thinking about postdoc applications depresses me. have couple of years until i finish phd but still. any advice?
thinking about postdoc applications depresses me. have couple of years until i finish phd but still. any advice?
it feels to me that as openAI launches another new product/agent, the gpt becomes less intelligent
Now in JHEP: Supersymmetric Horizon Stabilization (SHS) β a way to realize large-field inflation in string-inspired SUGRA, consistent with swampland/UV bounds.
βοΈ
t.co/gHpWlXIryE
Now in JHEP: Supersymmetric Horizon Stabilization (SHS) β a way to realize large-field inflation in string-inspired SUGRA, consistent with swampland/UV bounds.
βοΈ
t.co/gHpWlXIryE
waiting for an orange foldable iPhone to upgrade my base 15.
skydive egitiminin 140 bin lira oldugunu ogrenince kiz arkadasimin yuzundeki mutluluk ve rahatlama inanilmazdi
tek cumlelik mail bile atamamak ama hicbir sey yapmadan bile mukemmel calisan uygulamaya laf soylemeye calismak. junkman ve begone sayesinde imessage kullanabilir hale geldim ve sacma sapan dolandirici aramalari tamamen bitti. bu vesileyle gec kalmis 5 yildiz appstore geri bildirimimi yapayim bari π
i got gpt5 on my ios app but neither on my mac app or the browser has it. why is that
started reading manga and watching anime again. feels nice. any recommendations are welcome
it has been a while since i posted but here is our new preprint
arxiv.org/abs/2506.18107
writing code for scientific purposes gives me anxiety because how ugly the equations are had to be typed into code. zero aesthetic.
adjust model params.
check models obs. results
sftp model to cluster
run model
sftp back results
check power spectrum and mass of pbh
repeat
it was working before and the code has not changed.
why would mpi fails to write to a file at rank 0 for a specific calculation even though the calculation is done w/o failing. i checked the gathering and all ranks gather w/o a fail, i have enough disk space, enough ram for each core etc. every calculation is done properly but still fails at rank 0.
it takes two was amazing. this is not surprising to me.
photo showing an ipad screen where it has the mentioned apps opened on a multitask screen
* openVPN to connect cluster vpn
* termius to ssh into cluster and sftp the result files
* google collab to use plotter notebook
all on ipad.
parallelization is cool and i like it now. mpi4py ftw π―
for more context, I gave it my code and wanted it to optimize it as much as possible and it instead checked a code that does similar calculation but not exactly the same on github and optimized that code and disregarded my own code. I think I am using this wrong.
chatgpt's deep research got confused while thinking on a prompt about cosmological calculation and randomly thought about foster and non-foster parents in Kazakhstan and examined factors like gender, age, eduation etc. for a short while.
uhmm what? sure.
hell week is over and it all turned out for the better
sometimes it feels like I live in the Darkest Timeline of the Bad Place and sometimes the Neutral Place. I want to experience also the Good Place a little bit please. Do I need a Good Janet and a Michael? Am I Chidi Anagonye? What is happening?
yesterday was the worst day of 2025 so far for me and my gf and there is a chance (hopefully not) that it can get much worse today but seeing this made me smile a bit
would appreciate a selection of community designed app icons for @bsky.app
just got time to look at it and looks clean. i will try for the next couple of days. thank you for the advice.
i love seeing photos of scientists working while traveling on train. there is a strange appeal of the theme and would love to do in the future
what is your daily usage of AI? and why do you use it for?
chatGPT, Claude, Perplexity, DeepSeek, grok etcβ¦
you know sometimes your hair is so bad that even face id has a hard time recognizing you, itβs that kind of a bad hair day today and i donβt even have that much hair
color options in matplotlib not giving the default colors is insane to me. you either not give any color option or know the hex of defaults or use the C0, C1β¦ etc notation to get the default ones.