Search

Context. Deep learning (DL) techniques are a promising approach among the set of methods used in the ever-challenging determination of stellar parameters in M dwarfs. In this context, transfer learning could play an important role in mitigating uncertainties in the results due to the synthetic gap (i.e. difference in feature distributions between observed and synthetic data). Aims. We propose a feature-based deep transfer learning (DTL) approach based on autoencoders to determine stellar parameters from high-resolution spectra. Using this methodology, we provide new estimations for the effective temperature, surface gravity, metallicity, and projected rotational velocity for 286 M dwarfs observed by the CARMENES survey. Methods. Using autoencoder architectures, we projected synthetic PHOENIX-ACES spectra and observed CARMENES spectra onto a new feature space of lower dimensionality in which the differences between the two domains are reduced. We used this low-dimensional new feature space as input for a convolutional neural network to obtain the stellar parameter determinations. Results. We performed an extensive analysis of our estimated stellar parameters, ranging from 3050 to 4300 K, 4.7 to 5.1 dex, and −0.53 to 0.25 dex for Teff, log 푔, and [Fe/H], respectively. Our results are broadly consistent with those of recent studies using CARMENES data, with a systematic deviation in our Teff scale towards hotter values for estimations above 3750 K. Furthermore, our methodology mitigates the deviations in metallicity found in previous DL techniques due to the synthetic gap. Conclusions. We consolidated a DTL-based methodology to determine stellar parameters in M dwarfs from synthetic spectra, with no need for high-quality measurements involved in the knowledge transfer. These results suggest the great potential of DTL to mitigate the differences in feature distributions between the observations and the PHOENIX-ACES spectra. [ABSTRACT FROM AUTHOR]

We determined effective temperatures, surface gravities, and metallicities for a sample of 343 M dwarfs observed with CARMENES, the double-channel, high-resolution spectrograph installed at the 3.5 m telescope at Calar Alto Observatory. We employed STEPARSYN, a Bayesian spectral synthesis implementation particularly designed to infer the stellar atmospheric parameters of late-type stars following a Markov chain Monte Carlo approach. We made use of the BT-Settl model atmospheres and the radiative transfer code turbospectrum to compute a grid of synthetic spectra around 75 magnetically insensitive Fe I and Ti I lines plus the TiO γ and ϵ bands. To avoid any potential degeneracy in the parameter space, we imposed Bayesian priors on Teff and logg based on the comprehensive, multi-band photometric data available for the sample. We find that this methodology is suitable down to M7.0 V, where refractory metals such as Ti are expected to condense in the stellar photospheres. The derived Teff, logg, and [Fe/H] range from 3000 to 4200 K, 4.5 to 5.3 dex, and −0.7 to 0.2 dex, respectively. Although our Teff scale is in good agreement with the literature, we report large discrepancies in the [Fe/H] scales, which might arise from the different methodologies and sets of lines considered. However, our [Fe/H] is in agreement with the metallicity distribution of FGK-type stars in the solar neighbourhood and correlates well with the kinematic membership of the targets in the Galactic populations. Lastly, excellent agreement in Teff is found for M dwarfs with interferometric angular diameter measurements, as well as in the [Fe/H] between the components in the wide physical FGK+M and M+M systems included in our sample. [ABSTRACT FROM AUTHOR]

Due to their ubiquity and very long main-sequence lifetimes, abundance determinations in M dwarfs provide a powerful and alternative tool to GK dwarfs to study the formation and chemical enrichment history of our Galaxy. In this study, abundances of the neutron-capture elements Rb, Sr, and Zr are derived, for the first time, in a sample of nearby M dwarfs. We focus on stars in the metallicity range − 0.5 ≲ [Fe/H] ≲ +0.3, an interval poorly explored for Rb abundances in previous analyses. To do this we use high-resolution, high-signal-to-noise-ratio, optical and near-infrared spectra of 57 M dwarfs observed with CARMENES. The resulting [Sr/Fe] and [Zr/Fe] ratios for most M dwarfs are almost constant at about the solar value, and are identical to those found in GK dwarfs of the same metallicity. However, for Rb we find systematic underabundances ([Rb/Fe] < 0.0) by a factor two on average. Furthermore, a tendency is found for Rb – but not for other heavy elements (Sr, Zr) – to increase with increasing metallicity such that [Rb/Fe] ≳ 0.0 is attained at metallicities higher than solar. These are surprising results, never seen for any other heavy element, and are difficult to understand within the formulation of the s- and r-processes, both contributing sources to the Galactic Rb abundance. We discuss the reliability of these findings for Rb in terms of non-LTE (local thermodynamic equilibrium) effects, stellar activity, or an anomalous Rb abundance in the Solar System, but no explanation is found. We then interpret the full observed [Rb/Fe] versus [Fe/H] trend within the framework of theoretical predictions from state-of-the-art chemical evolution models for heavy elements, but a simple interpretation is not found either. In particular, the possible secondary behaviour of the [Rb/Fe] ratio at super-solar metallicities would require a much larger production of Rb than currently predicted in AGB stars through the s-process without overproducing Sr and Zr. [ABSTRACT FROM AUTHOR]

