2023, how is your research progressing? Do you have any insights or experiences you'd like to share with everyone?

Please translate the content into English, following the following principles (in descending order of priority)：1. Output the translated markdown in code block format (required).2. Do not include Chinese characters in the output content.3. Filter out links and images from the translated content.The translated content is as follows：As mentioned above.

Dec 31, 2023 research, academic, phd, post -doctoral, researchers

Academic Year Achievements

Grant
This year marked a bountiful harvest with two independent PI R01s, one Co-PI R01, and a state-level New Investigator Award grant. The guiding principle is to keep applying broadly, keep publishing, keep submitting, and keep networking. I hope to organize and share these experiences in detail later.
Paper
Published a research article in a decently recognized journal within the field. Also published a review, an editorial in a high impact journal, and several co-authored articles. Two potential high impact articles have been submitted for review.
Grant and Paper Review Invitations
Participated in several AHA grant reviews and European grant reviews. Received multiple article review invitations from top journals within the field.
Invited Talks
Delivered a total of 15 invited talks this year, including conference talks, different schools' invited talks (with honorarium), and talks related to interviews.
Tenure
Submitted a tenure application package one year ahead of the schedule.
Offer
Received two formal job offers. I hope to find time to write about some of these experiences later.

Reflecting on the Journey of a Postdoctoral Researcher

There has been no substantial progress in research, mainly focusing on finishing the work from my doctoral period to prevent the project from being abandoned and taken over by others - a recurring cycle.

I’ve noticed that the work of my peers is gradually becoming incomprehensible to each other, which I suppose is the essence of pursuing a doctorate.

My reputation in the field has been growing; several papers submitted have been confidently accepted into top journals, with total citations heading swiftly towards 2k. Attending conferences now feels more upright, although not without its share of issues (to be set aside for now).

I’ve continued collaborating with my undergraduate mentor, recently submitting another paper to Accounts, feeling as though I’ve gone through two PhDs.

Last month, I flew back to LA to give my exit talk. Once I submit my thesis, I’ll be ready to graduate, but I’m not in a hurry as I haven’t found a job yet.

In the second half of the year, I decided not to pursue a postdoc but to enter the TTAP market directly, thus spending a lot of time writing applications, soliciting recommendations, and reliving the nightmare of application season from four years ago.

My next phase involves shifting my focus from learning new techniques and completing tasks assigned by my mentor to proving the forefront and independence of my work, demonstrating that I am not a lower version of my mentor, and validating that my direction is fundable.

While many friends and theoretical colleagues I’ve met at conferences have secured academic positions, I’m facing peer pressure with two failed interviews. I have a big interview next week and hope to seize the opportunity.

I’ve become less inclined to mentor others, realizing the uniqueness of career paths and research tastes. Most advice or requests for help seem futile, saving both parties time and effort.

Interestingly, it’s the activities outside of research that pique my interest, as research is only a small part of life. It’s important to experience the beauty of reality, now embracing a more fulfilled life. This year, I undertook two major road trips exceeding 3k miles, along with organizing hikes, BBQs, camping, and urban exploration, finally matching the average leisure level of North American international students. I’ve also rehabbed my guitar skills and started playing in a band again after a decade, playing some classic rock and popular songs, which has been very enjoyable.

My greatest realization: Work-life balance is crucial. It’s important to socialize, spend time with family, and cultivate hobbies.

I hope to resolve my residency status and secure a teaching position before turning 27, otherwise, I joke about joining the 27 Club like Cobain.

The most significant insight is to devote 20% of research efforts to understanding human relationships and social norms, as this can lead to greater efficiency and effectiveness.

A Roller Coaster Year

This year has been quite eventful. In the first half, the job search was filled with ups and downs, and it wasn’t until April that things finally settled with an offer from a QS top 10 university. It took almost a week of interviews to seal the deal. No articles were accepted in the first half of the year, and it continued that way until September. I felt like I was going to pull my hair out… But things took a turn for the better in October when an article was accepted by AFM, and in November, the third article of my life was finally accepted by Nature Materials. Just a week later, in early December, after nearly half a year of effort, Science Advances was also accepted. Christmas is right around the corner, with streets adorned with lights, and suddenly, I feel like this year has warmth.

Next year, there will be a new role and a more challenging job, but I hope to find a partner first… Friends with resources, feel free to message me!

Reflections on an Eventful Year - 2023

Thank you for the invitation. I have published a total of 6 articles, with five of them being published from July to November this year. Additionally, I successfully completed my Ph.D. and started my job in 2023. It can be said that 2023 has been a bountiful year for research, as detailed below:

January to June 2023: In the Creative Zone

The first half of the year was dedicated to creative work, as follows:

Thesis: The outline was already completed, and in March, I finished the initial draft, made two revisions, and finalized it in June. I also prepared the presentation slides and speeches.

Paper 1: In collaboration with my advisor, I was the first author. The initial draft was completed, followed by two major revisions.

Paper 2: A computer-related paper, a collaboration with my own team, where I was the corresponding author. The initial draft was completed in January, followed by revisions in March, submission in April, and major revisions in June after the first review.

Paper 3: In collaboration with my advisor, I was the first author. It was completed and submitted in December 2022 but got rejected in April. After minor revisions, it was submitted to a new journal (CAS Zone II) in May.

Paper 4: In my field, self-organized collaboration, and I was the corresponding author. The first draft in Chinese was completed early in 2023. After rejection and minor revisions, I translated it into English and submitted it to MDPI in June, which was also rejected. In frustration, I submitted it to the Nature Portfolio series.

