Upe Blog: April 2015

Sunday, April 19, 2015

Launching simple echo experiment in Apache Airavata

Apache Airavata is a science gateway application that enables managing different scientific computational tasks among computational resources. Here I describe how to register a simple echo application in Airavata and launch it using Airavata API

1. Clone Airavata source https://github.com/apache/airavata/tree/0.14_release and build

2. Start Airavata server as given in https://cwiki.apache.org/confluence/display/AIRAVATA/XBAYA+Quick-Start+Tutorial

3. Run Sample class to register echo application
https://github.com/apache/airavata/blob/0.14_release/airavata-api/airavata-client-sdks/java-client-samples/src/main/java/org/apache/airavata/client/samples/RegisterSampleData.java

4. Run following command to create and run the experiment

Airavata.Client airavataClient = AiravataClientFactory.createAiravataClient("127.0.0.1", 8930);

String appId = "Echo_e82aa96b-66ea-4f31-97e7-1182a32e55d2";

List exInputs = new ArrayList();
InputDataObjectType input = new InputDataObjectType();
input.setName("Input_to_Echo");
input.setType(DataType.STRING);
input.setValue("Echoed_Output=Hello World");
exInputs.add(input);

List exOut = new ArrayList();
OutputDataObjectType output = new OutputDataObjectType();
output.setName("Echoed_Output");
output.setType(DataType.STRING);
output.setValue("");
exOut.add(output);

Experiment simpleExperiment = ExperimentModelUtil.createSimpleExperiment("default", "admin", "echoExperiment", "Echo Exp", appId, exInputs);
simpleExperiment.setExperimentOutputs(exOut);

Map computeResources = getClient().getAvailableAppInterfaceComputeResources(appId);
String id = computeResources.keySet().iterator().next();
String resourceName = computeResources.get(id);
System.out.println(computeResources.size());
System.out.println(id);
System.out.println(resourceName);
ComputationalResourceScheduling scheduling = ExperimentModelUtil.createComputationResourceScheduling(id, 1, 1, 1, "normal", 30, 0, 1, "sds128");
UserConfigurationData userConfigurationData = new UserConfigurationData();
userConfigurationData.setAiravataAutoSchedule(false);
userConfigurationData.setOverrideManualScheduledParams(false);
userConfigurationData.setComputationalResourceScheduling(scheduling);
simpleExperiment.setUserConfigurationData(userConfigurationData);

String exp = airavataClient.createExperiment(simpleExperiment);
airavataClient.launchExperiment(exp,"sample");

We (Dimuthu Upeksha, Chamila Wijayarathna, Maduranga Siriwardan, Lahiru Lasadun) started SinMin - Sinhala Corpus Project as final year undergraduates for our final year project under the supervision of Dr. Chinthana Wimalasuriya, Mr. N. H. N. D. de Silva and Prof Gihan Dias.

A rich language corpus enables a wide area of research topics for a language. Most of them include

1. Statistical analysis of the language usage pattens
2. Translators
3. Spell and Grammar Tools
4. Backend support to third party applications like OCR tools

Usually a corpus contains a collection of authentic texts of the language. However rather than storing them as raw text files, Sinmin further stores them in different databases with different schemas. This enables Sinmin to easily process language data in realtime. In addition to that SinMin Corpus contains a REST API that enable querying and finding data through third party applications.

SinMin web interface provides ability to illustrate and find patterns that occur in Sinhala Language over different time periods and different categories.