T.Brada Experimental Grid

T.Brada Experimental Grid (abbreviated TBEG) was a volunteer computing project built on the BOINC (Berkeley Open Infrastructure for Network Computing) platform. Founded in February 2019 by Slovak developer Tomáš Brada, the project served as both a personal BOINC development testbed and a home for several mathematical sub-projects in number theory and combinatorics. The project ran until December 2022, when its final administrator announced it would no longer distribute compute work.

T.Brada Experimental Grid
Project
Status	Completed
Category	Mathematics
Compute	CPU
Requires	None
Development
Developer	Tomáš Brada
Author	Tomáš Brada
Initial release	February 3, 2019  (7 years ago)
Discontinued	December 24, 2022  (4 years ago)
Software
Operating system	Windows, Linux (64-bit)
Metadata
Website	https://boinc.tbrada.eu/

File:Boinc logo.png

The BOINC platform logo. T.Brada Experimental Grid ran on BOINC server software.

Background and founding

T.Brada Experimental Grid was created by Tomáš Brada, a software developer from Slovakia, who described the project's primary purpose plainly on its home page: "This project serves as a development system for the project administrator." Rather than being organized around a single narrow research goal, the project was an experimental platform where Brada could develop and test new BOINC application technologies while simultaneously offering useful mathematical computation to the volunteer computing community.

The project launched under the name Pseudo Associative DLS before being renamed T.Brada Experimental Grid. By September 2019 the project's name had been officially updated in the BOINCstats database at the administrator's request.

The Scottish BOINC Team noted in January 2020 that T.Brada Experimental Grid had "been active since late spring 2019" and that it had become a marathon project in Formula BOINC 2020, though they described it candidly as "definitely an alpha project."

Sub-projects

At its most active, T.Brada Experimental Grid ran three distinct sub-projects simultaneously, each tackling a different area of mathematics or software development.

Symmetric Prime Tuples (SPT)

The most prominent sub-project was the search for Symmetric Prime Tuples, launched on 18 October 2019. This sub-project was a direct continuation of the defunct Stop@home project, which had been searching for symmetric prime k-tuples of consecutive primes in the range from 0 up to $2^{64}$.

A symmetric prime tuple of length $k$ with diameter $d$ is a prime tuple with a pattern $H=(h_1,h_2,\dots ,h_k)$ satisfying the condition:

$h_i+h_{k+1-i}=d\text{for all }i$

This means the prime elements of the tuple are distributed symmetrically about a common center. The diameter $d$ of a k-tuple is the difference between its largest and smallest elements.

The Hardy-Littlewood constant for a pattern ${\mathscr{H}}$ of length $k$ is:

$\mathfrak{𝔖}({\mathscr{H}})=\prod _p(1-\frac{{\nu }_p({\mathscr{H}})}{p}){(1-\frac{1}{p})}^{-k}$

where ${\nu }_p({\mathscr{H}})$ is the number of distinct residues modulo $p$ covered by the pattern. The symmetry of the pattern imposes strict constraints on these residues: for $p=2$, it is always the case that ${\nu }_2({\mathscr{H}})=1$, since all elements of the tuple are odd.

The SPT application extended the original Stop@home search in several important ways. In addition to searching for symmetric prime tuples, it also searched for twin prime tuples, symmetric twin prime tuples, and prime tuple gaps, and the search interval was extended beyond what Stop@home had covered. An early result reported by Stop@home was the discovery of the minimal 17-tuple starting at 159,067,808,851,610,411 with offsets {0, 42, 60, 96, 102, 186, 210, 240, 246, 252, 282, 306, 390, 396, 432, 450, 492}.

Once the SPT application had covered the full range previously handled by Stop@home while also scanning for the new tuple types up to the same range, Brada issued four new collector badges to thank contributors. He also announced plans to submit results to the On-Line Encyclopedia of Integer Sequences (OEIS). Later batches pushed the search interval up to $5\times 10^{17}$ and beyond.

The source code for the SPT application was developed by Tomáš Brada and is publicly available on GitHub.

PADLS Total

The second sub-project, PADLS Total, aimed to find new pseudo-associative diagonal Latin squares (PADLS), a structure described by mathematician Natalia Makarova. This work placed T.Brada Experimental Grid alongside related projects such as ODLK and ODLK1, which searched for orthogonal diagonal Latin squares of order 10.