Paper 5: In my field, leaning towards the social sciences, self-organized collaboration, and I was the corresponding author. The outline had been prepared since March-April, but I lacked a definitive new idea. Finally, in mid-May, I found it, completed the initial draft in June, submitted it to MDPI at the end of June, and it was rejected immediately. Annoyed, I submitted it to a CAS Zone I journal.

Paper 6: Similar to Paper 1, in my field, in collaboration with my advisor, and I was the first author. The initial draft was completed, followed by one major revision.

July to September 2023:

Thesis: Successfully defended in mid-July, with a few minor revisions completed by the end of July. The final version was submitted at the end of August, and I officially graduated in early October.

Paper 1: Submitted to CAS Zone I, received major revisions in the first review, and the revisions were completed by the end of September.

Paper 2: Accepted at the end of July, but no further communication until mid-September when they requested formatting.

Paper 3: Major revisions in the first review in mid-July, second-round major revisions at the end of August, and acceptance on September 2nd. Online publication in mid-September, and indexing in late October.

Paper 4: Major revisions in the first review in early September, revisions completed by the end of September.

Paper 5: Major revisions in the first review at the end of July, revisions completed in mid-August. Minor revisions in the second review at the end of August, accepted in early September, online publication at the end of September, and indexing in early November.

Paper 6: Second major revision completed.

Paper 7: A semi-finished paper from before, strengthened in August-September, and submitted to CAS Zone I in late September.

October to December 2023:

Thesis: Officially graduated in mid-October, received the electronic diploma in mid-November, and the physical diploma in mid-December. Completed the study abroad certification process.

Job Entry: Started the new job in mid-late October and formally entered in early December.

Paper 1: Minor revisions in the second review in mid-November, review completed in mid-November, acceptance at the end of November, just finished proofreading, not yet online.

Paper 2: Surprisingly, it went online at the end of October, but I received an official email from the journal several days later. The publication process was indeed mysterious. Indexed at the end of November.

Paper 4: Accepted at the end of November, online proofreading completed in early December, waiting for online publication.

Paper 6: Final draft completed in early November, submitted to a CAS Zone I journal.

Paper 7: Under review.

Paper 8: Came up with an idea at the end of October, completed the initial draft in early November, submitted to a CAS Zone I journal, and was suggested to transfer to another CAS Zone I journal with a social science focus. Under review.

Paper 9: Started in mid-September, mostly completed by mid-October, but I forgot about it due to starting my job. Remembered it in early November, finished it in mid-November, and submitted to the Nature Portfolio series. Under review.

Up to this moment, Papers 1-2-3-4-5 have been accepted. Papers 6 and 7 are in the first review stage. Paper 6 has one reviewer who has finished the review, but we are yet to find a second reviewer. If everything goes well, Papers 6 and 7 should return with feedback before the Chinese New Year. Papers 8 and 9 might also have new developments.

Reflections and Thoughts

I have many reflections and thoughts, but here are some key points for a brief summary. Feel free to message me if you’d like more details.

The institution has been very supportive, with funding for settling in, research, and low eligibility criteria for receiving these grants. I am grateful for the opportunities and the beautiful surroundings.
Research requires bold ideas and speed. I recently found a master’s thesis online, which closely resembled my Paper 3’s concept (two out of three innovative points were the same). At the time, when writing Paper 3, I combined these three innovations when my team members were not entirely convinced. If I had waited to convince them first, the results might not have been as good as they are now.
For papers that I lead, I usually submit them after they are written, using reviewer comments to supplement and revise the paper. My advisor prefers a more thorough approach, making major revisions before submission. My personal suggestion is that newcomers to research and those making major life decisions (such as buying a car, house, or accepting a job offer) should consider the latter approach—carefully plan before acting. For those with solid writing skills, the former approach can be more effective. In either case, both methods are valid for publishing articles.
In my research, I prioritize innovative ideas, focusing on what “no one has done before.” My advisor’s approach is more detailed, seeking innovation while also caring about the logic and writing quality of the paper. I believe both approaches have their merits. Writing an article is essentially storytelling, and a good story is interesting and well-founded. Currently, I have published three articles as a corresponding author in my self-organized group and three articles as the first author under my advisor’s guidance.

Expectations for 2024

In 2024, I will continue to transition into a new role, adapting to the responsibilities of a university teacher, writing, teaching, participating in competitions, and more. Perhaps I will have less time for research compared to my Ph.D. and early career days, but I am excited about new challenges. Writing, sharing knowledge in class with students, and engaging in competitions to exchange creative ideas may lead to new discoveries and more papers in 2024.

Today is a beautiful sunny day, with clear skies and a gentle breeze. Facing the warmth of the winter sun, I am extremely satisfied and grateful for 2023, and I look forward with great anticipation to 2024!

Reflecting on Progress in 2023

As 2023 comes to a close, it’s a good time to provide an update on the progress made throughout the year. This update reflects on your expectations for your research journey in 2023, highlighting both achievements and some areas of regret.

In our research group, we submitted four papers to ICCD 2023, focusing primarily on performance optimization of Transformer-based neural networks like LLM and ViT. This research traverses the realms of software-hardware co-design and hardware architecture design. For instance, in the paper titled “ViTframe: Vision Transformer Acceleration via Informative Frame Selection for Video Recognition,” we reduce the amount of data fed into the neural network using clever frame selection strategies and effectively support customized software algorithms through a reconfigurable array.

This year, my personal focus has been on the optimization of point cloud neural networks, given the importance and advancements in the autonomous driving industry. Consequently, my students have continued to build on the work that received the Best Paper Award at DATE 2023, submitting two papers on point cloud acceleration to DATE 2024. These papers primarily address performance bottlenecks in the three stages of point cloud processing (FPS, Neighbor search, and Feature computation) and propose innovative solutions. Additionally, I published an architecture design proposal in IEEE Transactions on Parallel and Distributed Systems (TPDS) regarding the optimization of multi-frame point cloud processing. This work extends spatial locality to temporal locality, further enhancing real-time point cloud processing.

