Results and new simulations // Resultados y nuevas simulaciones
Message boards :
News :
Results and new simulations // Resultados y nuevas simulaciones
Message board moderation
| Author | Message |
|---|---|
 Jesús Carro Send message Joined: 18 Mar 15 Posts: 283 Credit: 2,748,608 RAC: 0     |
Dear volunteers, I am writing to update you on how the project is going and the results we are obtaining. First of all, the optimization of the new model we are developing has already finished (the first one we released). Unfortunately the results have not been as expected. The optimization has worked very well and many markers have been improved, but one of the model issues that we are trying to fix has reappeared after the optimization. We think we understand where the problem is coming from and why it seemed like we fixed it, but really didn't. If we are right, this will require rethinking the model and starting over. Before starting again, we are testing various optimization settings to try to optimize the model without getting to the point where the problem reappears. We're not sure if it will work (this is science), but before we have to start over, we want to test this option. On the other hand, some of you have already noticed that we are launching work in the Beta version. One of our pending projects was to use DENIS not only to improve the model we were developing, but also to try to improve other existing models. Let's start with this part of the project. For now, we will use the O'Hara and Rudy model, one of the most used in research. We have released the simulations in Beta version because we have never tested this model in DENIS before, but they will soon be released in HuVeMOp. Finally, you may have noticed that we have run out of space a couple of times lately: the increase in the number of simulations requires more speed to get the results out and post-process them. We are working on further automating these processes in order to continue growing. Thank you all very much for your support of the project and we will keep you informed of new developments. Sincerely, Jesus. ===================================== Estimados voluntarios: Os escribo para actualizaros cómo va el proyecto y los resultados que estamos obteniendo. En primer lugar, la optimización del nuevo modelo que estamos desarrollando ya ha terminado (la primera que lanzamos). Lamentablemente los resultados no han sido como esperábamos. La optimización ha funcionado muy bien y se han mejorado muchos marcadores, pero uno de los problemas del modelo que estabamos intentando solucionar, ha reaparecido tras la optimización. Creemos entender de dónde viene el problema y por qué parecía que lo habíamos solucionado, pero realmente no lo habíamos hecho. Si estamos en lo cierto, esto requerirá replantear el modelo y comenzar de nuevo. Antes de recomenzar de nuevo, estamos probando varias configuraciones de la optimización para intentar optimizar el modelo sin llegar al punto en el que reaparece el problema. No estamos seguros de si funcionará (esto es ciencia), pero antes de tener que volver a empezar, queremos probar esta opción. Por otro lado, algunos ya habéis notado que estamos lanzando trabajo en la versión Beta. Uno de nuestros proyectos pendientes, era utilizar DENIS no solo para mejorar el modelo que estabamos desarrollando, sino para intentar mejorar otros modelos ya existentes. Vamos a arrancar con esta parte del proyecto. De momento lo haremos con el modelo de O'Hara y Rudy, uno de los más utilizados en investigación. Hemos lanzado las simulaciones en la versión Beta porque nunca habíamos probado este modelo en DENIS, pero en breve pasarán a lanzarse en HuVeMOp. Finalmente, habréis notado que nos hemos quedado un par de veces sin espacio últimamente: el aumento en el número de simulaciones requiere mayor rapidez a la hora de sacar los resultados y post procesarlos. Estamos trabajando en automatizar más estos procesos para poder seguir creciendo. Muchas gracias a todos por vuestro apoyo al proyecto y os mantenemos informados de los nuevos avances. Atentamente, Jesús. Jesús Carro Universidad San Jorge @InSilicoHeart |
|
Send message Joined: 24 Sep 22 Posts: 9 Credit: 108,298 RAC: 0 
|
Great news! I assume there are so many parameters and relations between them (also undiscovered) that it would be miracle if this worked at first time as expected. Also numerical errors play role and are not trivial to handle. Problems with server is nothing compared to the potential future benefits of results. It's normal that for DDOS-like requests server may sometimes be unavailable - it's not the problem, sending of results can be retried later. |
|
Send message Joined: 2 Aug 22 Posts: 39 Credit: 1,017,028 RAC: 0
|
Other than the occasional server issue (growing pains) it sounds like your theories are progressing well. Are you monitoring your system on the weekend for troubles? I run my system 8 or so hours at night and 24hrs a day saturday and sunday. So if I get a server error I will know you saw it and will fix it. I hate having a backlog of tasks sitting on my system. Is there some place we can go to look up what you are referring to? Any way to provide a screen shot of what a "model" looks like? What is the O'Hara and Rudy model? |
|
Send message Joined: 24 Sep 22 Posts: 9 Credit: 108,298 RAC: 0
|
Information can be easily found on the internet, for example here: https://www.scaptest.com/pages/models.html and here https://www.semanticscholar.org/paper/Evolution-of-the-Seminal-O’Hara-Rudy-Model-to-More-Bartolucci-Paci/6c9788287fbf3bcba247e2b950affd6f1c9932d4. |
|
Jesús Carro Send message Joined: 18 Mar 15 Posts: 283 Credit: 2,748,608 RAC: 0 |
Other than the occasional server issue (growing pains) it sounds like your theories are progressing well. Yes, we monitoring on the weekend, but with less frequency than other days. I understood you want to upload your tasks as soon as possible, and for that reason we are working on prevent the issues. On the other hand, it is very complicated to explain how a model looks like. They are a lot of equations trying to represent how the different variables of a cell interact. The O'Hara and Rudy model is described in https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002061. At some point I want to start writing some more informative articles on the basis of our research so that I can explain more complex things later. Classes have finished and a quieter period is approaching... Let's see if I can start with this idea. Best, Jesús. Jesús Carro Universidad San Jorge @InSilicoHeart |
|
Send message Joined: 2 Aug 22 Posts: 39 Credit: 1,017,028 RAC: 0
|
Other than the occasional server issue (growing pains) it sounds like your theories are progressing well. I had a look, the "model" is more of a graph I guess. I was still thinking in terms of protein models like Rosetta. That paper makes my brain freeze! I have done a lot of medical events as a AV tech in my earlier years (15-20 years ago) but this is way over my head! As far as a informative artcles go, Looking forward to it! Good luck!, I hope you can simplify the complex science into something we can understand. I had no idea this was so complex! |
jay Send message Joined: 28 Jun 15 Posts: 8 Credit: 574,508 RAC: 0
|
Greetings, I remember the growing pains of the SETI@Home project. As an engineer, I was fascinated with the details of problems and their solutions. Their forums were excellent in sharing info and knowledge. I especially admire and appreciate the DENIS decision to shorten the required response time - to minimize WU that were not started by their clients. (This also allows other clients, possibly new clients, to get WU. I wonder if it would be appropriate to post a informational topic on how WU are distributed and how clients request and process the workflow effect processing times for the project. Thanks, Jay |