A diagonal Latin square is an $n\times n$ Latin square in which both the main diagonal and the back diagonal are also transversals (containing each symbol exactly once). A pair of diagonal Latin squares is orthogonal (ODLS) if, when superimposed, each ordered pair of symbols appears exactly once.

Over the first three months of the PADLS Total rule 51 experiment, more than 700,000 unique canonical forms of orthogonal diagonal Latin squares were found, including five rare groups of ODLS pairs in groups of three and four. The PADLS Total sub-project remained running until the end of the project, though Brada throttled its workunit output in later years to conserve power while he focused on other areas.

Lua Alfa

The third sub-project, Lua Alfa, was purely an experimental software development effort. Its application was used to test an in-development Lua runtime environment for BOINC. The goal was to expand what is possible under the BOINC platform and improve ease of deployment by enabling script-based applications rather than traditionally compiled binaries. The application was limited to Linux, and tasks took only a few minutes at most.

File:Latin-square-1.png

An example of a diagonal Latin square. T.Brada Experimental Grid's PADLS Total sub-project searched for pseudo-associative variants of such structures.

Infrastructure and history

Database crashes

The project experienced significant technical difficulties during its operation. Around 20 February 2020, a database crash occurred. Fortunately no data was lost, as all data had been backed up. Brada posted a news item titled "Database Crash and Recovery" to reassure volunteers, and the project came back online shortly afterward.

An expired SSL certificate in early 2020 (valid only from 5 January 2020 to 5 April 2020) caused connection problems for volunteers and led some in the community to assume the project had been abandoned.

COVID-19 response

On 16 March 2020, in response to the global COVID-19 pandemic, Brada disabled distribution of new PADLS Total workunits with a note to volunteers: "Because of the global situation, I am disabling distribution of new PADLS tot5 workunits in an attempt to divert the compute attention to Folding@Home."

Closure

On 24 December 2022, an administrator signing as Veronika posted a farewell message to the project's community:

Dear Crunchers of this project. Let me wish you merry christmas and happy winter solstice. It weights my hearth to write this message, but things had come to a point where it is better to end things, rather than endlessly drag them out to no end. Since the summer, I have been hopelessly stomped by a stream of activity, both happy and necessary. It turned out to be impossible to care about this project... It is time to announce that this project will no longer provide any compute jobs. I will, of course, keep the site running, but since there is no work, no new users can get enough points to post, the forums will effectively be dead.

The BOINCstats thread for the closure was reported in December 2022.

Legacy and continuation

Following the project's closure, the volunteer community showed continued interest in preserving the SPT work. In June 2023, a call was posted on the T.Brada Experimental Grid front page from colleagues attempting to restore the Symmetric Prime Tuples project on a new BOINC server, using Brada's publicly available source code. As of the post, the code had not yet been successfully recompiled.

The broader work on orthogonal diagonal Latin squares in the PADLS Total sub-project was carried forward by the continuing ODLK and ODLK1 projects, and the newer ODLK2025 project launched in December 2024. The SETI Germany wiki noted that T.Brada Experimental Grid was essentially the continuation of Stop@home in the area of prime tuple research.

Software and applications

The project supported 64-bit Windows and Linux platforms. The three application types were:

• PADLS Total 5.10 for Windows and Linux (64-bit)
• Symmetric Prime Tuples 3.00 for Windows and Linux (64-bit)
• Lua Alfa (beta) 0.04 for Linux only

The project did not use GPU computing. Workunits for the SPT and PADLS sub-projects ran on CPU only, and tasks were designed to be short-running, especially for the Lua sub-project where completion times were described as taking "a few minutes at most."

See also

• BOINC
• ODLK
• ODLK1
• ODLK2025
• PrimeGrid
• SRBase
• Gerasim@home
• RakeSearch
• BOINC projects

External links

• https://boinc.tbrada.eu/ -- Official project website (no longer distributing work)
• GitHub: tbboinc source code by Tomáš Brada
• BC-Wiki entry (English)
• BOINCstats project news archive
• SETI Germany Wiki entry