Imperial College London

Dr Selim Hotinli

Faculty of Natural SciencesDepartment of Physics

Academic Visitor







Blackett LaboratorySouth Kensington Campus




Kinetic Sunyaev Zel'dovich tomography

In the recent years I have authored research papers which demonstrated that the kinetic Sunyaev Zel'dovich (kSZ) tomography (or radial-velocity reconstruction) (Cayuso et al 2020) will provide the most competitive constraints on primordial non-Gaussianity (Kumar et al 2022) and isocurvature (Hotinli et al 2019, Kumar et al 2022), smoking-gun signatures of inflationary scenarios or unknown components that may have been active during the early epoch of our Universe. In the former study, we have also demonstrated an opportunity to unambiguously probe unknown components in the early Universe by comparing different forms of higher-order statistics of the density fluctuations. These efforts are complementary to the core science goals of upcoming CMB experiments, such as measuring the effective number of relativistic species. My research has shown, for example, that the Figure of Merit (FoM) characterising the statistical information content of the joint analysis of large-scale galaxy clustering and velocity fields (latter reconstructed from measurements of the kSZ effect) with Simons Observatory and DESI data by 2026 can be greater by a factor of over an order-of-magnitude compared to considering DESI galaxy clustering in isolation. After the first year data release of LSST, this difference could reach over two orders of magnitude compared to considering LSST in isolation.

Kinetic Sunyaev Zel'dovich (kSZ) effect is a doppler effect due to CMB photons scattering off on energetic electrons in the large-scale structure that have some non-zero radial velocity. This effect is very significant and will dominate the small-scale observations of the black-body CMB in the upcoming decade. Below, you can see how this effects looks like from the websky simulation. It is   these small-scale distortions in the CMB intensity that we measure.


What is important about this effect is that it traces the bulk velocity fields of electrons — in that when electrons have larger radial velocity, the effect is more apparent, and when they have lower velocity effect is less apparent. As a result we can write estimators that capitalize on this statistical anisotropy to reconstruct the large scale velocity field — and the methods we have been developing are quite promising suggesting we will be able to recover information about the cosmological velocity field on large scales very efficiently this way (see e.g. Cayuso et al 2020). The tomographic reconstruction is demonstrated using a simation below for a narrow redshift bin near z=0.5 with noise and foreground specifications matching S3 survey specifications.



Probing helium reionization with CMB x LSS

show research

In addition to probing large-scale deviations from LCDM, the reconstruction program can contribute significantly to the characterisation of both hydrogen (Hotinli and Johnson 2020) and helium (Hotinli et al 2022) reionization (image below).


Characterising helium reionization has great significance for understanding galaxy formation, quasar activity and cosmology, and may open a new window on big bang nucleosynthesis.  Since photons emitted by the first stars (sourcing the reionization of hydrogen) are not energetic enough to fully ionize helium, helium reionization occurs only after the emergence of a substantial number of quasars. As a result, the history of helium reionization strongly depends on the properties of quasars, such as their luminosity function, accretion mechanisms and other astrophysics, clustering, variability, lifetimes, as well as the general growth and evolution of super-massive black holes. Since essentially all of the helium in the Universe is ultimately doubly ionized, the total change in the ionization fraction is also a measure of the primordial helium abundance—a sensitive probe of big bang nucleosynthesis. Furthermore, probing helium reionization can also improve our understanding of relativistic species through improving the primordial helium fraction measurement and breaking the degeneracy between the number of relativistic species and primordial helium fraction. As the primordial helium abundance depends on the weak interaction rates as well as the neutron lifetime, improving its measurement can also allow further valuable insights into our cosmological history. Observational challenges posed by helium emission line measurements make additional probes extremely valuable.

One of the main research programs I lead focuses on characterisation of both hydrogen (Hotinli and Johnson 2020) and helium reionization (Hotinli et al 2022)  from joint analyses of LSS and CMB data. In particular, characterising helium reionization has great significance for understanding galaxy formation, quasar activity and cosmology.  Since photons emitted by the first stars (sourcing the reionization of hydrogen) are not energetic enough to fully ionize helium, helium reionization occurs only after the emergence of a substantial number of quasars.