In the realm of research, the most notable accomplishment this year has been the acceptance of a paper that had been in the pipeline for two years, which will be presented at ACM Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) 2024. It’s a testament to the challenging journey from conceiving an idea to finally publishing a paper.

Furthermore, the course I taught, “Parallel Computing and Program Design,” has concluded. It covered modern GPU architecture, CUDA parallel programming, and related technologies. For those interested, course materials are available for download on my homepage: {Zhuoran Song}.

No diploma, no cap, zero papers.

Busy as a bee, but the financials look like a year-long gap.

My Second Year of PhD: Challenges, Growth, and Aspirations

Thank you for the invitation.

2023 marked the second year of my doctoral journey, stepping into the second year of the Ph.D. program. Although I’m not as accomplished as some of the senior researchers, I’d like to share my experiences and gains from this year.

In September 2022, I joined a research group with a returning overseas-educated professor who is only seven years older than me. He is energetic, approachable, and very nice. We immediately defined our research direction, primarily focusing on multimodal learning, encompassing both understanding and generation tasks, as well as 2D and 3D aspects. To be honest, this field was quite competitive, with numerous research groups and new papers appearing on arXiv daily. Fortunately, we identified a research problem and made progress, leading to a submission to CVPR 2023, which was subsequently rejected. We then submitted to ICCV, only to face rejection once more. This phase lasted for about 3-4 months. During this time, I experienced severe emotional turmoil, avoiding conversations with others, struggling to accept rejection, and allowing my research life to spiral into chaos. It took me nearly six months to regain my composure.

Despite the rejections, we managed to complete a piece of work. We incorporated feedback from multiple rounds of reviews, conducted experiments, improved results, restructured the paper, refined details, and finally submitted it to TPAMI in late October of this year. The waiting game began…

From April to August, the entire research group was heavily involved in project applications. Although it was cumbersome, our mentor emphasized its necessity for further development. I consoled myself by considering it a necessary phase of growth that would eventually pass. By September, the project applications had reached a conclusion, and I entered the second year of my Ph.D. program. At this point, I didn’t have a concrete research idea but focused on the field of image editing and generation, hoping to make some progress. In September and October, I experimented with various code and ideas until I finally identified a viable research problem. I’ve been working diligently on it, and as of December 2, 2023, I’ve made some progress. I hope to wrap up this work successfully before New Year’s.

In 2023, although I didn’t achieve significant research results, I strived to continue my personal growth. On January 1, 2023, I co-founded a paper reading community on Bilibili called PaperABC (a name my advisor suggested, which I really like). We consistently explained the latest arXiv papers, and our fan base grew to over 6000 followers with more than 100,000 views. Due to personal reasons, I couldn’t maintain regular updates, but I’ve recently started updating it gradually.

I also attended an academic conference, China3DV, where I had the opportunity to meet many distinguished researchers. I hope to learn from them and continue progressing in my academic journey.

I aspire to remain focused on my research endeavors, overcoming emotional challenges, and patiently waiting for the day when my hard work bears fruit.

Year in Review: Research Papers and Stories

I took a look at my Google Scholar profile, and this year has been quite productive. Since arriving in Hong Kong and Shenzhen, I’ve been deeply immersed in academia. I often joke with my friends that I’ve become a full-time writer - either I’m editing papers or on my way to do so.

I’ve selected ten journal papers accepted this year, and I’m looking forward to the first half of next year, which seems relatively free from teaching commitments. This will give me the chance to organize and share the achievements of my research group. Here, I’ll briefly discuss the stories behind these papers.

[1] W Jiang, C Lu, C Wu, Robust Scheduling of Thermostatically Controlled Loads with Statistically Feasible Guarantees, IEEE Transactions on Smart Grid, 2023.

This paper marked the beginning of our research group’s exploration into “Statistical Feasibility.” When it comes to stochastic optimization, whether it’s chance constraints or robust optimization, we often overlook the impact of sample observations. For example, in chance-constrained stochastic optimization, we typically estimate distribution information within the chance constraints. However, if our sample observations are unreliable, this estimation may fall short, affecting the effectiveness of stochastic optimization. Robust optimization, including distributed robust optimization, also overlooks this aspect. Distributed robust optimization mainly focuses on assuming a distribution, which may not be the best fit. From this perspective, we combined chance constraints and robust optimization.

The mathematical solution is relatively simple, adding an outer layer of chance constraints to account for the uncertainty in sample observations. This nested chance constraint is then transformed into a robust optimization problem, specifically one that characterizes uncertainty sets. If you’re interested in the details, feel free to read our paper.

[2] Y Zhang, Q Liu, J Sun, C Wu, The optimal dynamic regret for smoothed online convex optimization with squared l2 norm switching costs, Journal of the Franklin Institute, 360 (6), 4297-4330, 2023.

[3] C Lu, J Liang, W Jiang, J Teng, C Wu, High-resolution probabilistic load forecasting: A learning ensemble approach, Journal of the Franklin Institute, 360 (6), 4272-4296, 2023.

These two papers were featured in a special issue of the “Journal of the Franklin Institute.” I’m not very familiar with this journal, but at the time, I had two students working on distinct but substantial projects. One focused on load forecasting, an area I’m not particularly fond of due to the need for extensive comparative experiments to impress reviewers. However, most of our papers emphasize theoretical analysis, especially tailored solutions for power system problems. While load forecasting can also be customized, this paper primarily contributed a method for improving linear ensemble with learning.

