Kappa-Dev
Detect (and flag) if a rule has ambiguous molecularity
When specifying a rule, one can write a LHS with two components that have no connection stated in that LHS. Depending on the model, those components might be already connected (e.g. through a third party). In such a case, the physical interpretation of the model would require the usage of a different rate constant, as the diffusion volume of connected components usually is much smaller than the one of disconnected ones.
It would be useful if the IDE could flag cases where the LHS might have such ambiguous molecularity (e.g. by checking the contact map). Full reachability analysis of the species space seems too costly, but just verifying through the contact map would suggest rules to re-check. Since this is not a sensa stricta error, these rules would just be flagged for revision (i.e. instead of 'warning', maybe just 'hint' ?).
Before being an UI issue, it is a static analysis one. I don't know if it is in the scope of KaSa but it is definitely a potential question for it. I leave to @feret or @lykimquyen the choice to keep or not this issue open...
Weakly relational domains (based on Shortest-path closure) offer a nice trade-off between the contact map approach and the complete exploration of the species space (the complexity of atomic instructions in the abstract domain is n^3, with n a number of (sub)-views, the notion of (sub)) being parametric. I have an internship offer on this topic since a couple of years. If I find nobody this year as well, we may do it with Ferdinanda and @lykimquyen next spring.
+1
On 4 November 2016 at 01:29, Jérôme Feret [email protected] wrote:
Weakly relational domains (based on Shortest-path closure) offer a nice trade-off between the contact map approach and the complete exploration of the species space (the complexity of atomic instructions in the abstract domain is n^3, with n a number of (sub)-views, the notion of (sub)) being parametric. I have an internship offer on this topic since a couple of years. If I find nobody this year as well, we may do it with Ferdinanda and @lykimquyen https://github.com/lykimquyen next spring.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Kappa-Dev/KaSim/issues/235#issuecomment-258370472, or mute the thread https://github.com/notifications/unsubscribe-auth/AAcCLo01plbMpf0DE8PPIT08LbpzLYIoks5q6uz2gaJpZM4Iq7PB .
