Starting today, 10,000 scientists across every field, from math to chemistry to physics and more, can get Claude through our new Claude Team plan for scientists. Standard seats are free, and premium seats with 5x usage limits are $15 per month, an 80% discount, for one year.
Claude is becoming increasingly capable of scientific work, with recent progress on problems from advanced physics calculations to protein design. Alongside that progress, we've been investing in the research community: Claude Science launched in June, and our AI for Science program funds high-impact projects with free credits. Today's expansion builds on both.
Principal investigators (or equivalent) at academic and nonprofit research institutions can sign up, then add the researchers in their group. Over the coming months, we plan to extend the program well beyond the initial 10,000 seats.
Learn more: https://t.co/RTG1JxWi4Q
The secret lecture from Gilbert Strang that MIT almost never released.
The harder one. The one that separates the engineers earning $180K from the ones earning $500K.
Senior simulation engineers at SpaceX, Boeing, and NVIDIA who can debug numerical instability make $400K to $600K a year.
The ones who cannot hand the problem to someone who can.
This lecture is the difference.
His name is Gilbert Strang. MIT, fifty years of teaching. Textbooks in 300 universities. The most watched mathematics professor on OpenCourseWare.
This is lecture 1 of 18.086. The course engineers actually need when the math stops being clean.
He opens with a question every simulation engineer faces and almost none of them can answer cleanly: when does a numerical method blow up?
Not approximately. Exactly. And why.
Two methods dominate. Forward Euler and backward Euler. One is fast. One is stable. You cannot have both.
At 16:35 he explains stiff equations — the class of problems that destroys explicit methods. Two decay rates in the same system. One slow. One a hundred times faster. The fast one forces you to take tiny time steps even when the slow one is what actually matters.
Every chemical simulation. Every financial model with multiple timescales. Every control system. Stiff problems everywhere.
At 31:41 he builds a matrix with eigenvalues minus 1 and minus 99. Then shows exactly what happens to forward Euler — why the fast eigenvalue kills your time step even when it contributes nothing to the answer you care about.
At 35:08 he derives the stability limit in one line. Go past it by a single percent and the solution explodes.
Then backward Euler. One change. The growth factor is always below 1, for any time step, for any negative eigenvalue. Absolutely stable. No limit.
ODE45 in MATLAB — running inside every simulation at Boeing, SpaceX, and every pharmaceutical lab computing drug kinetics — is built on exactly this tradeoff.
A simulation engineer I know says this lecture is what got him from $160K to $480K in three years. Said he was the only person on his team who could explain why the solver was diverging.
The lecture was recorded in 2006 and barely noticed. The engineers who found it anyway are the ones debugging simulations everyone else gave up on.
The Fourier transform runs inside every MRI machine, every audio compressor, every signal processor on earth. JPMorgan pays $350K to engineers who can derive it from scratch. There is one professor alive who explains it the way nobody else can.
His name is Gilbert Strang. MIT. The most brilliant mathematical mind of his generation. His textbook sits in over 10 million homes. No other person on earth holds this combination of depth and clarity in one head.
He opens with one confession.
The Fourier transform is unreasonably effective. It solves problems it was never designed for. The goal is not to compute it. The goal is to understand why it works at all.
Then the core idea. A periodic signal breaks into pure sine waves. Each with a frequency, amplitude, and phase. The transform finds all three simultaneously. A complicated signal in time becomes a simple picture in frequency.
Then 1807. Fourier invented the transform not for sound but for heat. How does warmth spread through a metal rod? In time the equation is a partial differential equation - hard. In frequency space it becomes ordinary - easy. Transform in, solve it, transform back.
Then convolution. Convolution in time equals multiplication in frequency. Filtering a signal, removing noise, compressing audio - all multiplication in frequency space. Without the transform: thousands of computations. With it: a few multiplications.
Then the delta function. Zero everywhere except one point where it is infinite. Integral equals one. Every mathematician in 1900 said it was not a function. Dirac used it anyway. It took 40 years to justify what engineers had been doing the whole time.
Watch the moment Strang shows that the Fourier transform of a delta function is a constant - every frequency in equal measure. The more concentrated in time, the more spread in frequency. This is the uncertainty principle. Not quantum physics. A theorem about any signal at all.
A signal processing engineer I know rewatched this before their first project at Apple. Said it was the first time the Fourier transform felt like a change of coordinates rather than a formula to memorize.
Free on YouTube, MIT OpenCourseWare.
bookmark this and watch later - after this lecture every sound, every image, and every signal will feel like a sum of sine waves waiting to be separated
New in Claude Code: your sessions can now message each other.
Instead of having to re-explain yourself in another session, you can now tell Claude to do it. It sends a summary (not your history or files), and the other session picks it up mid-task.
Lost my phone at the office and spent 30 minutes turning the place over. Find My was disabled by MDM.
Out of ideas, I asked Claude how I could find it. It suggested tracking the Bluetooth signal strength, then wrote me a meter in about a minute.
I walked around watching the number climb. Found it.
Apparently you can just make the tool you need now.
Code: https://t.co/fmnISzHfZ2
6 months of Claude Max 20x, on us.
We're expanding Claude for Open Source to more of the community.
If you're a maintainer, a core contributor, someone landing PRs across the ecosystem, or someone keeping a critical package alive, apply today!
@LyndaShacks@AlettaHarrison@JHBWater@LyndaShacks - its North Road, Rivonia Road (Madison Square) and most of surrounds - we on corner of North & Rivonia - no water since yesterday mid day.
@JHBWater@JohnSmi42289760 Many thanks - just a bit worried that there has been no feedback since early yesterday afternoon; would have thought there would at least be something now.
Valve has officially released the files needed to build the Steam Machine’s e-ink front display, called the “Inkterface.”
The DIY kit includes:
- 3D printable files
- A complete bill of materials
- Step-by-step assembly instructions
- An official build video
The 5.83-inch e-ink screen connects over Bluetooth and can display system information like performance stats on the front of the Steam Machine.
If you don’t want to build it yourself, JSAUX is also working on a pre-built Inkterface.
It’s still listed as “coming in 2026,” but no release date or price has been announced yet.