The other paper may find applications in data centers or online scheduling for microgrids, but its relevance to large power grids appears limited for now. Given the opportunity, we submitted it to this mathematically-oriented journal.

If anyone has ideas on how to customize these methods further, your input would be appreciated.

[4] N Gu, J Cui, C Wu, An Auto-Tuned Robust Dispatch Strategy for Virtual Power Plants to Provide Multi-Stage Real-time Balancing Service, IEEE Transactions on Smart Grid, 2023.

To address the customization issue mentioned for [2] and [3], we tackled a problem related to virtual power plants and real-time balancing services within distribution grids. This is an online real-time decision-making problem, and we aimed to ensure system stability while incorporating machine learning or computational flavors. We combined dynamic programming and distributed robust optimization, which, under certain conditions, are equivalent modeling methods.

Subsequently, we explored how machine learning techniques can speed up dynamic programming. While some reviewers considered this a reinforcement learning problem, our problem’s clear physical structure and deterministic state transitions didn’t necessitate reinforcement learning. Instead, we simply estimated the value function within dynamic programming to avoid dimensionality issues.

[5] C Lu, N Gu, W Jiang, C Wu, Sample-Adaptive Robust Economic Dispatch With Statistically Feasible Guarantees, IEEE Transactions on Power Systems, 2023.

This paper is the second in the “Statistical Feasibility” series and applies the concept to economic dispatch problems. It further demonstrates the effectiveness of this method. If you have the time, I highly recommend delving into this valuable tool.

[6] Q Huang, W Jiang, J Shi, C Wu, D Wang, Z Han, Federated Shift-Invariant Dictionary Learning Enabled Distributed User Profiling, IEEE Transactions on Power Systems, 2023.

The inspiration for this paper came from a discussion with my students about Dictionary Learning, which I found to have similarities with federated learning. After all, one of federated learning’s core advantages is the ability to train models without sharing data. Dictionary learning can achieve similar results.

So, we applied these methods to user profiling tasks in the power system, combining two interesting concepts. The methods we used for dictionary learning and federated learning are relatively straightforward, and if you’re interested, we welcome suggestions for improvement.

The reason this paper has co-first authors is due to my student’s heavy workload after graduation. They were working in a demanding job at a large tech company, facing long working hours and weekend overtime. As a result, I asked Wenqian Jiang to take over.

[7] W Jiang, J Huang, G Xu, C Wu, Sample-Oriented Electricity Storage Sharing Mechanism Design With Performance Guarantees, IEEE Transactions on Smart Grid, 2023.

This paper is a result of training my students in research paper writing. Since exploring “Statistical Feasibility,” we’ve been more focused on concepts like sample complexity. I encouraged my students to expand upon my previous work in the sharing economy. In previous work, we assumed abundant data and could estimate the distributions of various random variables well. Now, we’re asking further questions: What if we don’t have enough data? How does the quantity of data affect our final decision when data is scarce?

If you’re interested in such questions, I invite you to read our paper and consider how this approach could be applied in your own research.

[8] C Lu, J Cui, H Wang, H Yi, C Wu, Privacy Preserving User Energy Consumption Profiling: From Theory to Application, IEEE Transactions on Smart Grid, 2023.

Surprisingly, the number of papers related to differential privacy has been relatively low this year. Those who follow my research may know that our group has made progress in understanding how different parameters of differential privacy impact tasks in power data analysis. This paper explores the effects of different differential privacy protection methods on the final energy user profile. We also begin to address the issue of data pricing. For instance, if you sell a piece of data without adding noise, the price might be p. But now that you’ve added noise, it’s unlikely you can still charge p. We establish the relationship between the magnitude of noise injection and the discount rate. However, we took a shortcut here, as fully exploring pricing would require more than a ten-page paper. We welcome input from experts in this area.

[9] C Lu, J Liang, N Gu, H Wang, C Wu, Manipulation-Proof Virtual Bidding Mechanism Design, IEEE Transactions on Energy Markets, Policy and Regulation, 2023.

We plan to extend this paper further. Gaming under network constraints is a challenging problem, but when it comes to virtual bidding, due to its relatively small trading volume, people often assume they don’t need to worry too much about network constraints.

But in any game, there’s room for strategic behavior. We explore whether mechanism design principles can encourage cooperation and reduce unnecessary complexity in decision-making. If you’re interested in these types of problems, we welcome you to read our paper and engage in discussion.

[10] J Cui, J Wu, C Wu, S Moura, Electric vehicles embedded virtual power plants dispatch mechanism design considering charging efficiencies, Applied Energy 352, 121984, 2023.

This paper addresses the concept of virtual power plants, which has gained recent attention, and makes electric vehicle scheduling more practical. For instance, the charging and discharging efficiencies of each electric vehicle can vary significantly and are often temperature-dependent. As such, how might these practical considerations change electric vehicle scheduling? We welcome you to read our paper and engage in a discussion.

Thank you for your attention!

Slowly getting back on track, aiming to maintain a pace of two papers per year. This year, one first-author paper got accepted in Nat. Comm., and I’m currently making minor revisions for a paper in Chem. Soc. Rev. Hoping for smooth submissions next year.

Academic Update

Papers

Published a paper in Nature Sustain on ammonia recovery from cattle manure. Surprisingly, a writer reached out to us voluntarily.
Collaborated with the advisors on a major news piece. Anticipating the results next year.

Funding

Collaborated with a professor to secure NSF FMRG funding with ease.
We’re seeking postdocs interested in flow cell and electro-synthesis for an opening in our lab.