As a result, the history of helium reionization strongly depends on the properties of quasars, such as their luminosity function, accretion mechanisms and other astrophysics, clustering, variability, lifetimes, as well as the general growth and evolution of super-massive black holes. Observational challenges posed by helium emission line measurements make additional probes extremely valuable. I anticipate a multitude of studies will follow my work, capitalising on the statistical power of cross-correlations between CMB and different high-redshift LSS probes. These results are extremely promising and will likely re-ignite the interest from the community for helium reionization. My efforts currently focus on advancing scientific programs involving characterisation of helium reionization with joint analyses of CMB and LSS including contributing to building up the science cases of the line-intensity signals, which will be a promising direction within this program, and transform this area into an active, evolving field of research. 


On this image: We now know that the hydrogen and helium fully ionizes in the Universe after the formation of first luminous objects. During this process, the ionization fraction (the fraction of free electrons) goes from zero to, say 1, for all the electrons in the hydrogen atom being ionized (kicked away of their proton). This happens soon after the first stars form. You may notice that this line does not end at one, however, which is because helium also ionizes, loosing its first electron around the same time as hydrogen. However the second electron in helium requires much more energy to ionize which only happens after sufficient numbers of much more luminous objects such as quasars, due to active galactic nuclei form… As a result, the history of helium reionization strongly depends on the properties of quasars, such as their luminosity, accretion mechanisms and other very much unknown astrophysics, clustering, variability, lifetimes, as well as the general growth and evolution of super-massive black holes. In a recent work, I have shown for the first time that the joint analysis of the CMB and upcoming high-redshift galaxy surveys can indeed probe this epoch. (See arXiv:2207.07660)

Other CMB secondaries

show research

In the last years I have also made valuable contributions to identifying various novel CMB signatures including the moving lens effect (Hotinli et al 2018, Hotinli et al 2020, Hotinli et al 2021), the pSZ (Lee et al 2022) and kpSZ (Hotinli et al 2022) effects. My efforts on these lines also explored the full reach of the prospects to probing fundamental physics with these signatures. The moving-lens effect is due to cosmological structure moving transverse to our line of sight and results of purely gravitational effects. As a result, it can be used to measure quantities which cannot be accessed directly with the kSZ effect alone, such as the growth rate of density fluctuations, which is useful for studying dark energy, modified gravity, and the effects of neutrino mass (Hotinli et al 2018). The polarized Sunyaev Zel'dovich (pSZ) effect is a linear blackbody polarization sourced by photons that scatter off free electrons in the large-scale structure, proportional to the remote CMB temperature quadrupole observed in the electrons’ rest frame. The measurement of pSZ effect allows reconstructing the linear polarization of the CMB as a function of redshift on the light cone. This allows probing the tensor to scalar ratio as well as the cosmic birefringence angle (Lee et al 2022) that is sourced if the physics behind dark energy and/or dark matter violates the parity symmetry assumed in the standard cosmological paradigm. Relativistic corrections to the remote quadrupole field also give rise to a non-blackbody polarization anisotropy proportional to the square of the transverse peculiar velocity field; this is the kinetic polarized Sunyaev Zel’dovich (kpSZ) effect. The kpSZ effect is also a probe of cosmic birefringence and primordial non-Gaussianity (Hotinli et al 2022).

Improving CMB measurements

show research

The effects of gravitational lensing and scattering on the CMB photons is both a help and hindrance to our understanding of the history and contents of the Universe. Both lensing and scattering effects manifest themselves as a distortion of the primary CMB anisotropies and also functions as an obstacle to analyses which rely on a pristine view of the last scattering surface. Simons Observatory and the succeeding experiments will map the CMB sky with unprecedented precision where these effects will be limiting factors to cosmological inference. My current and future efforts also focus on mitigating the effects of lensing and scattering on the primary CMB with the goal of improving the forthcoming cosmological parameter constraints. 

