Application Development and Deployment With MATLAB

Transcription

1 Application Development and Deployment With Jean-Philippe Villaréal Application Engineer Applications Engineering Group MathWorks Benelux June 11, The MathWorks, Inc. 1

2 Typical Industry Challenges How do I build robust applications? How do I collaborate in a team? What is the most efficient way to share my algorithms? How do I get sufficient computing performance? 2

3 From Data to Decisions and Design Observation Organization Understanding Decisions & Design Physical Sensors Data Information Knowledge Action 40 MPG 20 Acceleration NN measured Displacement Weight active power per-unit Horsepower time secs MPG Acceleration Displacement Weight Horsepow er 3

4 Decisions & Design Build Robust Applications Collaborate in a Team Generate Custom reports Deploy and Scale your Applications 4

5 Building Robust Applications Which programming technique? Scripting Procedural Object-oriented Easily create user interfaces Graphical environment Automatic code generation Build Unit Tests Built-in unit test framework 5

6 Developing in a Collaborative Environment Version Control Integration GIT, SVN Simulink Projects Organize, compare and maintain your models 6

7 Automatic Report Generation Create pdf, word documents, LateX, html pages in one click! 7

8 Create APPs and Toolboxes to Easily Share Your Projects Package your code and distribute to users 8

9 Deploy and Scale up your application Integration into Existing Systems & Product Deployment Excel 2015 The MathWorks, Inc. 9

10 Share Programs Outside of Excel add-ins Desktop Royalty-free Encryption to protect intellectual property Production Server(s) Web Server(s) Web & Enterprise 10

11 Application Deployment from Platform independent Royalty Free Scalable Flexible Code Generation Coder HDL Coder C/C++ HDL.exe.NET Java Excel Application Packaging Compiler Compiler SDK 11

12 Benefits of Deploying Code Easily adopt algorithm improvements throughout lifecycle Domain experts maintain ownership of ideas, algorithms, and applications Flexibility to integrate with different programming languages Implement a common algorithm on different platforms Avoid time consuming and error prone re-coding 12

13 The Range of Application Platforms 13

14 Which Product will Fit Your Needs? Compiler Compiler SDK Standalone Application Excel Add-in Hadoop C/C ++ Java.NET Production Server Compiler for sharing programs without integration programming Compiler SDK provides implementation and platform flexibility for software developers Production Server provides the most efficient development path for secure and scalable web and enterprise applications 14

15 Sharing Applications with Compiler Standalone desktop applications Compiler Add-ins for Excel Run code against Hadoop Standalone Application Excel Add-in Hadoop Create professional software with customizable installers, icons, and splash screens without integration programming 15

16 A 3-step Workflow Application Author Toolboxes 1 2 Compiler End User Standalone Application Excel Add-in Hadoop 3 Runtime 16

17 Integrating Applications with Compiler SDK Integrate with C/C++,.NET, Java Compiler SDK Develop applications for Production Server C/C ++ Java.NET Production Server A flexible toolkit for software developers to create professional robust components without recoding! 17

18 Integrating -based Components Application Author Toolboxes Application author and software developer might be same person 1 Software Developer 2 Compiler SDK C/C ++ Java.NET Production Server 3 4 Runtime 18

19 R2015a: Functionality Transition R2014b and earlier R2015a and later Builder NE Compiler SDK.NET assembly.net assembly Builder JA.jar files.jar files Builder EX Excel add-ins for Production Server Excel add-ins for desktop Tools for Production Server - Excel add-ins for servers - Deployable archives - Client libraries (new in 15a) - Debugging tools (new in 15a) C/C++ shared libraries Compiler Deployable archive for Production Server C/C++ shared libraries Hadoop Standalone applications Excel add-ins for desktop Hadoop Standalone applications Compiler 19

20 Scale up with Production Server Directly deploy programs into production Centrally manage multiple production applications Automatically deploy updates without server restarts Reduce risks Production Server(s) Scalable & reliable Service large numbers of concurrent requests Add capacity or redundancy with additional servers Use with web, database & application servers Lightweight client library isolates processing Access programs using native data types Web Server(s) Database Integration Take the algorithm to the data Request analytics directly from database servers Java.Net HTML XML Java Script 20

21 Production Deployment Workflow Development Developer Algorithm Compiler SDK Enterprise Application Developer Web Application CTF Production Server Client Library Function Call Production. Web Application Client Library Production Server 21

22 Application Deployment from Share programs with people who do not have Royalty-free distribution Encryption to protect your IP Create both standalone applications and components for integration Deploy to desktop, web, and enterprise applications 22

23 Scaling up Computational Intensity only Desktop Computer Local + PCT Desktop Computer Local Distributed Computing Server Computer Cluster Cluster Simulink, Blocksets, and Other Toolboxes Simulink, Blocksets, and Other Toolboxes Profile (Local) Scheduler Profile (Cluster) - Prototype code - Run a single process - Refine code to leverage Parallel Computing - Leverage you hardware - Scale Parallel Algorithm to high computational resource 23

24 Deployed Applications with Hadoop Using Compiler Datastore HDFS runtime Node Data Map Reduce Node Data Map Reduce Node Data Map Reduce MapReduce Code 24

25 Explore and Analyze Data on Hadoop Using Distributed Computing Server Datastore HDFS Distributed Computing Server Node Data Map Reduce Node Data Map Reduce MapReduce Code Node Data Map Reduce 25

26 Practical use case: Load Demand Forecasting Database Access Neural Network Parallel Computing on HPC Sharing on the web 26

27 Success Stories 2015 The MathWorks, Inc. 27

28 UniCredit Bank Austria Develops and Rapidly Deploys a Consistent, Enterprise-Wide Market Data Engine Challenge Improve risk management operations throughout a multinational financial institution Zero-coupon yield curve plot in UniCredit Bank Austria s UMD environment. Solution Use, Compiler, and Compiler SDK to build and rapidly deploy a consistent enterprise-wide data warehouse into J2EE Web Architecture Results Development time reduced by 50% Risk management improved across the bank Operational, audit, and maintenance costs reduced With, we can focus on business logic instead of implementation details. We can deploy an algorithm in a Java environment the same day, without any additional coding. This approach enabled us to cut our development time in half, if not more weeks, instead of months." Peter W. Schweighofer UniCredit Bank Austria Link to user story 28

29 Halliburton Makes Oil Exploration Safer Using and Neural Network Toolbox Challenge To improve the ability to detect detonation of explosives used to perforate the well bore Solution Use MathWorks products to develop an adaptive, predictive neural network filter that cleanses the detonation signal of contaminating noise from onsite machinery Results Authentic simulation on the desktop An accurate, production-standard algorithm Dramatic time savings Using and Compiler, I can develop an application at least 100 times faster than I could with Visual Basic or C. The time we saved on the very first application that we wrote in more than paid for the software. Roger Schultz Halliburton Energy Services Link to user story 29

30 Take-aways is a very efficient algorithm development environment Built-in libraries Utilities for sharing and collaborating is highly scalable Parallel computing capabilites Integration with clusters for distributed computing bridges the business integration gap Centralize your application and access it from anywhere 30

31 Thank You for Attending! 2015 The MathWorks, Inc. 31