Projects

Continuing our research on cattle manure to investigate electron transfer.
Pursuing an exciting project, although it leans more towards chemistry than engineering, we’re determined to make progress. Hoping to wrap it up next summer.

Reflections

Spent a joyful two months at home during the summer break. Realized that not everything is under my control, and going with the flow can be satisfying.
While I often yearn for some downtime, it seems that everyone expects me to stay productive.
Engaged in gardening for six months, and it feels great.
Taking a laid-back approach to finding a postdoc position, not expecting much, but maintaining good mental health. Embracing both laziness and happiness.

Insights

Reflecting on four years of experimental work in the lab, two and a half years of teaching, and one and a half years of grant-writing and project management. Each stage brings its own challenges.
Well-funded research groups are becoming rare, and it’s clear that academic positions are not easy to come by.

Academic Progress Update

It seems there weren’t suitable questions in the subfield for me to answer, considering I’m not well-versed in mathematics, despite my nominal affiliation with the Mathematics department. Overall, this year’s research has progressed relatively smoothly, although the challenge is that the issues of interest remain within the established framework of 2019/2020. However, there have been notable technical breakthroughs, such as merging the concepts of space-bounded quantum computation from 2019 and QSVT from 2021.

Since the end of last year until February this year, I dedicated time to several less critical issues:

Firstly, I worked on the polarization lemma for quantum distances, leading to some ideas about open problems in QSZK. This also exposed some deep-seated misconceptions.
Secondly, I explored structural complexity results for stateQMA, significantly enhancing my understanding of recent results in state-synthesizing complexity classes.

By mid-February, I rushed to submit two papers to conferences, sticking to my prior strategy. However, rushing has never been a good idea, and this year’s conference submissions were exceptionally competitive, with one paper receiving toxic reviews. Nonetheless, this strategy proved successful, helping me secure a fellowship from the university, temporarily alleviating my financial constraints. However, this fellowship may not cover my entire academic journey. Overall, the university’s funding support for international students is relatively friendly compared to other institutions in Japan.

It’s worth mentioning the submission experience for the QSZK paper. In hindsight, starting with a more prestigious conference might have been a better choice. Initially, I received feedback on the polarization lemma from an expert that surpassed my understanding, and it took me several days to grasp their insights. Unfortunately, when submitting to another conference (referred to as T2), I didn’t make significant revisions and even had errors in side results. Surprisingly, the overall review quality was good. Upon returning in September, I spent over a week revising it thoroughly. Later, when submitting to a different conference (not the one on par with F), I received a toxic review among the three provided. In the rebuttal, I meticulously addressed every issue, nearly reaching the 1,000-word limit. Each toxic review serves as a reminder that some PC members are irresponsible and do a poor job.

In February, I also achieved a space-efficient QSVT for the sign function. Interested readers can refer to my previous answers on Singular Value Decomposition (SVD) applications and determining the strong connectivity of a directed graph with logarithmic space algorithms for background. In early March, I noticed that a fellow postdoc’s work in my group could potentially reach quantum logspace. Consequently, I dedicated a month to submit to F conferences, focusing on space-bounded state testing problems. This paper is close to tier 1 results, and it’s interesting to note that for two-sided (or one-sided) error state testing (state certification):

In the time-bounded case, the computational difficulty depends on the selected distance measure (l1 is much harder than l2).
However, in the space-bounded case, the computational difficulty remains the same.

Space-efficient QSVT also exhibits some excellent properties, allowing us to naturally formulate a one-sided error version of a new BQL-complete problem. Rushing is never a good idea, although involving my advisor in the initial introduction writing improved its quality, roughly half of it was barely readable technical details. After receiving F conference reviews, we revised the introduction’s narrative, demonstrating that even with expert involvement, the introduction doesn’t necessarily reach its final form, despite saving one or more submissions objectively. The revision took the entire summer. Subsequently, I submitted to I and Q conferences, but both received borderline rejections. It’s worth noting that I feel if the inclusion work of some (top) conferences in this field is treated as a source, it might yield some random bits for seeded extractors.

Towards the end of last year, I received two review requests related to my interest at the time (a direct proof of QIP being in BQPSPACE). I was too busy with administrative tasks during the first request, but during the second request, I dedicated a whole week to carefully examine it. It turned out that the technical details were poorly written. Still, I eventually understood the work and derived different results based on their techniques.

In total, I received seven review requests this year (excluding one that was far from relevant and declined). Only one was related to my current interests, but after reviewing the proof, I found the technical aspects less appealing. The remaining four were related to my prior work, but I’ve lost interest in this topic. There were also two that I understood but were less relevant. Unfortunately, this year did not bring the same luck as the end of last year in terms of review assignments, and these reviews did not significantly benefit my research.

Research Progress and Encouragement

This year, my advisor changed positions, and my research progress has been quite modest, leaving me feeling somewhat inadequate.

However, today, while checking ResearchGate, I discovered that a research work I published earlier this year has received 10 citations, which has greatly encouraged me.

Although I worked on this project for a long time, faced several rejections, and couldn’t achieve the initial expectations when submitting it, acknowledging my limitations, I’m genuinely pleased with these citations. It was even mentioned in a mini-review by a prominent figure in the field, reigniting my interest.

In the final year ahead, I aim to produce some inspiring work, rectifying past regrets. It’s a test of my ability to conduct independent research and determine whether I can continue in this field.

This year’s biggest research achievement is that I didn’t do anything. I’ve been super busy, but I feel light-hearted deep down.

— Screw research papers, screw projects, screw hats.

In addition, the biggest achievement of the year is that both of my apprentices successfully found jobs in such a chaotic job market.

