Amirhossein Samandar

Amirhossein Samandar

Cosmology · statistical inference · probabilistic machine learning · computational methods

Sitemap

A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.

Pages

Amirhossein Samandar

Statistical inference for cosmological data — likelihood-ratio methods and their neural counterparts.

Posts

Future Blog Post

less than 1 minute read

Published:

This post will show up by default. To disable scheduling of future posts, edit config.yml and set future: false.

Blog Post number 4

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 3

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 2

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 1

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

portfolio

publications

Cosmic topology. Part IVa. Classification of manifolds using machine learning: a case study with small toroidal universes

Published in Journal of Cosmology and Astroparticle Physics 09 (2024) 057, 2024

Classifiers trained on simulated CMB maps recover near-likelihood-level discrimination between manifolds at a fraction of the cost. A rare setting in which the ground-truth likelihood is computable, so the learned model’s discrepancy from the optimal decision rule can be measured rather than assumed.

Download Paper

Don’t Pass@k: A Bayesian Framework for Large Language Model Evaluation

Published in International Conference on Learning Representations (ICLR) 2026, 2026

Replaces Pass@k and avg@N with posterior estimates and credible intervals on a model’s underlying success probability. Outcomes are modelled as categorical with a Dirichlet prior, giving closed-form posterior mean and uncertainty for any weighted rubric. Achieves faster convergence and greater rank stability than Pass@k at far smaller sample counts.

Download Paper

Test-Time Scaling in Reasoning LLMs: Inference Regimes, Evaluation, and Reproducibility

Published in arXiv preprint, 2026

Formalizes test-time scaling as budgeted inference over the implicit prefix tree of an autoregressive transformer, distinguishing sequential, leaf-level and prefix-level scaling. Treats the entire inference system as the evaluated object and specifies reproducibility requirements separating exact replay from distributional reproducibility. Over 2 billion reasoning traces released.

Download Paper

The Topology of the Universe

Published in Nature Astronomy, 2026

Invited review of the observational status of cosmic topology and the prospects for CMB (LiteBIRD, Taurus) and large-scale-structure searches.

Download Paper

Beyond the Kullback–Leibler divergence: the likelihood-ratio distribution as a detectability measure for cosmic topology

Published in In preparation, 2026

The KL divergence is an expected information gain and cannot, by itself, give a detection probability on a single realization. This work derives the exact sampling distribution of the log-likelihood ratio between two zero-mean Gaussian models — a generalized non-central chi-squared — and uses it to convert expected information gain into a detection probability with ROC curves and calibrated interval coverage. Supplies the frequentist validation layer that learned ratio estimators currently lack, and connects to Bayesian experimental design. Model-agnostic; cosmic topology is the case study.

talks

LLM Evaluation and Test-Time Scaling

Published:

Invited seminar on calibrated Bayesian evaluation of large language models and the formalization of test-time scaling as budgeted inference.

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.