Talk:Petri net

Add PIPE link to See also

Why I came to this article was not to understand Petri Nets. I wanted unbiased information on the PIPE open source program. There are several versions. It is not mentioned in 'see also' or any section of the article. I think that this would add something to the article. 184.144.115.253 (talk) 15:28, 26 March 2022 (UTC)

Citation needed on Introduction

The reference section of this article point to several mathematical paper on Petri nets. Petri nets have limitations and various form exist to try and address the limitation of the instantaneous/timed petri nets. However, the reference section clearly shows several mathematical papers on a formalism to Petri nets. This is to contrast with a UML sequence diagram which is only a pictorial to diagram the information or execution flow, but may not be able to 'prove' anything more advanced that what is depicted in the diagram. This contrast to a Petri net where properties can be derived from the diagram. Is the reference section a suitable citation? The tagger who left no text here seems to want an article on the articles in the reference section. 184.144.115.253 (talk) 15:18, 26 March 2022 (UTC)

In fact the whole article is rather mathematically based. There is little on the pragmatic application of a Petri net. This is completely counter to what is on the UML Sequence diagram page. The next topic is that the article is too pendantic. As well, Petri Net is assigned to the 'mathematics category' and a sequence diagram is not. It seems clearly more mathematical/formal and I am left to wonder why it was tagged? Maybe User:Ahmadadam96 could explain? 184.144.115.253 (talk) 15:44, 26 March 2022 (UTC)

Pedantic

This page seems overly pedantic. The opening, for instance, is filled with terms needing additional clarification:

discrete distributed systems generalize automata theory

Seems like a simpler, clearer statement is needed--and I'm not qualified! Can someone make this accessible to a layman like me?

No problemo! I'll get right on it. Vonkje 03:04, 8 Jun 2005 (UTC)

I think the graphic relating to state machines would be helpful to move to the introduction. This frames why Petri nets are more complex. It is difficult to make a complex topic simple. A simple example might help. Customers at a bank/DMV/waterslide and the number of employees needed to keep the customer queue short? 184.144.115.253 (talk) 15:33, 26 March 2022 (UTC)

"Also known as"

Vonkje, I am not happy with the "also known as" clauses in your formal definition. E.g., a place is not a "file" or "other container", it is a place, an abstract notion. Similarly, I think your introduction uses the word transaction where it really doesn't belong. (One of the problems with Petri nets is that they cannot easily express transactions.)

yer right! ... I'll 'up 'n fix it. Vonkje 20:37, 9 Jun 2005 (UTC)

The formal part is good, although I have never seen a separation between arcs and arc weights before, nor have I seen the use of place capacities; I would suggest to omit the latter. Can you give a reference for this definition?

--Rp 17:17, 9 Jun 2005 (UTC)

I hadn't seen the separation either, until I attended a tutorial by Jörg Desel at the 4th ACPN in 2003. ... makes sense when you think about it more. The paper is called "What is a Petri Net" by Desel and Juhás.

By the way: Warm greetings from Leiden to Eindhoven (though I'm here in Florida for the summer). Vonkje 20:37, 9 Jun 2005 (UTC)

Just a detail regarding "Restrictions", 1. state machine

It is claimed within the mentioned section, that in a state machine type petri net there cannot be concurrency. However, the only constraint given for the state machine type is $\forall t\in T:|t^{\bullet }|=|{}^{\bullet }t|=1$

I believe this is not enough, in addition the whole net must be one connected component. Consider a net $(S,T,W)$ where $S=\{p_{1},p_{2}\},T=\{t_{1},t_{2}\},W=F\cup F^{-1}$ and $F=\{(p_{1},t_{1}),(p_{2},t_{2})\}$. I.e. 2 places with self loops. — Preceding unsigned comment added by RasTaIARI (talk • contribs) 12:28, 10 August 2012 (UTC)

You are correct; alternatively, the initial marking may be required to have exactly one token. I don't know what is common. Rp (talk) 14:23, 23 September 2013 (UTC)

Relationship with state machines

Dear all,

in Section Restrictions it is stated the following:

1. In a state machine (SM), every transition has one incoming arc, and one outgoing arc. This means, that there can not be concurrency, but there can be conflict (i.e. Where should the token from the place go? To one transition or the other?). Mathematically: $\forall t\in T:|t^{\bullet }|=|{}^{\bullet }t|=1$

Unless I missed something this is not true. The concurrency in a Petri net is not due to the in/out degree of its transitions but to the fact that several places can contain tokens in a given marking. To get (something which can be seen as) a state machine one thus needs to consider a safe Petri net such that every transition has an in/out degree of one.

--Lowije (talk) 12:26, 23 September 2013 (UTC)

You are correct; alternatively, the initial marking may be required to have exactly one token. I don't know what is more common. Rp (talk) 14:42, 23 September 2013 (UTC)

I don't know what is more common either. In fact I am not sure that this is a classical relationship between Petri nets and FSA, I can't remember having read it anywhere else than on this wikipedia page (and on the corresponding french page which has the same mistake in it). Maybe requiring the initial marking to have exactly one token is better, as it is a simple structural property. Lowije (talk) 12:26, 23 September 2013 (UTC)

I agree; made the change. The observation that a finite state machine is a Petri net in which all branching happens on places is definitely commonplace. Rp (talk) 10:15, 24 September 2013 (UTC)

By the way, how would you propose to fix the way Petri nets are defined? My proposal would be to adopt the definitions in Reisig, 2013.

Notation is mixed

In the definitions of L and R, the notation L(N) and R(N) is used (with an obviously missing parameter). More downward on the page, L(N,M_0) and R(N,M_0) are used. — Preceding unsigned comment added by 2001:980:93A5:1:74EB:4D18:FB5D:EBB2 (talk) 18:18, 17 October 2016 (UTC)