Alireza Doostan

Life, in all its many forms, is full of uncertainty. Materials often come with manufacturing imperfections. Gravity perturbations alter the orbits of unwitting spacecraft and satellites.

In complex systems, uncertainty comes from a variety of sources, making accurate simulations difficult to produce and computationally taxing. Dr. Alireza Doostan is seeking to develop less-costly means of quantifying uncertainty in multi-physics systems, improving the accuracy and utility of real-world models.

Uncertainty quantification (UQ) is an emerging field, transcending both academia and industry. As Doostan explains:

“Historically, people ran codes on HPC platforms that took an order of months to run. The outcomes of these simulations soon started to be questioned – how credible were these models, really? How predictive were they? What sources of uncertainty did they ignore?” 

Even once efforts increased to incorporate uncertainty into models, the computing power and time required to run these simulations was exorbitant:

“People can do uncertainty quantification for single-physics problems, but when you look at multi-component, multi-scale problems - the types of problems that you actually encounter in mechanical, aerospace and civil engineering - there are no scalable, efficient ways of doing UQ. The cost of doing UQ becomes prohibitive when you have many sources of uncertainty.”

Doostan’s goal is to develop algorithms and methodologies that are “more scalable,” decreasing UQ’s computational demands.

Beginning his UQ research at the Center for Turbulence Research at Stanford University as a Postdoc Fellow, Doostan joined the CU faculty in 2009 as a member of the Center for Aerospace Structures (CAS). Doostan explains CAS’s appeal:

“CAS maintains lots of research in computational mechanics; it is known for its pioneering work in multi-physics problems. There was great synergy between the existing faculty in the center and myself, as far as my research supporting their activities.”

Through his research, Doostan cultivates an extensive network of collaborations. Within CU, Doostan frequently works with professors from the Applied Mathematics Department on algorithm development and with Principal Investigators from both CCAR and CAS. Nationally, Doostan collaborates closely with academics from Stanford University through the Predictive Science Academic Alliance Program.

In 2015, Doostan won a NSF Early Career Award of $500,000 for his project: “Fast Surrogate Modeling for Design under Uncertainty of Complex Engineering Systems." These funds will not only be used to support Doostan’s research, but also to establish an “outreach and education effort to attract the next generation of engineers.”

Doostan is partnering with the GoldShirt Program through CU’s BOLD Center to develop short-course modules that will be used to “recruit [high school] students into engineering programs at CU-Boulder and increase retention of students from minority and female communities.”

Already, Doostan is an active contributor to GoldShirt’s Summer Bridge Program, delivering a Research 101 lecture that “lets students know about research in academia and how to be successful researchers.” At the end of this year’s Summer Bridge Program, Doostan will select a student to mentor throughout the upcoming academic year.

Doostan’s notability extends beyond his research into his teaching ability. In 2015, he was awarded the Outstanding Junior Faculty Award of the Aerospace Engineering Sciences Department and the AIAA-Rocky Mountain Educator of the Year (College) award.

Doostan summarizes his teaching approach:

“The key [to effective teaching] is to be flexible in the approach you take, and to see what is the best approach to getting students interested. The best approach to getting students engaged may not always be same as your ‘ideal’ approach (what works best for me may not what works best for my 140 students). However, if you pay attention to the methods that motivate students, such as teaching through real-world engineering applications, everything converges nicely.”

- Written By: Ari Sandberg, Intern