Existing and upcoming instrumentation is collecting large amounts of astrophysical data, which require efficient and fast analysis techniques. We present a deep neural network architecture to analyze high-resolution stellar spectra and predict stellar parameters such as effective temperature, surface gravity, metallicity, and rotational velocity. With this study, we firstly demonstrate the capability of deep neural networks to precisely recover stellar parameters from a synthetic training set. Secondly, we analyze the application of this method to observed spectra and the impact of the synthetic gap (i.e., the difference between observed and synthetic spectra) on the estimation of stellar parameters, their errors, and their precision. Our convolutional network is trained on synthetic PHOENIX-ACES spectra in different optical and near-infrared wavelength regions. For each of the four stellar parameters, Teff, log g, [M/H], and v sin i, we constructed a neural network model to estimate each parameter independently. We then applied this method to 50 M dwarfs with high-resolution spectra taken with CARMENES (Calar Alto high-Resolution search for M dwarfs with Exo-earths with Near-infrared and optical Échelle Spectrographs), which operates in the visible (520–960 nm) and near-infrared wavelength range (960–1710 nm) simultaneously. Our results are compared with literature values for these stars. They show mostly good agreement within the errors, but also exhibit large deviations in some cases, especially for [M/H], pointing out the importance of a better understanding of the synthetic gap. [ABSTRACT FROM AUTHOR]

Context. The relevance of M dwarfs in the search for potentially habitable Earth-sized planets has grown significantly in the last years. Aims. In our on-going effort to comprehensively and accurately characterise confirmed and potential planet-hosting M dwarfs, in particular for the CARMENES survey, we have carried out a comprehensive multi-band photometric analysis involving spectral energy distributions, luminosities, absolute magnitudes, colours, and spectral types, from which we have derived basic astrophysical parameters. Methods. We have carefully compiled photometry in 20 passbands from the ultraviolet to the mid-infrared, and combined it with the latest parallactic distances and close-multiplicity information, mostly from Gaia DR2, of a sample of 2479 K5 V to L8 stars and ultracool dwarfs, including 2210 nearby, bright M dwarfs. For this, we made extensive use of Virtual Observatory tools. Results. We have homogeneously computed accurate bolometric luminosities and effective temperatures of 1843 single stars, derived their radii and masses, studied the impact of metallicity, and compared our results with the literature. The over 40 000 individually inspected magnitudes, together with the basic data and derived parameters of the stars, individual and averaged by spectral type, have been made public to the astronomical community. In addition, we have reported 40 new close multiple systems and candidates (ρ <  3.3 arcsec) and 36 overluminous stars that are assigned to young Galactic populations. Conclusions. In the new era of exoplanet searches around M dwarfs via transit (e.g. TESS, PLATO) and radial velocity (e.g. CARMENES, NIRPS+HARPS), this work is of fundamental importance for stellar and therefore planetary parameter determination. [ABSTRACT FROM AUTHOR]

The He I infrared (IR) triplet at 10 830 Å is known as an activity indicator in solar-type stars and has become a primary diagnostic in exoplanetary transmission spectroscopy. He I IR lines are a tracer of the stellar extreme-ultraviolet irradiation from the transition region and corona. We study the variability of the He I triplet lines in a spectral time series of 319 M dwarf stars that was obtained with the CARMENES high-resolution optical and near-infrared spectrograph at Calar Alto. We detect He I IR line variability in 18% of our sample stars, all of which show Hα in emission. Therefore, we find detectable He I variability in 78% of the sub-sample of stars with Hα emission. Detectable variability is strongly concentrated in the latest spectral sub-types, where the He I lines during quiescence are typically weak. The fraction of stars with detectable He I variation remains lower than 10% for stars earlier than M3.0 V, while it exceeds 30% for the later spectral sub-types. Flares are accompanied by particularly pronounced line variations, including strongly broadened lines with red and blue asymmetries. However, we also find evidence for enhanced He I absorption, which is potentially associated with increased high-energy irradiation levels at flare onset. Generally, He I and Hα line variations tend to be correlated, with Hα being the most sensitive indicator in terms of pseudo-equivalent width variation. This makes the He I triplet a favourable target for planetary transmission spectroscopy. [ABSTRACT FROM AUTHOR]