Especially the one I was most worried about, unexpectedly secured a job at a 211 university in the fall.

Suddenly, I feel like thanking the heavens.

Thank you, God, for your blessings.

Grateful that all my apprentices come from families with virtues.

Reflections on Research and Life

I haven’t made much progress in research this year, but I have gained some valuable insights. I’d like to share my thoughts here, like a confessional, for the encouragement of fellow researchers.

I’m a second-year direct Ph.D. student in theoretical physics, specializing in quantum physics, at a research institute in China.

In the first half of the year, I completed enough credits and passed the Ph.D. qualification exam after returning from the National Day holiday. Next week, I will be presenting my thesis proposal, which is a tentative step into the world of research. (However, according to my advisor, I’m not fully on the right track yet.)

This period has been quite challenging for me. Since my second year of college, I had been exposed to and interested in gravitational physics. My undergraduate research was related to black holes, and I was quite familiar with the principles of general relativity. However, after extensive research, discussions with my undergraduate advisor, and conversations with professors in the field during the National Day holiday, I decided to temporarily leave this area and venture into a relatively unfamiliar field.

Now, I am primarily working on problems related to quantum decoherence. It’s not entirely unfamiliar because whether it’s quantum optics or open quantum systems, it’s fundamentally the application of quantum mechanics. I started learning about “quantum mechanics” in my sophomore year.

I say it’s unfamiliar because compared to gravity and black holes, whether it’s the history, development, cutting-edge research, or specific techniques in this new field, I’m still in the learning phase. If I had started working in this field a few years ago, I would have been more comfortable.

So, now when I look ahead, I see a myriad of challenges and uncertainties. I don’t know which specific techniques to use for certain problems, what to learn from the thick reference books, and where to start. I can only experiment and try different approaches.

Of course, I don’t mean I have absolutely no idea; it’s just an exaggeration. When my attempts fail or I’m clueless, discussing with my advisor always provides me with valuable insights. However, research cannot solely rely on the advisor; they guide and collaborate, but many problems require personal solutions.

Fortunately, my research group has an excellent environment and is considered one of the top in our field in China. My advisor, despite being experienced, is still actively involved in research. He stepped down from an administrative position this year, giving him more time to interact with us. It’s easy to schedule meetings with him, and he often holds discussions, even after a full day of teaching. When it comes to selecting research topics, it’s mostly based on scientific merit rather than anything else.

In summary, I am content with my external environment and often feel that my abilities do not match the quality of the platform I have.

The most significant difficulty I’ve faced is related to my health, primarily my physical health. I’ve come to realize that in the end, research demands a lot from your physical well-being, and this has been my biggest insight this year.

I won’t dwell on the increasing number of white hairs; the real issue is my physical health. After an illness in May, I developed severe brain fog that persists to this day. I suspect I might have long COVID, which has significantly impacted my memory, concentration, and overall cognitive abilities.

During the same period, some dormitory-related issues disrupted my sleep patterns. I would often wake up multiple times at night, have early awakenings, or find it extremely difficult to fall asleep. This has taken a toll on my overall well-being, especially considering the seasonal changes, and the fact that I lost contact with my only confidant due to unforeseen circumstances after the National Day holiday.

It’s a frustrating state to be in. Most of the time, I have clear and rational thoughts, but I can’t control my emotions and feelings, which in turn affect my behavior. I’ve been familiar with this condition for many years, but the long-term physiological effects of such states on neurotransmitters inevitably impact my mental faculties, leading to difficulties and pain.

These challenges have become somewhat manageable and adaptive over time. However, there were times when I became overwhelmed, fearing making mistakes, and this led me to seek medical help at a local hospital. The doctor encouraged me, saying that reaching this point after years of dealing with such issues is already quite an achievement. Medication can only help to a limited extent and often swings my mood from one extreme to another, like a pendulum. So, unless it’s unbearable, I only take medication when I struggle to fall asleep and need something to force myself to sleep.

At present, there doesn’t seem to be a better solution. Over the years, it’s been challenging for me to discern whether my emotions are normal or not. I tend to suppress my feelings because I’m unsure whether my thoughts and emotions are normal. Sometimes, within seconds, my thoughts and emotions can change, and I don’t know how to determine what’s genuine. Regardless, whether through medication, religion, personal experiences, or certain responsibilities, I’m constantly trying to find a balance.

So, researchers, remember to prioritize your health, especially sleep. Sleep isn’t just about the total duration; maintaining a regular sleep-wake cycle is crucial, and ensuring adequate exposure to natural light is also essential. While exploring the boundaries of science in basic research, we often lack a sense of achievement and positive feedback. It’s important to intervene early if your initial emotional challenges turn into severe physiological issues. Once emotional problems transform into severe physical illnesses, it becomes a complex and difficult situation.

Another realization with age is that many of my former classmates have stable jobs, some have even settled down with families, homes, and cars. Although some have modest incomes, they lead happy lives in smaller towns. Looking at myself, around the same age, I sleep in a small dorm room on a narrow wooden bed, with no personal space. I move every one or two years, and my monthly income is not much higher than that of a bubble tea shop employee (even though this income is not considered low among Ph.D. students in China, it’s still not enough to afford a single room outside in Beijing). This situation sometimes leaves me with an indescribable feeling.

Objectively, compared to most graduate students, my conditions are already quite good. It’s a far cry from the conditions our predecessors had in the past when they were in remote areas like Jinshitan and Luobupo. It’s a world of difference from those who are struggling to make ends meet. Logic tells me to avoid any romanticization of bourgeois desires, but there are times when I feel like my current state is merely survival, far from actually living.

(This contradiction is also one of the sources of my pain. I know I shouldn’t have certain thoughts, but they inevitably creep in.)

