Making advanced simulation tools more accessible and impactful
Improving Usability and Broadening Adoption of Simulation with Microworkflows
- related publication
motivation
-
advanced modeling and simulation of products leads to
- faster time to market
- cost saving on prototypes
-
bu there are barriers to adoption
- gen-purpose sim codes, software licenses, training,
- expert sim-workflow consultation for complexity
goals
-
improve usability
- show only whats needed for specific use cases
- hide the machine-readable complexity
-
broaden adoption of sim tech
- computational science engineers handle complexity
- stakeholders can use results to perform simulations and run analysis standalone
simulation adoption is cumbersome to use, steep learning curves, and analyzing test results is hard to comprehend
Workflows components
-
Define Experiments
- model verification, parameter estimation, optimizations, and inference
-
Depict Domain
- space of interest, discrete geometry, CAD models
-
Describe Simulation
- collection of files for roadmap for simulation
-
Discretize Domain
- discretize continuous surface or volume …
- Execute Simulation
- Analyze and Visualize Results
How do we?
- reduce the complexity involved in carrying out a sim workflow?
- make it easier to learn how to configure componentss of a workflow
- andd what about communicating with stakeholders?
Microworkflows
- bundle workflow components inttot small complete applications
- expose modeling and simulation capabilities through a limited set of predefined use cases
- define use cases with templates and present limited set of configuration options
Defining a microworkflow
- Computational Science Engineer(CSE) w/ knowledge of full simulation workflow formulates a use casse with sstakeholders
- Stakeholder and CSE id parameteers toi explore for sim exp
- CSE maps these parameters ….
The web is a pseudo application delivery mechanism for those under IT administration.
the workflow is well-defined, basically a application core-driven workflow,
the micro part is the laser-focus goal that we are doing. you pick a template, define parameters, and run simulation, then partition results you want.
- the template structure is creating the simulation itself
the orchestration of core app functionality in the microworkflow simpler to debug because it is hiding behind integrated cross-platform libraries, right?
what are the disadvantages relying on efficient application delivery mechanism through microworkflows?
- maybe resource-allocation
Also what environments will work best in for microworkflows?
- computational environments
- local-first and serverless
- remote connection to your browser running on an HPC GUI-less