Dependency list removal on dependent variables

[Clément Michaud]

Hello Community,

Glad to join it. I found a first paper on NAL 2 weeks ago and I find the topic very inspiring. I read the book and went through most of the forum and papers and I saw in multiple threads that the dependency list of dependent variables has been removed in the implementation of NARS but I did not quite find why the dependency list is not needed and can be removed. Maybe I missed something.

Can anyone help me understand why it can be done or point me to the paper explaining it if there is any.

Thanks,

Clément

[Clément Michaud]

To complement my question, I mean I understand that the list can be implicitly represented if it is empty, but what if it is not? Say #x(#y).

[Pei Wang]

Hi Clément,

The following is a note in the 2nd edition of the NAL book, which I'm working on:

This definition is a simplification of the definition in the previous edition of this book, where a dependent variable is explicitly specified as a function of some independent variables, and the prefixes are different from the ones introduced in this definition. Though conceptually a dependent variable can be a function of other variables, the dependency is already implicitly represented by the \emph{scope} (to be defined later) of the variables involved, that is, if the scope of an independent variable includes the scope of a dependent variable, the latter depends on the former. Since the processing of variables starts from larger scopes, in the known usages of NAL there has not been a case that demands to explicitly associating a dependency list to a dependent variable.

Regards,

Pei

--
You received this message because you are subscribed to the Google Groups "open-nars" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-nars+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-nars/1b7f075c-f7d3-48ed-bb30-cf3675810017n%40googlegroups.com.

[Clément Michaud]

Hi Pei,

Glad to hear that you are working on a v2. I loved the v1 so much that I read it in almost one go ah ah. I'd be happy to help with doing an external review on v2 anytime if needed.

Thanks for your answer, I think it's clearer now. Let me try to summarize to see if I got it right:

In the most general case, the dependency list must be specified otherwise the semantic of the dependent variable might be ambiguous, but in the cases handled so far in NARS, the dependency list has been implicit because there has not been any issue with ambiguity. This is due to the fact that the scopes of the variables prevented any kind of ambiguity. Am I right?

Regarding your remark on the prefixes, I indeed saw that when I went through the wiki on github. I'm sorry I still had the one in the book in mind when writting.

Thanks,

Clément.

[Pei Wang]

Hi Clément,

Your summary is accurate.

I should be able to finish the first draft of the 2nd edition of the NAL book soon (in 2023?) and do welcome comments and reviews. I'll probably post an announcement here when it is ready.

Regards,

Pei

To view this discussion on the web visit https://groups.google.com/d/msgid/open-nars/4e86f919-f3f3-46a5-9295-44dd21d35a37n%40googlegroups.com.

[Clément Michaud]

Hi Pei,

Thanks for the confirmation.

I'd be happy to review and comment. I'll stay tuned. Thanks.

Clément