Difference between revisions of "Explanation of Malariacontrol.net tasks"

From BOINC Wiki
Jump to: navigation, search
(Saving FAQ 304)
 
(adding table)
 
(One intermediate revision by one other user not shown)
Line 1: Line 1:
 
The following was written by Nicolas Maire.  
 
The following was written by Nicolas Maire.  
 
-------------------------------------------------
 
-------------------------------------------------
 +
  
 
Ok, I'll try and explain how you can tell from the workunit names roughly what the corresponding simulation is about. This list covers the current simulation runs and a few more we were running since the beginning of the year.
 
Ok, I'll try and explain how you can tell from the workunit names roughly what the corresponding simulation is about. This list covers the current simulation runs and a few more we were running since the beginning of the year.
 
First a short reminder of what the different components of the workunit names mean for the malariacontrol application (workunits that start with wu_), for example:
 
First a short reminder of what the different components of the workunit names mean for the malariacontrol application (workunits that start with wu_), for example:
 +
  
 
wu_133_234_108426_0_1206358864
 
wu_133_234_108426_0_1206358864
 +
  
 
The first number (here 133) is the run id. We group simulations into runs, and some of the recent run ids are listed and explained below.  
 
The first number (here 133) is the run id. We group simulations into runs, and some of the recent run ids are listed and explained below.  
 +
 
The second number (234) is the scenario id, where scenario stands for a set of input data to the simulation. In the process of estimating model parameters, we reuse the same set of scenarios over and over again, and see how well a given set of parameters fits the observations (the real world data from malariological field studies) that the particular scenario describes. In runs that aim to predict the impact of control interventions, the different scenarios correspond to assumptions about the details of the intervention.
 
The second number (234) is the scenario id, where scenario stands for a set of input data to the simulation. In the process of estimating model parameters, we reuse the same set of scenarios over and over again, and see how well a given set of parameters fits the observations (the real world data from malariological field studies) that the particular scenario describes. In runs that aim to predict the impact of control interventions, the different scenarios correspond to assumptions about the details of the intervention.
 +
 
The third number (108426) is the id of the parameter set we use. These parameters define how the mathematical functions that describe the simulated processes are shaped. Have a look at the published papers if you are interested in the details.
 
The third number (108426) is the id of the parameter set we use. These parameters define how the mathematical functions that describe the simulated processes are shaped. Have a look at the published papers if you are interested in the details.
 +
 
The forth number (0) is a seed for the random number generator, and the last number is a unix time stamp, the time the wu was created.
 
The forth number (0) is a seed for the random number generator, and the last number is a unix time stamp, the time the wu was created.
 +
  
 
And here now a few more details on what the current run numbers mean:
 
And here now a few more details on what the current run numbers mean:
Line 21: Line 28:
  
 
Finally, there is the other science application, optimizer, which has workunits named opt_* .
 
Finally, there is the other science application, optimizer, which has workunits named opt_* .
 +
 +
  
 
Michael writes:
 
Michael writes:
  
 
opt: optimizer application
 
opt: optimizer application
 +
 
36: run number 36 (each run represents a different model being fitted to the dataset.Currently 48 different models are being tested. For details see post "a third science application...")
 
36: run number 36 (each run represents a different model being fitted to the dataset.Currently 48 different models are being tested. For details see post "a third science application...")
 +
 
3700: the 3700th workunit belonging to this run
 
3700: the 3700th workunit belonging to this run
 +
 
5: number of config parameters, same for all runs.
 
5: number of config parameters, same for all runs.
 +
 
790480104 : random number to prevent two workunits from having the same name
 
790480104 : random number to prevent two workunits from having the same name
  
 
More information as to what a workunit is doing is in the input file or, more conveniently for users, in the standard error of a result after it was completed. Just at the end of the stderr_out,  after "file content":
 
More information as to what a workunit is doing is in the input file or, more conveniently for users, in the standard error of a result after it was completed. Just at the end of the stderr_out,  after "file content":
 +
 
The first number, ideally around 9000 for the moment, but up to 1E65, is the "lossfunction", a measure of how closely our models predict the field data. We are searching for the parameter values that have the LOWEST value of the lossfunction. After the lossfunction, all the parameters are listed.
 
