Hanbin Lee

Just took a bus around the North campus and even the dorms are sprawled in this area. Apartments are all maximum 2.5 floors high and more than half of the land is parking lots. Damn cursed.

I feel very stupid after searching "next letter of omega" on google. I badly needed something for an n-form :(

My experience in the US is that the lack of physical connectivity in many parts of the country, of which have been deliberately created over decades, considerably hampers the dissemination of modern policies. I wonder how the numbers look like after conditioning on rural vs suburb vs urban.

Another thing that bothers me is ergodicity. Distribution over one dimension (e.g. time) coincides with the distribution over another dimension (e.g. replicates) if certain conditions are met. Is this obscuring us from recognizing something important?

Am I a MAGA fan now?

And, do we even care about IQ? It is like the Impact Factor of intelligence...

www.biorxiv.org/content/10.6...
What surprised me is Figure 6 in which the distribution of the points looks very similar to what I've encountered in a very different setting. I wonder if this is some sort of a very general statistical phenomenon.

end of class! no more daily homeworks :)

The academics who stuck by financier to the end – and those who didn’t | The Observer

Don't forget all these repulsive men who helped boost Jeffrey Epstein's reputation even after his horrific deeds were well known.

observer.co.uk/news/interna...

People like Martin Nowak, Steven Pinker, Alan Dershowitz, Larry Summers, Lawrence Krauss, et al

I wonder how administration and representation can be separated looking at gerrymendered maps.

On the Approximation of Phylogenetic Distance Functions by Artificial Neural Networks Inferring the phylogenetic relationships among a sample of organisms is a fundamental problem in modern biology. While distance-based hierarchical clustering algorithms achieved early success on this ...

New paper on minimal NNs for phylogenetic inference, with former grad student, Ben Rosenzweig:

arxiv.org/abs/2512.02223

I remember you asking about the purpose of omitting a information (here, location of materialized mutations). My tentative answer is that the we get a simpler model (no confounding) at a cost of perhaps lower precision (e.g prediction accuracy of BLUP) due to omitting information.

That was part of my convoluted thoughts. All, if not most, the papers making the claim that "GRM is enough" is marginalizing possible evolutionary paths, removing those non-direction flunctuations away from the model. This is obvious in our paper because mutations are treated as random.

It certainly depends on the model assumption. I have a few convoluted arguments in my mind, perhaps the most intuitive one will be that such a directional LD can't be there in the absence of selection.

The overall gaussian-ish assumption allows us to discard many of the higher-order terms and makes it clearer how these terms could be handled in more general cases.

I think I finally follow the parent-offspring regression shit. This sounds like conditioning on a pedigree, but it's in fact not because parents are being randomly selected. Hence, it's equivalent to forming an infinite gamete pool first and then forming a new individual out of it.

An interesting rabbit hole that I have yet dug into is that population structure alone cannot drive genetic variance inflation in the estimates. We did a spatial simulation exhibiting a very strong isolation-by-distance pattern. The estimated variance components are consistent.

Observational epidemiological studies can mitigate genetic confounding with the genetic relatedness matrix Observational studies are commonly used in psychology and epidemiology to identify risk factors correlated with health outcomes. However, these studies are vulnerable to confounding when shared geneti...

Excited to share work from my postdoc with @docedge.bsky.social and collaborators Matt Pennell and @jgschraiber.bsky.social, newly out over the weekend: www.biorxiv.org/content/10.1... (1/6)

Depictions of evolution where a phylogeny often has humans on the far right or top can give an impression of evolution being progressive of leading to ‘increased complexity’ when it does not. Top figure shows such a phylogeny which can look the same as ‘the March of progress’ depiction most commonly used to depict evolution (showing monkey to man erroneous march of evolution) - instead swiveling some nodes on a phylogeny where humans are shown closer to the center (which doesn’t change relationships) can lead to better ‘tree thinking’

New post by me on #MITPressReader @mitpress.bsky.social

On the 100th anniversary of the #ScopesMonkeyTrial
the ways we depict #evolution can still give an erroneous progressive view (that evolution leads to humans or ‘increased complexity’).

thereader.mitpress.mit.edu/is-our-pictu...

I now automatically think of gerrymendering whenever I see decision boundaries.

Single-cell phylodynamics reveal rapid late-stage colorectal cancer expansions Single-cell whole-genome sequencing of 335 cells from seven colorectal cancers, coupled with Bayesian phylodynamic modeling, revealed tumors often originate decades before diagnosis, remain indolent, ...

www.biorxiv.org/content/10.1... hm?

빈도주의적 통계로써 신뢰구간은, 내가 하는 방법론의 신뢰성을 드러내는거지 내가 한 번 만들어낸 결과값의 신뢰성을 드러내는게 아니다. 근데 아마 베이지언적으로 확신구간으로 연산해도 사실상 비슷한 범위가 나오긴할꺼다. 무슨 전국여조서 전국 1%p 변동이나 개별지역, 성연령 십프로 안 변동을 갖고 썰 푸는게 일상인 보도 판에, 표본편향 고민하며 접촉률 협조율에 갤럽만큼의 애쓰고 신중한 표현하는 것만으로도 감지덕지하다. 업력이 괜히 나오는게 아니다.

We tried to make the paper more focused on variance component estimation and genetic prediction. Other points raised in the earlier paper will be discussed in future works.

Genetic prediction with ARG-powered linear algebra Ancestral recombination graphs (ARGs) are an attractive means for quantitative genetic analysis of complex traits because they encode the realized genetic relatedness between a sample of individuals i...

www.biorxiv.org/content/10.1...

We finally submitted the earlier preprint to a journal after massive restructuring.
We've expanded the REML section for those interested in the method. We clarify that ARG-LMM estimates mutational variance and not additive variance.

Transit should be at forefront of any climate resiliency plan. Take snow. A bus is heavy and piloted by a trained driver; much safer in snow (or heavy rainfall) than the relatively light vehicles driven by civilians. Bus riders take vehicles off the streets, which speeds plowing, emergency response.

techbros everywhere lol

Curriculum shapes more than skills as taking STEM in high school boosts tech careers—but also shifts politics as boys grow more conservative and girls more progressive, from Robert Ainsworth, Rajeev H. Dehejia, Andrei Munteanu, Cristian Pop-Eleches, and Miguel Urquiola www.nber.org/papers/w34502

www.biorxiv.org/content/10.1... looks useful?

pubmed.ncbi.nlm.nih.gov/41001573/

I got something to tell to my thesis committee member