| [stem-dev] Notes from the STEM Community Call on May 11, 2017 |
Notes from the STEM Community Call on May 11, 2017 On the call: James Kaufman, Stefan Edlund, Kun Hu, Taras Gunther, Armin, and Judy Douglas Next Call: Thursday, June 9th, 9 am Pacific Time 1. STEM 3.0.0 Milestone 5: Release to follow solution of problem with launcher on Mac This effort to make the build compatible with Eclipse is a major undertaking and awaits resources (summer intern? post doc?). Until completed, use the workaround available on the wiki at https://www.eclipse.org/forums/index.php/mv/msg/1084123/1753241/#msg_1753241 2. The Stochastic Solver Taras asked about the differences between the stochastic solver in STEM and the stochastic Gillespie algorithm to solve ODEs. Jamie noted that there is only one stochastic solver available in STEM. Emily discussed and resolved setting the seed on the stochastic solver and using the nonlinearity coefficient on email with Stefan and Jamie. *TO DO: Stefan to post relevant emails on stem-dev. *TO DO: Stefan will hide random seed and the stochastic true/false option that are in the e-core of the disease model. Neither works in the current version, but neither can be removed without breaking the build; hence they will be hidden. 3. Model Development The problems Emily noted in April with predicates have been resolved. Emails have been posted to stem-dev. *TO DO: Emily will close bug. Issues Taras experienced with birth and death rates during model development have been posted as bugs; for details see item 4. 4. Bugs/Feature Requests 515648: Mixing edges when migrating to another polygon. Fixing this bug would be a major development effort and is essentially a feature request. *TO DO: Taras will consult with Matthias; if this is important for BfR, Jamie and Stefan will join in a call with Matthias. IBM does not now have time to resolve. 513636: Create option to log internal parameters over time. Another feature request. 515654: Calculating disease model with model generator; problem with multiple aging populations. Also a feature request. 5. New Contributions Available to STEM Users Taras has posted a links to number of resources on the wiki and/or stem-dev (lecture series at the Koch Institute in Berlin; course material from BfR’s course on STEM; Taras’s own presentation on modeling). *TO DO: Taras will post on the wiki if not now there, also on the STEM Newsgroup. 6. Items from Participants EMILY: Wants to make the process of being infected stochastic while keeping treatment definite. Per Jamie, this is a feature request. *TO DO: Emily will add a bug. However, every transition has some noise added. Thus there are workarounds to try. *WORKAROUNDS: (1) Hack it using math.rand from java math to add noise to the infection process and add new hidden compartment in susceptible line code. (2) Try with finite difference solver and jiggle the “f” value.. (3) Go to the model creator on wiki and use the _expression_ editor to create a transition, “f”=random (0-1). *TO DO: Emily will test, write up what works for others. TARAS: Wants to respond to paper using deterministic model work he disagrees with. *TO DO: Jamie will send Taras the line of code where noise is computed. In response to query from Emily: On wiki, when going from Triggering Interventions to Importing Vaccination Data, click no on Vaccination to go to Isolation. |