The first number, ideally around 9000 for the moment, but up to 1E65, is the "lossfunction", a measure of how closely our models predict the field data. We are searching for the parameter values that have the LOWEST value of the lossfunction. After the lossfunction, all the parameters are listed.
  
 
Nick
 
Nick
 +
 +
 +
{| class="wikitable" style="text-align: center;"
 +
! Original writer
 +
! Original FAQ
 +
! Date
 +
|-
 
  | Jorden
 
  | Jorden
  | 37
+
  | [http://boincfaq.mundayweb.com/index.php?language=1&view=304 304]
| 1
+
  | 26-03-2008
| 22:12:12
+
  |-
| 2008-03-26
+
|}
| 45325
 
| 1
 
| 304
 
  | 2008-03-26 22:12:12
 
  | 2008-03-26 22:12:12
 
|-
 

Latest revision as of 15:54, 29 August 2016

The following was written by Nicolas Maire.



Ok, I'll try and explain how you can tell from the workunit names roughly what the corresponding simulation is about. This list covers the current simulation runs and a few more we were running since the beginning of the year. First a short reminder of what the different components of the workunit names mean for the malariacontrol application (workunits that start with wu_), for example:


wu_133_234_108426_0_1206358864


The first number (here 133) is the run id. We group simulations into runs, and some of the recent run ids are listed and explained below.

The second number (234) is the scenario id, where scenario stands for a set of input data to the simulation. In the process of estimating model parameters, we reuse the same set of scenarios over and over again, and see how well a given set of parameters fits the observations (the real world data from malariological field studies) that the particular scenario describes. In runs that aim to predict the impact of control interventions, the different scenarios correspond to assumptions about the details of the intervention.

The third number (108426) is the id of the parameter set we use. These parameters define how the mathematical functions that describe the simulated processes are shaped. Have a look at the published papers if you are interested in the details.

The forth number (0) is a seed for the random number generator, and the last number is a unix time stamp, the time the wu was created.


And here now a few more details on what the current run numbers mean:

The workunits with run id >=63 and id<=68 are part of a heterogeneity study. By heterogeneity in this case we mean heterogeneity in transmission (the number of infectious bites) and the human response to malaria infection. These runs try to determine how well different formulation of such models of heterogeneous transmission can explain field observations. This article contains some more information on the rational for such models.

Runs 71-74, 105 to 110, and 140-149 were all simulations that predicted the impact of different vaccination schemes under various conditions. Runs 136-137 and 150-153 predicted the impact of a control intervention called Intermittent Preventive Treatment in infants. These runs are over, and most of the data has been analyzed.

Runs 111-135 are all part of an attempt to learn something about the decay of (human-)acquired immunity to the malaria parasites. Just like runs 63-68, these runs are still ongoing, and will take some more time.

Finally, there is the other science application, optimizer, which has workunits named opt_* .


Michael writes:

opt: optimizer application

36: run number 36 (each run represents a different model being fitted to the dataset.Currently 48 different models are being tested. For details see post "a third science application...")

3700: the 3700th workunit belonging to this run

5: number of config parameters, same for all runs.

790480104 : random number to prevent two workunits from having the same name

More information as to what a workunit is doing is in the input file or, more conveniently for users, in the standard error of a result after it was completed. Just at the end of the stderr_out, after "file content":

The first number, ideally around 9000 for the moment, but up to 1E65, is the "lossfunction", a measure of how closely our models predict the field data. We are searching for the parameter values that have the LOWEST value of the lossfunction. After the lossfunction, all the parameters are listed.

Nick


Original writer Original FAQ Date
Jorden 304 26-03-2008