The high-resolution, dual channel, visible and near-infrared spectrograph CARMENES offers exciting opportunities for stellar and exoplanetary research on M dwarfs. In this work we address the challenge of reaching the highest radial velocity precision possible with a complex, actively cooled, cryogenic instrument, such as the near-infrared channel. We describe the performance of the instrument and the work flow used to derive precise Doppler measurements from the spectra. The capability of both CARMENES channels to detect small exoplanets is demonstrated with the example of the nearby M5.0 V star CD Cet (GJ 1057), around which we announce a super-Earth (4.0 ± 0.4 M⊕) companion on a 2.29 d orbit. [ABSTRACT FROM AUTHOR]

We present precise photospheric parameters of 282 M dwarfs determined from fitting the most recent version of PHOENIX models to high-resolution CARMENES spectra in the visible (0.52–0.96 μm) and NIR wavelength range (0.96–1.71 μm). With its aim to search for habitable planets around M dwarfs, several planets of different masses have been detected. The characterization of the target sample is important for the ability to derive and constrain the physical properties of any planetary systems that are detected. As a continuation of previous work in this context, we derived the fundamental stellar parameters effective temperature, surface gravity, and metallicity of the CARMENES M-dwarf targets from PHOENIX model fits using a χ2 method. We calculated updated PHOENIX stellar atmosphere models that include a new equation of state to especially account for spectral features of low-temperature stellar atmospheres as well as new atomic and molecular line lists. We show the importance of selecting magnetically insensitive lines for fitting to avoid effects of stellar activity in the line profiles. For the first time, we directly compare stellar parameters derived from multiwavelength range spectra, simultaneously observed for the same star. In comparison with literature values we show that fundamental parameters derived from visible spectra and visible and NIR spectra combined are in better agreement than those derived from the same spectra in the NIR alone. [ABSTRACT FROM AUTHOR]

Although M dwarfs are known for high levels of stellar activity, they are ideal targets for the search of low-mass exoplanets with the radial velocity (RV) method. We report the discovery of a planetary-mass companion around LSPM J2116+0234 (M3.0 V) and confirm the existence of a planet orbiting GJ 686 (BD+18 3421; M1.0 V). The discovery of the planet around LSPM J2116+0234 is based on CARMENES RV observations in the visual and near-infrared channels. We confirm the planet orbiting around GJ 686 by analyzing the RV data spanning over two decades of observationsfrom CARMENES VIS, HARPS-N, HARPS, and HIRES. We find planetary signals at 14.44 and 15.53 d in the RV data for LSPM J2116+0234 and GJ 686, respectively. Additionally, the RV, photometric time series, and various spectroscopic indicators show hints of variations of 42 d for LSPM J2116+0234 and 37 d for GJ 686, which we attribute to the stellar rotation periods. The orbital parameters of the planets are modeled with Keplerian fits together with correlated noise from the stellar activity. A mini-Neptune with a minimum mass of 11.8 M⊕ orbits LSPM J2116+0234 producing a RV semi-amplitude of 6.19 m s−1, while a super-Earth of mass 6.6 M⊕ orbits GJ 686 and produces a RV semi-amplitude of 3.0 m s−1. Both LSPM J2116+0234 and GJ 686 have planetary companions populating the regime of exoplanets with masses lower than 15 M⊕ and orbital periods <20 d. [ABSTRACT FROM AUTHOR]

