Debasmita Lohar
About News Projects Publications Service Teaching Honors Theses
About News Projects Publications Service Teaching Honors Theses

Assistant Professor · IT University of Copenhagen

Debasmita Lohar

"The true measure of success is how many times you can bounce back from failure." — Stephen Richards

Email: debloh [at] itu [dot] dk
Address: SQUARE Group, Software Engineering Section
IT University of Copenhagen
Rued Langgaards Vej 7, 2300 Copenhagen
Debasmita Lohar

About Me

Hi! I am an Assistant Professor in the Software Engineering section at the IT University of Copenhagen.

Before this, I was a Postdoctoral Researcher at the Karlsruhe Institute of Technology, working with Bernhard Beckert. I obtained my PhD from Saarland University in collaboration with the Max Planck Institute for Software Systems (MPI-SWS), Saarbrücken, where I was advised by Eva Darulova.

Research Interests

Program Analysis Floating-Point Arithmetic Mixed-Precision Tuning Neural Network Verification Probabilistic Analysis Approximate Computing Embedded Systems Formal Methods

News

Service2027

I will be chairing ETAPS Workshops 2027 (together with Raúl Pardo).

TeachingFall 2026

In Fall 2026, I will be teaching Analysis, Design and Software Architecture.

Projects

Infinite-Safety

2025 · 2026

Guaranteeing safe neural control under finite precision and budgeted perturbations.

Paper 1 Paper 2 (under review)

Aster

2023 · 2024

Sound mixed fixed-point quantization of neural networks. Produces quantized networks with soundness guarantees on the numerical error.

Paper 1 Paper 2 Code

Blossom

2021

A guided blackbox fuzzer to infer numerical kernel ranges.

Paper Code

PrAn

2018 · 2019

Sound probabilistic numerical error analysis. Computes probabilistic bounds on floating-point errors for programs with uncertain inputs.

Paper 1 Paper 2 Code

ProPFA

2015 · 2016 · 2017

Probabilistic failure analysis and reliability estimation of control software and component-based systems using formal methods.

Paper 1 Paper 2 Paper 3

Publications

Bookkeeping for Bad Angels: Safe Neural Control under Budgeted Updates and Perturbations

Samuel Teuber, Debasmita Lohar, and Bernhard Beckert (Under Review), 2026.

Neural NetworkSafetyFixed-PointMixed-Precision

Service

  • 2027: ETAPS Workshops Co-Chair (with Raúl Pardo)
  • 2026: SOAP @ PLDI Co-PC Chair (with Yue Li)
  • 2026: ACM REP'26 PC, Science of Computer Programming Reviewer, SOSYM Reviewer
  • 2025: EMSOFT'25 AEC, PLDI'25 AEC
  • 2024: SETTA'24 PC
  • Older: CAV'23 AEC, TACAS'22 AEC, CAV'21 AEC, TACAS'21 AEC, EMSOFT'19 Work-in-progress PC

Teaching

IT University of Copenhagen, Denmark

Karlsruhe Institute of Technology, Germany

  • Advanced Program Analysis (Summer 2019), Saarland University, Germany
  • Program Analysis (Winter 2018/2019), Saarland University, Germany
  • Fault Tolerant Systems (Spring 2016, 2015, 2014), Indian Institute of Technology Kharagpur, India
  • Theory of Computation (Fall 2015), Indian Institute of Technology Kharagpur, India
  • Computer Organization and Architecture Lab (Fall 2014), Indian Institute of Technology Kharagpur, India

Honors and Awards

  • Selected to participate in the 10th Heidelberg Laureate Forum, 2023.
  • Best Presentation Award (iFM PhD Symposium), 2019.

Theses