Search for: All records

There is a widespread feeling among potential employers that the traditional PhD model—largely unchanged since the years after World War II—no longer fits the world that our graduates are entering. The recently held 'National Workshop on the Formation of Industry-University Partnerships for Doctoral Training', has yielded a road map to address the current challenges with specific recommendations for academia, industry, government, and doctoral students. Based on this new knowledge, it is our opportunity—and responsibility—to guide the next transformation of doctoral training, ensuring that the PhD remains both rigorous and relevant in an era of rapid discovery and innovation. 

If the STEM PhD were a product sold in stores, customers would have long ago called for a redesign. The National Academies, Council of Graduate Schools, and Nature have all rightly advocated for reforms to PhD training. They've echoed calls from industry and society for researchers to be more responsive and more quickly generate innovative solutions to pressing problems. We encourage our fellow engineering educators to join with us in reimagining graduate education through P3 consortia or by driving similar innovation at their institutions. 

By querying approximate surrogate models of different fidelity as available information sources, Multi-Fidelity Bayesian Optimization (MFBO) aims at optimizing unknown functions that are costly or infeasible to evaluate. Existing MFBO methods often assume that approximate surrogates have consistently high or low fidelity across the input domain. However, approximate evaluations from the same surrogate can have different fidelity at different input regions due to data availability and model constraints, especially when considering machine learning surrogates. In this work, we investigate MFBO when multi-fidelity approximations have input-dependent fidelity. By explicitly capturing input dependency for multi-fidelity queries in a Gaussian Process (GP), our new input-dependent MFBO (iMFBO) with learnable noise models better captures the fidelity of each information source in an intuitive way. We further design a new acquisition function for iMFBO and prove that the queries selected by iMFBO have higher quality than those by naive MFBO methods, with a derived sub-linear regret bound. Experiments on both synthetic and real-world data demonstrate its superior empirical performance. 

We examine results from two transient modeling experiments that simulate the Last Interglacial period (LIG) using the state-of-the-art Community Earth System Model (CESM2), with a focus on climate and ocean changes relevant to the possible collapse of the Antarctic ice sheet. The experiments simulate the early millennia of the LIG warm period using orbital forcing, greenhouse gas concentrations and vegetation appropriate for 127ka; in the first case (127ka) no other changes are made; in the second case (127kaFW), we include a 0.2 Sv freshwater forcing in the North Atlantic. Both are compared with a pre-industrial control simulation (piControl). In the 127ka simulation, the global average temperature is only marginally warmer (0.004 degrees C) than in the piControl. When freshwater forcing is added (127kaFW), there is surface cooling in the NH and warming in the SH, consistent with the bipolar seesaw effect. Near the Antarctic ice sheet, the 127ka simulation generates notable ocean warming (up to 0.4 degrees C) at depths below 200 m compared to the piControl. In contrast, the addition of freshwater in the North Atlantic in the 127kaFW run results in a multi-century subsurface ocean cooling that rebounds slowly over multiple millennia near the Antarctic ice sheet. These results have implications for the thermal forcing (and thereby mass balance) of the Antarctic Ice Sheet. We explore the physical processes that lead to this result and discuss implications for climate forcing of Antarctic ice sheet mass loss during the LIG. 

We examine results from two transient modeling experiments that simulate the Last Interglacial period (LIG) using the state-of-the-art Community Earth System Model (CESM2), with a focus on climate and ocean changes relevant to the possible collapse of the Antarctic ice sheet. The experiments simulate the early millennia of the LIG warm period using orbital forcing, greenhouse gas concentrations and vegetation appropriate for 127ka; in the first case (127ka) no other changes are made; in the second case (127kaFW), we include a 0.2 Sv freshwater forcing in the North Atlantic. Both are compared with a pre-industrial control simulation (piControl). In the 127ka simulation, the global average temperature is only marginally warmer (0.004 degrees C) than in the piControl. When freshwater forcing is added (127kaFW), there is surface cooling in the NH and warming in the SH, consistent with the bipolar seesaw effect. Near the Antarctic ice sheet, the 127ka simulation generates notable ocean warming (up to 0.4 degrees C) at depths below 200 m compared to the piControl. In contrast, the addition of freshwater in the North Atlantic in the 127kaFW run results in a multi-century subsurface ocean cooling that rebounds slowly over multiple millennia near the Antarctic ice sheet. These results have implications for the thermal forcing (and thereby mass balance) of the Antarctic Ice Sheet. We explore the physical processes that lead to this result and discuss implications for climate forcing of Antarctic ice sheet mass loss during the LIG. 

Abstract. We examine results from two transient modeling experiments that simulate the Last Interglacial period (LIG) using the state-of-the-art Community Earth System Model (CESM2), with a focus on climate and ocean changes relevant to the possible collapse of the Antarctic ice sheet. The experiments simulate the early millennia of the LIG warm period using orbital forcing, greenhouse gas concentrations, and vegetation appropriate for 127 ka. In the first case (127ka), no other changes are made; in the second case (127kaFW), we include a 0.2 Sv freshwater forcing in the North Atlantic. Both are compared with a pre-industrial control simulation (piControl). In the 127ka simulation, the global average temperature is only marginally warmer (0.004 °C) than in the piControl. When freshwater forcing is added (127kaFW), there is surface cooling in the Northern Hemisphere (NH) and warming in the Southern Hemisphere (SH), consistent with the bipolar seesaw effect. Near the Antarctic ice sheet, the 127ka simulation generates notable ocean warming (up to 0.4 °C) at depths below 200 m compared to the piControl. In contrast, the addition of freshwater in the North Atlantic in the 127kaFW run results in a multi-century subsurface ocean cooling that rebounds slowly over multiple millennia near the Antarctic ice sheet. These results have implications for the thermal forcing (and thereby mass balance) of the Antarctic ice sheet. We explore the physical processes that lead to this result and discuss implications for climate forcing of Antarctic ice sheet mass loss during the LIG. 

Not Available.