To this end, I co-developed a code for robust treatments of CMB ‘delensing’ on the curved sky, demonstrating the impact that removing the lensing effect from the CMB map will have with future surveys. Together with the forecasting framework I co-developed, these softwares are now becoming parts of the main analysis pipelines of the upcoming CMB experiments. I have also authored research papers on assessing the prospects to delens the CMB temperature and polarisation maps as well as to remove the kSZ effect using external galaxy templates, or ‘de-kSZing' (Foreman, Hotinli et al 2020).

The delensed CMB spectra have sharper acoustic peaks and more prominent damping tails, allowing for improved inferences of cosmological parameters that impact those features. Delensing reduces the B-mode power, aiding the search for primordial gravitational waves and allowing for lower variance reconstruction of lensing and other sources of secondary CMB anisotropies. Lensing-induced power spectrum covariances are reduced by delensing, simplifying analyses and improving constraints on primordial non-Gaussianities. Biases that result from incorrectly modelling nonlinear and baryonic feedback effects on the lensing power spectrum are mitigated by delensing. All of these benefits are possible without any changes to experimental or survey design. 

The situation with the kSZ effect is analogous to that of gravitational lensing. In a recent work, we investigated the prospects for external “de-kSZing'' of CMB temperature maps, whereby an external galaxy survey is used to construct a template for the kSZ effect, which is then subtracted from the observed maps. The kSZ effect can significantly bias lensing reconstruction, affecting our ability to constrain neutrino mass, dark energy, and dark matter. Furthermore, kSZ fluctuations at small scales add noise to measurements of the Silk-damped regime of the primary CMB, which can be used to probe Neff, the primordial helium abundance, and the spectrum of primordial scalar perturbations. The kSZ effect also acts as noise for measurements of other CMB secondaries, such as the moving-lens effect.

Delensing the CMB: An all-orders, full-spectra, iterative approach

show research

esa_lensingImage copyright: ESA and the Planck Collaboration

In this near-complete project, we have calculated and coded a fast and accurate delensing algorithm of CMB TT, TE, EE and BB spectra with the quadratic estimator up-to all orders in lensing potential. The scientific goal here is assessing how much we can recover the peak locations and the damping tail of the primary CMB spectra from a rigorous application of delensing with the quadratic estimator.

Project will produce public code to calculate the delensed CMB spectra and lensing reconstruction noise accompanied with an analysis and forecast paper in the near future. The code is planned as a user-friendly extension to the publicly available CLASS Boltzmann solver. 

Weak gravitational lensing plays a crucial role in many areas in cosmology including the potential detection of gravitational waves from the CMB polarization; determining the radiation content of the Universe; and uniquely probing dark matter at high redshifts. Mitigation procedure of this effect, also called 'delensing', improves the constraining power of the cosmological surveys significantly on a wide range of parameters and understood to be essential to future cosmological inference. My research focuses on developing the analytic framework for estimating prospects of optimal delensing the CMB observables with high fidelity. In this upcoming work we will aim leading the community towards the most fruitful directions forward through understanding the benefits of delensing.

Utilizing 21cm Velocity Acoustic Oscillations for fundamental physics

show research

As matter clusters under gravity, its components can behave very differently. While the majority of matter is collisionless, dark and cold, a fraction of it are baryons which couple to photons before the recombination at redshift z≈1100, giving rise to the baryon acoustic oscillations (BAOs) observed from the CMB and galaxy-surveys. The same physics also induces a bulk relative velocity between DM and baryons [1,2]. More recently, it was shown in [3] that for reasonable models of epochs up-to the end of reionization, detecting a velocity-induced acoustic feature, so called velocity acoustic oscillations (VAOs), may be possible from the measurement of 21cm hydrogen-line with the upcoming experiments such as HERA. The VAO feature provides an effective probe of the early Universe physics that affect the relative behaviour of DM and baryons. Unlike the matter power-spectrum, where the effect of BAO is small, the VAO feature is O(1) in the 21cm power-spectrum.

