(x) = \mathcal {C}(x)\) for every place x. Thus, when firing T, the resulting state covers \(\mathcal {C}\) and T is a feasible solution of \(\mathcal {I}\).
We now assume there exists such a feasible solution \(T = (t_1, t_2, \dots , t_l)\) of \(\mathcal {I}\) and prove \(\mathcal {J}\) contains a clique of size \(l+1\). Let \(\gamma _{i,j}\), for \(i < j \in [\![1;l ]\!]\) be the states describing the tokens produced by \(t_i\) and consumed by \(t_j\) ; let \(\gamma _{i,l+1}\) be the tokens produced by \(t_i\), not consumed by any next transition and used to cover \(\mathcal {C}\) ; and finally let \(\gamma _{i,l+1}'\) be the not consumed tokens of \(t_i^+\) that are not used to cover \(\mathcal {C}\). Thus, we have firstly \(t^+_i = \gamma _{i,l+1}' + \sum _{j=i+1}^{l+1} \gamma _{i,j}\), secondly \(t^-_j = \sum \limits _{i=1}^{j-1} \gamma _{i,j}\) and thirdly \(\sum \limits _{i=1}^{l} \gamma _{i,l+1} = \mathcal {C}\). For each set \(V_i\) with \(i \le l\), we select the node \(v_i\) associated with \(t_i\), the decomposition \(\{\gamma _{j,i}, j \le i-1\}\) of \(t_i^-\) and the decomposition \(\{\gamma _{i,i+1}, \dots , \gamma _{i,l}, \gamma _{i,l+1} + \gamma _{i,l+1}'\}\) of \(t_i^+\). We finally choose the node \(v_{l+1}\) of \(V_{i+1}\) associated with the decomposition \(\{\gamma _{j,l+1}, j \le l \}\) of \(\mathcal {C}\). If \(i < j \le l\), then \(\gamma _{i,j}\) is the j-th state of the decomposition of \(t_i^+\) and the i-th state of the decomposition of \(t_j^-\). By construction the edge \((v_i, v_j)\) belongs to G. Moreover, if \(i \le l\), the last state \(\gamma _{i,l+1} + \gamma _{i,l+1}'\) of the decomposition of \(t_i^+\) covers the i-th state \(\gamma _{i,l+1}\) of the decomposition of \(\mathcal {C}\). By construction the edge \((v_i, v_{l+1})\) belongs to G. Consequently, \(\{v_1, v_2, \dots , v_{l+1}\}\) is a clique of size of \(l+1\) in \(\mathcal {J}\).
As a result and by Theorem 10, FS–MLWSP is W[1]-Complete in l and \(\pi \). \(\Box \)
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this paper
Cite this paper
Watel, D., Weisser, MA., Barth, D. (2017). Parameterized Complexity and Approximability of Coverability Problems in Weighted Petri Nets. In: van der Aalst, W., Best, E. (eds) Application and Theory of Petri Nets and Concurrency. PETRI NETS 2017. Lecture Notes in Computer Science(), vol 10258. Springer, Cham. https://doi.org/10.1007/978-3-319-57861-3_19
Download citation
DOI: https://doi.org/10.1007/978-3-319-57861-3_19
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-57860-6
Online ISBN: 978-3-319-57861-3
eBook Packages: Computer ScienceComputer Science (R0)