Context. M dwarfs are known to generate the strongest magnetic fields among main-sequence stars with convective envelopes, but we are still lacking a consistent picture of the link between the magnetic fields and underlying dynamo mechanisms, rotation, and activity. Aims. In this work we aim to measure magnetic fields from the high-resolution near-infrared spectra taken with the CARMENES radial-velocity planet survey in a sample of 29 active M dwarfs and compare our results against stellar parameters. Methods. We used the state-of-the-art radiative transfer code to measure total magnetic flux densities from the Zeeman broadening of spectral lines and filling factors. Results. We detect strong kG magnetic fields in all our targets. In 16 stars the magnetic fields were measured for the first time. Our measurements are consistent with the magnetic field saturation in stars with rotation periods P < 4 d. The analysis of the magnetic filling factors reveal two different patterns of either very smooth distribution or a more patchy one, which can be connected to the dynamo state of the stars and/or stellar mass. Conclusions. Our measurements extend the list of M dwarfs with strong surface magnetic fields. They also allow us to better constrain the interplay between the magnetic energy, stellar rotation, and underlying dynamo action. The high spectral resolution and observations at near-infrared wavelengths are the beneficial capabilities of the CARMENES instrument that allow us to address important questions about the stellar magnetism. [ABSTRACT FROM AUTHOR]

Chromospheric modeling of observed differences in stellar activity lines is imperative to fully understand the upper atmospheres of late-type stars. We present one-dimensional parametrized chromosphere models computed with the atmosphere code PHOENIX using an underlying photosphere of 3500 K. The aim of this work is to model chromospheric lines of a sample of 50 M2–3 dwarfs observed in the framework of the CARMENES, the Calar Alto high-Resolution search for M dwarfs with Exo-earths with Near-infrared and optical Echelle Spectrographs, exoplanet survey. The spectral comparison between observed data and models is performed in the chromospheric lines of Na I D2, Hα, and the bluest Ca II infrared triplet line to obtain best-fit models for each star in the sample. We find that for inactive stars a single model with a VAL C-like temperature structure is sufficient to describe simultaneously all three lines adequately. Active stars are rather modeled by a combination of an inactive and an active model, also giving the filling factors of inactive and active regions. Moreover, the fitting of linear combinations on variable stars yields relationships between filling factors and activity states, indicating that more active phases are coupled to a larger portion of active regions on the surface of the star. [ABSTRACT FROM AUTHOR]

We use spectra from CARMENES, the Calar Alto high-Resolution search for M dwarfs with Exo-earths with Near-infrared and optical Echelle Spectrographs, to search for periods in chromospheric indices in 16 M0–M2 dwarfs. We measure spectral indices in the Hα, the Ca II infrared triplet (IRT), and the Na I D lines to study which of these indices are best-suited to finding rotation periods in these stars. Moreover, we test a number of different period-search algorithms, namely the string length method, the phase dispersion minimisation, the generalized Lomb–Scargle periodogram, and the Gaussian process regression with quasi-periodic kernel. We find periods in four stars using Hα and in five stars using the Ca II IRT, two of which have not been found before. Our results show that both Hα and the Ca II IRT lines are well suited for period searches, with the Ca II IRT index performing slightly better than Hα. Unfortunately, the Na I D lines are strongly affected by telluric airglow, and we could not find any rotation period using this index. Further, different definitions of the line indices have no major impact on the results. Comparing the different search methods, the string length method and the phase dispersion minimisation perform worst, while Gaussian process models produce the smallest numbers of false positives and non-detections. [ABSTRACT FROM AUTHOR]

Context. The Calar Alto high-Resolution search for M dwarfs with Exo-earths with Near-infrared and optical Echelle Spectrographs (CARMENES) survey is searching for Earth-like planets orbiting M dwarfs using the radial velocity method. Studying the stellar activity of the target stars is important to avoid false planet detections and to improve our understanding of the atmospheres of late-type stars. Aims. In this work we present measurements of activity indicators at visible and near-infrared wavelengths for 331 M dwarfs observed with CARMENES. Our aim is to identify the activity indicators that are most sensitive and easiest to measure, and the correlations among these indicators. We also wish to characterise their variability. Methods. Using a spectral subtraction technique, we measured pseudo-equivalent widths of the He I D3, Hα, He Iλ10833 Å, and Pa β lines, the Na I D doublet, and the Ca II infrared triplet, which have a chromospheric component in active M dwarfs. In addition, we measured an index of the strength of two TiO and two VO bands, which are formed in the photosphere. We also searched for periodicities in these activity indicators for all sample stars using generalised Lomb-Scargle periodograms. Results. We find that the most slowly rotating stars of each spectral subtype have the strongest Hα absorption. Hα is correlated most strongly with He I D3, whereas Na I D and the Ca II infrared triplet are also correlated with Hα. He Iλ10833 Å and Paβ show no clear correlations with the other indicators. The TiO bands show an activity effect that does not appear in the VO bands. We find that the relative variations of Hα and He I D3 are smaller for stars with higher activity levels, while this anti-correlation is weaker for Na I D and the Ca II infrared triplet, and is absent for He Iλ10833 Å and Paβ. Periodic variation with the rotation period most commonly appears in the TiO bands, Hα, and in the Ca II infrared triplet. [ABSTRACT FROM AUTHOR]