They say research often means giving up material comforts, but it still feels surreal to me. On one hand, I’m pushing my intellectual limits while compromising my physical health. On the other hand, I’m barely making ends meet. It’s not easy to ask for money from my family at this age, so after deducting expenses for necessities like food, shelter, transportation, water, electricity, and the internet, there’s hardly anything left if I want to do anything more.

Of course, if I aim for a 100% Engel coefficient, I would have a comfortable life. However, one’s life cannot be just about eating. With an Engel coefficient of around 0.5, my life often involves calculating how much money I’m spending on each meal, avoiding going to the supermarket, rarely indulging in expensive fruits, and occasionally splurging on beverages. In my dorm, we use large 5L bottles of water to refill our empty water containers for free, as it’s cheaper than buying bottled water.

I can’t afford to get sick either, especially when I think about the medical expenses. These are expenses I cannot bear on my own. When I visited the hospital in the past two months, I had to ask the doctor to prescribe the cheapest medication possible, and I only manage to afford it with medical insurance.

Sometimes, I have to borrow money from my family to get through the month when I overspend. During these times, I feel a significant sense of guilt.

However, looking back, there weren’t many other choices. Even now, there are countless opportunities. Even during the toughest moments, when I’m so exhausted that even breathing feels tiresome, I think back to the numerous crossroads where I had to make choices. If given the chance, I would still choose the path I’m on now. I remember the difficult two months spent in the hospital seven or eight years ago. If I could survive those challenging days, what’s there that I can’t overcome?

In many difficult moments, I considered giving up, but I’ve never actually given up, and I’ve never regretted my choices. I find ways to get some sleep, and the next day, I wake up and continue to engage my brain in solving those challenging problems.

I’ve shared a lot, but in the end, let’s encourage each other.

Reflections on Research and Life

It’s that time of year again for year-end reflections. I’ve noticed that last year’s year-end reflection in my drafts folder contained only my gloomy mood from the end of the year and this link: “Mixing in Circles Hinders the Birth of New Ideas - Let Science Return to its Original Intent.”

This year is different from the last. Work has been slow and filled with confusion, but as a working person/social being, I’ve grown a lot. For the first time, I’ve felt a complete sense of control over planning my life based on my own decisions, and the freedom to say, “I won’t do it if I don’t like it.” People often grow through significant ups and downs, and note that these ups and downs are subjective, not objective. My “ups and downs” this year have mostly been in my social life. The first big change made me completely disillusioned with authority and eliminated my idol worship; the second made me almost give up on academia. Although I bounced back temporarily and didn’t quit, mentally I’ve separated my work from academic pursuits, keeping the flexibility to leave at any time.

Firstly, I have genuinely accepted the idea that “where there are people, there’s a community.” One should never judge the world by a “good” standard. Every decision and action should depend on “what the world is like” rather than “what the world should be like.” In academia, everyone knows that it should be equal, free, objective, and neutral, but as long as there are people involved, these ideals cannot be achieved. These can only be good visions, and we can’t even demand them. Academic ability and personal character are actually two separate dimensions, and I believe that in a professional context, the first dimension is more important. It’s important to separate these two dimensions. Moreover, sometimes having good character is not enough. Human energy is limited, and if you want to change the academic culture with your own power, it’s inevitable that you’ll sacrifice time spent on actual research. Understanding this trade-off is crucial. In most cases, I’m against extremes. You can’t ignore the influence individuals have on the culture, but you can’t spend 100% of your time on it either. Change takes time, especially for such a massive system. You can spend all your time trying to change the culture, but if you can’t see the results in your lifetime, is it worth it?

Furthermore, suffering comes from having too many desires. But people’s (at least mine) innate desires are not that numerous. After reflecting on a round of suffering, I found that the source was mixing many common desires with my personal desires, creating immense pressure on myself. It’s essential to clarify which desires belong to the people around us/society and which desires genuinely belong to us. Only by recognizing our differences and being willing to admit that we are different can we genuinely focus our time on satisfying our own desires. I used to think that if you were a genuinely “good” researcher, your desires should focus on doing meaningful non-trivial work as much as possible. Now I think that I was too narrow-minded. By the way, something a junior student said to me yesterday struck a chord. I said I envied the monks who sweep the floors; why can they be so calm and composed? She said it’s because they have books they love to read in the scripture library.

The last point is a bit tricky, and I haven’t thought it through clearly, so I may not explain it well, but it’s the main argument I use to convince myself not to try too hard. “Effort” is, in fact, one of the means to success summarized by society, aside from the old adages of “the scope is unclear,” “there are more efficient means in today’s society,” and “society’s success is different from personal success.” One of the shortcomings I find more convincing is that these are laws and means of success summarized under societal significance, and they are means to success under societal significance. But the laws of society and the laws of nature often contradict each other, and this is also a point where I convince myself not to care too much about instrumental rationality. From the macroscopic perspective of all humanity, one of the favorable factors in species evolution is variation and diversity. At the beginning of variation, “nature” (or God/fate) also doesn’t know which direction is right, so it undergoes random variation. Therefore, on this micro-scale of an individual, before reaching the end of life, we cannot arrogantly assert that our variation has what effect on the entire species. Even if we reach the end of life and discover that we have taken the wrong path, the fact that we “took this wrong path” in itself has meaning. This is, indeed, a bit tricky, and it seems like I’m analyzing a shortcoming with a more cold-blooded approach, and I’m not sure if it’s correct. But recently, I’ve found a lot of stability in this idea.