Measurements of the 21cm hydrogen-line provides a unique window into the cosmic-dawn era when the first stars were formed (z=15-20). During these early times the typical mass of collapsed baryonic objects fall near the critical mass below which gas pressure prevents their collapse. The abundance of Lyman-α photons that excite the hyperfine transition in neutral hydrogen and allow it to absorb 21cm photons from the CMB will depend on the collapse-fraction of baryons and will be directly impacted by effects that alter early structure growth, such as local modulations of relative DM-baryon densities due to bulk CIPs. Furthermore, as is the case for the BAO feature in the CMB and LSS observables, some characteristics of the VAO will be unaffected from the complicated local physics related to various feedback mechanisms that play role during the epoch of reionization and can be utilised to constrain effects that have a coherent impact of the observables on large-scales, such as CIPs. In this paper we discuss the detection significance of CIPs from measurement of the 21cm hydrogen line.


[Artist impression of the first stars, image credit: APS, Adolf Schaller for STScI]

While measurements of the 21cm brightness temperature power-spectrum allow discriminating between models of cosmology, measured variances of the modes (as well as the features of the power-spectrum themselves) are a set of cosmological observables too, whose properties can be described independently of any model. Some characteristics of these observables such as the amplitudes of individual Fourier modes are expected to be sensitive to the details of local mechanisms such as baryonic feedback. These effects need to be modelled and the model parameters need to be marginalised, weakening the constraining power of the measurements, making it hard to unambigiously distinguish between fundamental physics and astrophysical effects, for example. Observations VAOs, on the other hand, provide a unique measure whose various large-scale properties that are unaffected under these effects.

The difficulty for local interactions to produce coherent effects on long distances outside the sound horizon of primordial inhomogeneities allow the VAO scale and the VAO peak locations to be protected from details of the baryonic feedback. In an upcoming work with Marc Kamionkowski, we demonstrate the use of VAO peak locations for inferring details of the fundamental properties of our Universe. 


[Demonstration of 21cm power-spectrum VAO peak locations, upcoming work]

Probing correlated compensated isocurvature perturbations using scale-dependent galaxy bias

show research

The equivalence principle dictates that no local interaction could care about the local gravitational potential, in turn protecting the longest cosmological modes against effects of complicated non-linear physics. This fundamental and simple fact in turn makes the large-scale observables a very powerful probe of the type of effects that are not caused by local interactions (i.e. that are non-local) such as those that determine the characteristics of the curvature fluctuations at the early Universe. It has long been understood that one such observable is the galaxy bias (i.e. the fudge-factor that relates the galaxy number-densities to the total-matter number-density). This bias turns out to be scale invariant on large scales in the absence of non-local effects due to equivalence principle. The bias becomes scale dependent, however, in the presence of local non-Gaussianity, for example, as well as a specific type compensated isocurvature perturbations (CIPs) that are correlated with the primordial curvature perturbations. CIPs are modulations of the relative baryon and dark matter density that leave the total matter density constant. The best current constraints from the primary cosmic microwave background (CMB) are consistent with CIPs some two orders of magnitude larger in amplitude than adiabatic perturbations, suggesting that there may be a huge gap in our knowledge of the early Universe! 


see arXiv:1908.08953

Combining a galaxy survey with an unbiased tracer of the density field facilitates a measurement of the amplitude of correlated CIPs that is free from cosmic variance, the main limitation on constraints from the primary CMB. This is often called the 'sample variance cancellation'.

Among the most promising tracers to use for this purpose is the remote dipole field, reconstructed using the technique of kinetic Sunyaev Zel'dovich (kSZ) tomography.

With James Mertens, Matthew Johnson and Marc Kamionkowski, we evaluated the detection significance on the amplitude of correlated CIPs possible with next-generation CMB and galaxy surveys using kSZ tomography.  We find that kSZ tomography can probe CIPs of comparable amplitude to the adiabatic fluctuations, representing an improvement of over two orders of magnitude upon current constraints, and an order of magnitude over what will be possible using future CMB or galaxy surveys alone.