Aims. The main goal of this work is to measure rotation periods of the M-type dwarf stars being observed by the CARMENES exoplanet survey to help distinguish radial-velocity signals produced by magnetic activity from those produced by exoplanets. Rotation periods are also fundamental for a detailed study of the relation between activity and rotation in late-type stars. Methods. We look for significant periodic signals in 622 photometric time series of 337 bright, nearby M dwarfs obtained by long-time baseline, automated surveys (MEarth, ASAS, SuperWASP, NSVS, Catalina, ASAS-SN, K2, and HATNet) and for 20 stars which we obtained with four 0.2–0.8 m telescopes at high geographical latitudes. Results. We present 142 rotation periods (73 new) from 0.12 d to 133 d and ten long-term activity cycles (six new) from 3.0 a to 11.5 a. We compare our determinations with those in the existing literature; we investigate the distribution of Prot in the CARMENES input catalogue, the amplitude of photometric variability, and their relation to v sini and pEW(Hα); and we identify three very active stars with new rotation periods between 0.34 d and 23.6 d. [ABSTRACT FROM AUTHOR]

We announce the discovery of two planetary companions orbiting around the low-mass stars Ross 1020 (GJ 3779, M4.0V) and LP 819-052 (GJ 1265, M4.5V). The discovery is based on the analysis of CARMENES radial velocity (RV) observations in the visual channel as part of its survey for exoplanets around M dwarfs. In the case of GJ 1265, CARMENES observations were complemented with publicly available Doppler measurements from HARPS. The datasets reveal two planetary companions, one for each star, that share very similar properties: minimum masses of 8.0 ± 0.5 M⊕ and 7.4 ± 0.5 M⊕ in low-eccentricity orbits with periods of 3.023 ± 0.001 d and 3.651 ± 0.001 d for GJ 3779 b and GJ 1265 b, respectively. The periodic signals around 3 d found in the RV data have no counterpart in any spectral activity indicator. Furthermore, we collected available photometric data for the two host stars, which confirm that the additional Doppler variations found at periods of approximately 95 d can be attributed to the rotation of the stars. The addition of these planets to a mass-period diagram of known planets around M dwarfs suggests a bimodal distribution with a lack of short-period low-mass planets in the range of 2–5 M⊕. It also indicates that super-Earths (>5 M⊕) currently detected by RV and transit techniques around M stars are usually found in systems dominated by a single planet. [ABSTRACT FROM AUTHOR]

Context. The CARMENES spectrograph is surveying ∼300 M dwarf stars in search for exoplanets. Among the target stars, spectroscopic binary systems have been discovered, which can be used to measure fundamental properties of stars. Aims. Using spectroscopic observations, we determine the orbital and physical properties of nine new double-line spectroscopic binary systems by analysing their radial velocity curves. Methods. We use two-dimensional cross-correlation techniques to derive the radial velocities of the targets, which are then employed to determine the orbital properties. Photometric data from the literature are also analysed to search for possible eclipses and to measure stellar variability, which can yield rotation periods. Results. Out of the 342 stars selected for the CARMENES survey, 9 have been found to be double-line spectroscopic binaries, with periods ranging from 1.13 to ∼8000 days and orbits with eccentricities up to 0.54. We provide empirical orbital properties and minimum masses for the sample of spectroscopic binaries. Absolute masses are also estimated from mass-luminosity calibrations, ranging between ∼0.1 and ∼0.6 M⊙. Conclusions. These new binary systems increase the number of double-line M dwarf binary systems with known orbital parameters by 15%, and they have lower mass ratios on average. [ABSTRACT FROM AUTHOR]

