# Suppression of 1/f noise in one-qubit systems

Pekko Kuopanportti, Mikko Mottonen, Ville Bergholm, Olli-Pentti Saira, Jun Zhang, K. Birgitta Whaley

## Main idea

• A more efficient way to model $$1/f$$ noise is proposed (compared to a sum of random telegraph fluctuators), by constructing the transition matrix from scratch

## Background

• $$1/f$$ noise is ubiquitous, yet the origins are generally unknown
• $$1/f$$ (or in general $$1/f^\alpha$$) refers to the scaling of the power spectrum at low frequency (and generally causes slow drifts in control signals)
• One way to model it is by a series of Random Telegraph Noise (RTN) fluctuators; a collection of noise variables that oscillate between two values
• With enough RTN fluctuators oscillating between approprately chosen frequencies, the spectrum indeed scales like $$1/f$$ (in the limit)
• Solving the associated equations exactly however (to get the mean effect of the noise) quickly becomes prohibitive, since the number of noise states scales as $$2^N$$, for $$N$$ oscillators

## Results

• In this work, they propose to model $$1/f$$ by constructing a transition matrix describing transitions between noise states “by hand”
• The general idea is that the transition matrix must satisfy certain properties to give a $$1/f$$ spectrum, which can be reverse engineered
• They argue that a size $$M$$ transition matrix constructed this way is equivalent to a system of $$M$$ RTN souces (which gives an exponentially larger transition matrix of size $$2^M$$), and therefore there is an exponential improvement
• This improvement stems from the observation that the constructed transition matrix is in some sense a more dense representation, and can represent more noise states, compared to an equivalent sized transition matrix arising from the RTN sources

tags: noise