Transverse Velocities with the Moving Lens Effect

show research

planck_cmb[CMB by PLANCK]

Thanks to experiments such as WMAP and PLANCK, statistics of the primary  cosmic microwave background (CMB) is very well described. We have constraints on densities, curvature, age of the univers as well as some properties of the initial fluctuations such as Gaussianity, scale-invariance and adiabaticity. 

Remaining questions for the next generations large-scale cosmology experiments such as SO, CMB-S4, LSST and others are related to the growth of structure, and will involve better understanding the nature of neutrinos, dark photons, dark energy, modified gravity, inflation and dark matter. 

But going beyond the primary CMB; we should note that a CMB photon traverses the Universe and is subject to many effects such as deflection by gravitational potential (weak gravitational lensing); scattering on baryons in clusters (Sunyaev Zel’dovich effects); and getting redshifted due to evolving potentials, which at linear level includes the Sachs-Wolfe effect and also integrated Sachs Wolfe effect (at the non-linear order, is often called the Rees-Sciama effect).  

In addition, gravitational potentials which change in time induce fluctuations in the observed CMB temperature. Cosmological structure moving transverse to our line of sight provides a specific example known as the moving lens effect.  


The moving lens effect (shown above) has a dipolar shape on the CMB, centred around the gravitational potential and aligned with the velocity. 

We showed recently that the upcoming CMB experiments together with large-scale structure experiments will be able to detect the moving lens effect to high signal-to-noise ratio.


You can find out more in our paper here

This project also involves many great scientists!


Effect of reheating on predictions following multiple-field inflation

show research

Inflation, a period of accelerated expansion in the early Universe, solves many of the classical problems associated with the hot Big Bang scenario, and provides a natural mechanism for generating primordial cosmological fluctuations.

Observations are currently consistent with the simplest single-field, slow-roll models of inflation, e.g., the Planck observations of the cosmic microwave background (CMB) indicate a featureless power-law shape for the primordial power spectrum of scalar fluctuations and no detectable primordial non-Gaussianity or tensor fluctuations.

Single-field models are phenomenologically successful, however, they often lack the generality of more complex scenarios. More notably, they are not always natural from a theoretical point of view, e.g., string compactifications often result in hundreds of scalar fields appearing in the low energy effective action.

Models with multiple fields naturally produce fluctuations that are non-adiabatic, whose presence allows the curvature perturbation and its correlation functions to evolve outside the Hubble radius. Therefore, in order to make predictions in multifield models, it is necessary to understand the evolution of the correlation functions until either the curvature fluctuations become adiabatic or they are directly observed. 

In a recent paper at, we provided a general methodology for calculating the adiabatic power spectrum of curvature perturbations after multifield inflation for any number of scalar fields. Scenarios with many fields also tend to predict an amount of isocurvature perturbations at the end of inflation which increases with the number of fields, thereby elevating the importance of studying the effects of reheating for these models.

We found that details of reheating can effect the inflationary observables significantly.


The search for anisotropy in the gravitational-wave background with pulsar-timing arrays and astrometry

show research

Pulsar-timing arrays (PTAs) are seeking gravitational waves from supermassive-black-hole binaries, and there are prospects to complement these searches with stellar-astrometry measurements.


Theorists still disagree, however, as to whether the local gravitational-wave background will be isotropic, as arises if it is the summed contributions from many SMBH binaries, or whether it exhibits the type of anisotropy that arises if the local background is dominated by a handful (or even one) bright source.

In a recent work with Prof. Marc Kamionkowski and Prof. Andrew Jaffe, we derive, using bipolar spherical harmonics, the optimal PTA estimators for anisotropy in the GW background and simple estimates of the detectability of this anisotropy.


We found that the GWB needs to be detected to very high precision before we will be able to tell, unambiguously, whether it is statistically isotropic or not.

You can read our paper here: