Announcement
Our next chapter: Building the VietAlpha lab
July 19, 2026 · Harry Tran
Vietnam's oldest known mathematics textbook survives only in copies made long after its author died, and scholars still argue about whether he wrote it at all.
This text, the Toán pháp đại thành, is a treatise on practical arithmetic attributed to Lương Thế Vinh. Folklore remembers Lương Thế Vinh as Trạng Lường, the measuring laureate. It credits him with weighing an elephant by floating it on a boat. His book focuses on the mathematics of ordinary life, such as how to measure a field or calculate a tax. Although generations of civil service candidates studied versions of the text, only later copies survive in Hanoi's archives. This gap means that the few scholarly studies of these copies (notably the work of Alexei Volkov) view the authorship as an open question.
This uncertainty reflects the state of Vietnam's entire mathematical history, where only a few dozen Hán and Nôm treatises survive. Most are late printings or undated manuscript copies. Because almost none have been digitized, they remain unread. They are the silent history of how mathematics was done and taught in Vietnam.
We believe this history belongs to the students who inherit it. Sharing it requires a new approach. Our transition from an exam-prep library to a research lab begins with our origin as a tutoring tool. From there, we are organizing our work into three research areas and planning our next steps.
Where we started
VietAlpha began with the cost of math tutoring in Vietnam. Premium exam preparation can reach 7.2 million VND a month per subject, while the average monthly income of a rural household is 4.5 million VND. To build a free alternative, we wrote Cadmus. This parsing pipeline converts scanned Vietnamese math papers into structured, verified data. Using this dataset, we launched LumiToán. It is a free library and Socratic tutor that now serves more than 10,000 students monthly. This scale helped us win first prize in the Edtech category of the ICIA National Youth Innovation Award.
Our work on LumiToán showed us that models themselves are not the hard part. Compute is rentable. Algorithms improve every quarter. The real barrier is the data: Vietnam's mathematical materials in a digital form clean enough to build on.
The work we do now
To rebuild this mathematical foundation, VietAlpha is a research lab of CS Inspiration, conducted in association with Deerfield Academy. We want to connect Vietnam's mathematical tradition with modern digital tools. We digitize the surviving texts and analyze their mathematical methods. We then transfer these historical systems into modern technologies, which we use to build classroom tools.
Digitization comes first because all other research depends on it. The history we cover spans five centuries, from the era of Lương Thế Vinh's Toán pháp đại thành to the current MoET curriculum. Digitizing these sources starts where scanning ends: we convert each page into structured, verified data for research and training, and publish these as open datasets. While our parser, Cadmus, handles modern papers, adapting this pipeline for historical manuscripts is our main engineering challenge.
In our second area, we study and dissect these historical mathematical systems. The surviving texts deal with abacus arithmetic, land measurement, tax computation, and currency conversion: the practical mathematics of running a country. By analyzing these documents, we study how these methods functioned and trace how Vietnam adapted ideas from neighboring cultures. We then compare nineteenth-century calculations directly to modern school lessons. We now focus on the history of science and digital preservation. This work extends our reach beyond our original tutoring tools. An empty field is a strange place for a student lab, but a small team can still build the foundation here.
The third part carries this work back into the present, where we transfer historical mathematical systems into modern technology. LumiToán is our first application. We use digitized texts and historical algorithms to train our models, and student corrections feed directly back into the archive. Because every future tool will draw on this same archive, this repository separates our lab from products that wrap external models.
What we keep, and where we go next
Even as the lab expands, some things do not change. Our work follows the MoET curriculum, because that is what Vietnamese students study. LumiToán is free, because the cost of tutoring started this project and the student who cannot pay is the person we build for. VietAlpha is still part of the CS Inspiration ecosystem, alongside the programs it grew up with.
What opens up is the historical data. We are selecting our first project now, and we will announce it once the work begins. Since everything we have shipped so far covers the modern curriculum, placing the era of the Toán pháp đại thành inside our scope is an ambition.
Over the coming months, we will update the VietAlpha site to match this direction and continue publishing our research notes. This work is a response to the fragility of these materials: five hundred years ago, someone wrote down how to measure a field, and that book survived only because a copyist decided it was worth the ink. Vietnam's mathematical history should not have to survive by luck. Bringing this history into a form that lasts is now our job, and we feel lucky to begin it.