August 15, 2026
Welcome to the third and last episode of my Retreat in Lozère series. Today I want to write about the experience of preparing a research grant.
I am writing this from a chalet in Lozère. This place has a special place in my heart. I spent two weeks here almost every summer from age zero to age 20. I came back less regularly after that, but I still come whenever I can.
This chalet is where I prepared some of my math exams during prep school. Between my two years of classes préparatoires, I spent one month here working hard and running on the trails. I know the forests and paths around here very well. Within a radius of 15 kilometers, almost every turn feels familiar.
This is also where I read Saint Augustine as a teenager and felt something close to faith for the first time. It was also the last time, but it still mattered to me. Some places keep the memory of who we used to be.
The chalet has now passed from my grandparents to my uncle and aunt. I am very grateful that they let me stay here. For ten days, I could work, eat, sleep, run, and think. This is a luxury.
In France in particular, and I'm sure elsewhere but I will not over-generalize, the idea of writing grants is very political. Researchers are very attached to permanent positions and long-term research with no immediate pressure. Who wouldn't? I understand this and I like this model—it gives people time to think and protects research from short-term fashion.
However, there is another side: there is less and less money, and funding is getting much harder. Us, young researchers cannot simply pretend this is not happening; we need PhD students and postdocs! We need resources if we want to build something larger than a few papers.
I sometimes feel that permanent researchers who reject grants on principle are defending a system that no longer really exists. (Yes, I know, another provocative post. This is my style!) I try and say this with respect, but also with great frustration. They lived in a different system with a very different funding landscape.
The discussion is difficult because it often becomes political. On the left, grants are seen as a problem, while on the right, they are seen as a solution. Reality is more complicated.
I felt the same thing when I wrote about AI: the discussion immediately became political. Some people did not want to discuss what AI can or cannot do. They only wanted to say that it eats bad text, digests bad text, produces bad text, and then eats its own bad text again. (I'm writing this in a much more polite way than what was actually said!) Other people immediately talked about AI water and electricity consumption. These are very important points but they absolutely do not remove the scientific question.
For grants, I feel the same. Yes, we can criticize the system, and we should! But we also have to live in it.
My first reason is that I'm a driven person and I need goals. After I got my permanent position at Inria, I sent an email to senior people in my lab. I asked them for advice about what to do next. Some of them basically answered: “You have tenure now: enjoy it! Why are you already thinking about the future?” Well, yes, this is me, hello!
Anyway, so I got a permanent position three years ago. Since then, I have completed around 70 or 80% of the research program I had in mind at the time; so, yes, I needed a new goal.
Writing a grant is a difficult exercise, but it is also a wonderful one. It forces you to read the current literature. And with AI, scientific machine learning, and numerical analysis moving so fast, this is already a huge task. It also forces you to step back and ask a hard question: what do I really want to do next?
It is easy to stay in a niche! As I wrote in a previous post, this is one way to do research: you take a known problem and generalize it. You write a few papers, you slowly improve the numerical method, you prove sharper convergence theorems, you apply it to harder and harder application problems. This is useful but I do not want to do only that.
So another reason is leaving the niche to properly think about where applied mathematics is going. In particular, I want to understand how AI can be used in numerical analysis and scientific computing without losing mathematical rigor. What I really want are mathematical criteria that tell us when these methods work. I will not say much more here, because this is the grant proposal and the core of the project.
There is also another reason. I like writing! It was passed on to me by my mother, who was a pharmacist but also loved writing and wrote a couple of novels in her free time. It was also passed on to me by my PhD advisor, whose writing skills always impressed and inspired me. So now, I really like putting documents together. I like taking messy ideas and turning them into something clear. And editing a document until it's polished. I do this in my research memos. I do this in my papers. I do this in my lecture notes. I do this on this blog! A grant proposal is another version of the same exercise.
It remains a hard process. Like trail running for a few hours with some good D+. Yes, it is hard, but it is the kind of hard I like.
This retreat has been so good for me! I spent my days reading, writing, thinking, running, and eating delicious local food. I was hosted and fed by my family, so I could put almost all my mental energy into the proposal. Again, that is a luxury.
When the writing became too much, I ran on the same paths I knew as a child. I recognized the trees and the stones. I felt tired, but I felt safe and at home.
I do not know whether this grant will succeed and, frankly, most grants do not. Nevertheless, the experience has already been useful. It forced me to think seriously about the next years of my research. It forced me to ask how I want numerical analysis to look like in this new AI era. For those reasons, I am glad I did it anyway.
Finally, I am grateful to this place, once again, for giving me the space to think.