Category Archives: housekeeping

Cross-platform ride-hail integration published

After an extended review process that lasted nearly two years, this paper finally came out in Transportation Research Part B.  Thanks Ruijie, the first author, for his extraordinary patience and commitment to scholarship, which have become a rare commodity in today’s hyper-competitive research environment that prioritizes productivity over everything else.

A pre-print of the paper was posted in September 2022 on SSRN, under a different title and with slightly different contents.  You may check it here if you don’t have Elsevier subscription.

Congratulations, Dr. Li

Aabout a week ago (on 7/19/2024), Jiayang Li defended his PhD thesis, entitled “Some New Perspectives on Games in Transportation Systems Analysis” .   His thesis has inspired a successful NSF proposal,  two publications in Transportation Science, one paper under review at Operations Research, and three conference papers in top Machine Learning conferences (ICML and NeurIPs) .

The 11th PhD student graduated from my group,  Jiayang is set to start in September as an Assistant Professor in the Department of Industrial Engineering and Manufacturing Systems at the University of Hong Kong. 

Congratulations, Dr. Li!

Four papers published recently

It so happened that four papers I co-authored came out in the past month – quite a coincidence. Here is the list.

1. A Day-to-Day Dynamical Approach to the Most Likely User Equilibrium Problem, published in Transportation Science, and to be presented in the International Symposium of Transportation and Traffic Theory next week.

2.  The sustainability appeal of urban rail transit, published in Transportation Research Part A.

3. An Autonomous Modular Public Transit service, joint work with Jane Lin and Xi Cheng at UIC, published in Transportation Research Part C, and to be presented in the International Symposium of Transportation and Traffic Theory next week.

4.Is order-2 proportionality good enough for approximating the most likely path flow in user equilibrium traffic assignment? Joint work with Jun Xie and Liyang Feng at Southwest Jiaotong University, published in Transportation Research Part B.

Hongyu Zheng’s PhD defense

Hongyu Zheng defended his Ph.D. thesis successfully today.  His thesis committee consists of Karen Smilowitz (first from the left), Jane Lin (second from left) and Hani Mahmassani (first from the right).

Hongyu is my eleventh PhD student and the tenth to complete the PhD thesis defense.  He is joining the Industrial Engineering Department at the University of Tennessee, Knoxville as a tenure-track Assistant Professor in August.

Congratulations, Hongyu!

Open-TNM

Finally, this little C++ project that I started more than 20 years ago went open source at Github, mainly thanks to the diligence and passion of Prof. Jun Xie, a former student and now collaborator at Southwest Jiaotong University.  As of now, it contains just one traffic assignment algorithm (the greedy algorithm we developed a few years back), but the plan is to gradually release many more in the coming months.

I began to develop TNM –which stands for Toolkit for Network Modeling– in 2003.  The project is a collection of C++ code I wrote to deal with many different transportation network problems, ranging from origin-destination estimation to microscopic traffic simulation (yes, I had even written a microscopic simulator based on Cellular Automata.   It seems like an extravagant waste of time, with the benefits of hindsight).   In March 2006, when I was wrapping up my PhD research at UC Davis, I wrote a document summarizing the development up to that point, which I have since frequently quoted in my papers that use the code to produce numerical results.  You may check it here.

My development of TNM would continue well into my tenure at Northwestern.  In fact, as late as in 2015, I was still busy putting stuff into TNM.  Over the years I developed a rather flexible and universal GUI based on MFC  that has the ability to accommodate many different types of network problems — at one point I even learned to deal with GIS  functions in my GUI.  I would be lying if I tell you I did not enjoy the thrill of making a machine do what I command….  However, eventually, I realize that I must stop indulging in this hobby as it has become a black-hole swallowing an undue amount of my energy that was in short supply given my increasing seniority and the  responsibilities  that came with it  (I am sure this epiphany came to me a bit too late).

Right around the time I decided it is the time to pass the torch to the next generation, Jun showed up in my lab, first as a visiting PhD student then as a visiting postdoc.   Unlike the vast majority of my other students, he did not mind wresting with the messy C++ code I wrote in  haste.  On the contrary, he seemed to truly enjoyed it.  So, Jun gradually became involved in the development of TNM  and, after  joining Southwest Jiaotong University, has added to the development team his own graduate students.  Without Jun, the publication of TNM on github probably would never have happened.

 

RIVER published.

The last paper I wrote with  Kenan Zhang, who graduated two years ago and is now joining EPFL as Assistant Professor, was finally published in Transportation Research Part B last week.  The paper first went on-line at SSRN in December of 2021.  So, it has been in the review process for nearly two years, during which it was reviewed by three different journals.   I am relieved the paper is finally in print; I know Kenan was exhausted by the lengthy back-and-forth with reviewers and editors.  Rewarding as the experience may be, one can only take so much for each paper without being demoralized.

The above link should work for two months.  After that, if you don’t have access to the journal, you may find a preprint here.

First published book review

David Boyce suggested I should submit my review of Vaclav Smil’s book for publication after I shared it with him.   Smil’s perspective on climate change initiatives might be of interest to regional scientists and transportation planners, he told me.   At his suggestion, I submitted the review to Papers in Regional Science.   To my pleasant surprise, the editor recommended acceptance with minor revision within a day, and  the review was published a week after the first submission, definitely a record for me.  If you don’t have subscription, read it here.  The published version is only slightly different from the blog.

Integrated bus-bike system

After much delay, the last paper Sida and I wrote together came out last week in Transportation Research Part C.   Growing out of the last chapter of Sida’s PhD dissertation, the first draft of the paper was completed before he went back to China in the summer of 2020, at the height of COVID19 pandemic.   In part, the long delay was due to Sida’s transition to his new job at Beijing Jiaotong University.   I am glad he pressed on despite the early setbacks and eventually published the paper  in a descent journal.  Here is the abstract for those who wonder what is an integrated bus-bike system.


Abstract: This paper examines the design of a transit system that integrates a fixed-route bus service and a bike-sharing service. Bike availability – the average probability that a traveller can find a bike at a dock – is modelled as an analytical function of bike utilization rate, which depends on travel demand, bike usage and bike fleet size. The proposed system also recognizes the greater flexibility provided by biking. Specifically, a traveller can choose between the closest bus stop and a more distant stop for access, egress or both. Whether the closest stop is a better choice depends on the relative position of the traveller’s origin and destination, as well as system design parameters. This interdependence complicates the estimation of average system cost, which is conditional on route choice. Using a stylized analysis approach, we construct the optimal design problem as a mixed integer program with a small number of decision variables. Results from our numerical experiments show the integrated bus-bike system promises a modest but consistent improvement over several benchmark systems, especially in poorer cities with lower demand density. We find a large share of travellers, more than 20% in nearly all cases tested, opt not to use the nearest bus stop in an optimally designed system. The system also tends to maintain a high level of bike availability: the probability of finding a bike rarely drops below 90% except in very poor cities.