Anyway, these are my gains for this year. Besides that, after teaching four quarters, I’m no longer afraid/disgusted by teaching, and I even quite like some of the courses. This can be considered a gain (in terms of teaching, I think I’m lucky. In the first three semesters, I taught precalc, calc, and optimization, covering various course types. In the fourth semester, I taught two courses, which was a stress test and I passed smoothly). Regarding soft skills in my work, I’m still learning and not doing well. Actually, the gains mentioned above may not necessarily be a good thing. These thoughts make me feel even more out of place in my current circle, and I acknowledge my non-conformity. Yesterday, a colleague said she felt exhausted, that there seemed to be no end, that she was constantly climbing upwards, but didn’t know how to stop/quit. I said I was in a state of floating around, ready to quit anytime, not caring about it. Nothing particularly profound. Wishing you all good luck and success.

Reflections on the Year

Published three papers, two in first-tier journals and one in a second-tier journal. When I first submitted as corresponding author, I often faced rejections. However, starting from this year, the journal level has reached the standard of a Ph.D./postdoc period.

Participated in a small collaborative project. Applied for a provincial-level project in the first half of the year, but it was unsuccessful.

Successfully guided my first graduate student to graduation, and I’m pleased to see them continuing in our field. We occasionally exchange messages on WeChat. The new graduate student I recruited seems to be a bit laid-back, and we’re still adapting to each other.

Attended conferences in many places and had insightful discussions with some senior researchers and new friends, deepening my understanding of the academic community.

Handled various miscellaneous tasks throughout the year. That’s about it for this year; more effort will be required next year.

Lastly, health comes first. Do your best in everything, but don’t force things. Hoping for fewer late nights in the future!

Reflections on the First Year of Postdoc

My first year as a postdoc has concluded. I relocated from the United States to Europe. During the first half of the year, I focused on advancing research projects, while in the second half, my primary effort went into seeking faculty positions. In June of this year, when I came across the DPO paper, I had a feeling that it would gain recognition, and today, I saw that DPO received an award at NeurIPS this year. I wonder if this is an improvement in my research taste.

The most significant change in my research projects was the initiation of collaborations with some PhD students. I developed a deeper understanding of collaboration dynamics. Typically, I propose projects, set goals, define technical routes, and anticipate results. I present contributions and feasibility in a visually appealing way to attract PhD students to collaborate.

The key principle here is to identify common interests among multiple parties in the collaboration and design projects based on these shared interests. This transformation involved two major shifts:

I transitioned from being the one who answers questions to being the one who asks questions. I ask what they want to do, what interests them, and what they aspire to achieve.
I moved from thinking about how to do things to determining what to do, why, and how it aligns with both parties' needs.

The guiding principle for initiating collaborations is not to make the other party discover the gold in your ideas but to help them find it and place it in front of them. It’s crucial to understand your role in the collaboration.

Specifically, I collaborate with PhD students who have their own advisors. In this context, I function more like a tech lead in a company. I provide ideas, concepts, and take responsibility for solving challenging technical issues on the project roadmap. My core belief is that if I can’t solve these technical challenges or provide a clear direction, it’s unreasonable to expect others to invest their time. There are three reasons for this:

They are unlikely to be more familiar with the problem than I am.
Their time isn’t compensated by me.
The problem was proposed by me, and if I can’t solve it but expect others to, it’s somewhat irresponsible and can easily lead to collaboration failure.

This collaboration model is also applicable for PhD students collaborating with each other. For faculty members, having their own PhD students provides some room for exploratory projects, but student acceptance and confidence should still be considered.

Another significant shift occurred in the establishment of a research pipeline. I prepared three long-term research directions, each with a grand objective, and decomposed each major goal into multiple sub-projects. As a result, I shouldn’t have any project-related worries for the next few years, as long as I follow the plan diligently.

Apart from these directions, given the current popularity of LLM, I’ve also delved into it, and there will be some projects involving LLM in the future. In June of this year, when I came across the DPO paper, I had a feeling that it would gain recognition, and today, I saw that DPO received an award at NeurIPS this year. I wonder if this is an improvement in my research taste.

Regarding the faculty position search, I’m starting to gain some insights. I’ll summarize the specifics in a future post.

Achievements – New Work:

A solo-authored paper at NeurIPS proposing a new contextual bilevel optimization framework. It is more versatile than traditional bilevel optimization (where the follower optimizes against the leader’s decisions) as it can handle scenarios with multiple followers, each having different optimal responses, or when the follower’s optimal response depends on both the leader’s decision and additional side information. The efficiency of our algorithm is not influenced by the number of followers or the support size of side information, making it as fast as solving with a single follower. In other words, we prove that a semi-infinite constrained problem can be solved as quickly as a single constraint.
Two collaborative papers at NeurIPS workshops. Hoping they get published in journals/conferences next year.
A paper on solving a class of nonconvex optimization problems to global optimality.
A paper on model-based robust Markov decision processes and using active learning to reduce the required number of samples.

Among the previously written papers:

A paper submitted to Operations Research received a major revision. The feedback was good, and our responses effectively addressed the raised issues. The paper’s title might be too realistic to convey its core points, but the content is quite interesting.
A paper on minimax generalization received positive reviews. Hopefully, it will be published next year.

Looking ahead to next year, I hope to achieve (in increasing order of difficulty):

Acceptance in Operations Research.
Resubmission to one or two Operations Research/Management Science journals or conferences.
Cross the 200 citations mark.
Collaborative papers with PhD students receiving awards.
Secure a research grant.
Receive a faculty job offer.
Complete and submit all ongoing projects (almost impossible).

A Fresh Understanding of Social Relationships:

Presenting oneself and creating a persona within the academic community is essential for effective promotion.

Maintain positive relationships with peers, and they will naturally promote you.