Context. The new CARMENES instrument comprises two high-resolution and high-stability spectrographs that are used to search for habitable planets around M dwarfs in the visible and near-infrared regime via the Doppler technique. Aims. Characterising our target sample is important for constraining the physical properties of any planetary systems that are detected. The aim of this paper is to determine the fundamental stellar parameters of the CARMENES M-dwarf target sample from high-resolution spectra observed with CARMENES. We also include several M-dwarf spectra observed with other high-resolution spectrographs, that is CAFE, FEROS, and HRS, for completeness. Methods. We used a χ2 method to derive the stellar parameters effective temperature Teff, surface gravity logg, and metallicity [Fe/H] of the target stars by fitting the most recent version of the PHOENIX-ACES models to high-resolution spectroscopic data. These stellar atmosphere models incorporate a new equation of state to describe spectral features of low-temperature stellar atmospheres. Since Teff, logg, and [Fe/H] show degeneracies, the surface gravity is determined independently using stellar evolutionary models. Results. We derive the stellar parameters for a total of 300 stars. The fits achieve very good agreement between the PHOENIX models and observed spectra. We estimate that our method provides parameters with uncertainties of σTeff = 51 K, σlog g = 0.07, and σ[Fe/H] = 0.16, and show that atmosphere models for low-mass stars have significantly improved in the last years. Our work also provides an independent test of the new PHOENIX-ACES models, and a comparison for other methods using low-resolution spectra. In particular, our effective temperatures agree well with literature values, while metallicities determined with our method exhibit a larger spread when compared to literature results. [ABSTRACT FROM AUTHOR]

CARMENES is a spectrograph for radial velocity surveys of M dwarfs with the aim of detecting Earth-mass planets orbiting in the habitable zones of their host stars. To ensure an optimal use of the CARMENES guaranteed time observations, in this paper we investigate the correlation of activity and rotation for approximately 2200 M dwarfs, ranging in spectral type from M0.0 V to M9.0 V. We present new high-resolution spectroscopic observations with FEROS, CAFE, and HRS of approximately 500 M dwarfs. For each new observation, we determined its radial velocity and measured its Hα activity index and its rotation velocity. Additionally, we have multiple observations of many stars to investigate if there are any radial velocity variations due to multiplicity. The results of our survey confirm that early-M dwarfs are Hα inactive with low rotational velocities and that late-M dwarfs are Hα active with very high rotational velocities. The results of this high-resolution analysis comprise the most extensive catalogue of rotation and activity in M dwarfs currently available. [ABSTRACT FROM AUTHOR]

Context. Targeted spectroscopic exoplanet surveys face the challenge of maximizing their planet detection rates by means of careful planning. For a large planet survey, the number of possible observation combinations, i.e., the sequence of observations night after night, both in total time and amount of targets, is enormous. Aims. Sophisticated scheduling tools and the improved understanding of the exoplanet population are employed to investigate an efficient and optimal way to plan the execution of observations. This is applied to the CARMENES instrument, which is an optical and infrared high-resolution spectrograph that has started a survey of about 300 M-dwarf stars in search of terrestrial exoplanets. Methods. We used evolutionary computation techniques to create an automatic scheduler that minimizes the idle periods of the telescope and distributes the observations among all the targets using configurable criteria. We simulated the case of the CARMENES survey with a realistic sample of targets, and we estimated the efficiency of the planning tool both in terms of telescope operations and planet detection. Results. Our scheduling simulations produce plans that use about 99% of the available telescope time (including overheads) and optimally distribute the observations among the different targets. Under such conditions, and using current planet statistics, the optimized plan using this tool should allow the CARMENES survey to discover about 65% of the planets with radial-velocity semi-amplitudes greater than 1 m s-1 when considering only photon noise. Conclusions. The simulations using our scheduling tool show that it is possible to optimize the survey planning by minimizing idle instrument periods and fulfilling the science objectives in an efficient manner to maximize the scientific return. [ABSTRACT FROM AUTHOR]

Context. CARMENES is a stabilised, high-resolution, double-channel spectrograph at the 3.5m Calar Alto telescope. It is optimally designed for radial-velocity surveys of M dwarfs with potentially habitable Earth-mass planets. Aims. We prepare a list of the brightest, single M dwarfs in each spectral subtype observable from the northern hemisphere, from which we will select the best planet-hunting targets for CARMENES. Methods. In this first paper on the preparation of our input catalogue, we compiled a large amount of public data and collected low-resolution optical spectroscopy with CAFOS at the 2.2m Calar Alto telescope for 753 stars. We derived accurate spectral types using a dense grid of standard stars, a double least-squares minimisation technique, and 31 spectral indices previously defined by other authors. Additionally, we quantified surface gravity, metallicity, and chromospheric activity for all the stars in our sample. Results. We calculated spectral types for all 753 stars, of which 305 are new and 448 are revised. We measured pseudo-equivalent widths of Hα for all the stars in our sample, concluded that chromospheric activity does not affect spectral typing from our indices, and tabulated 49 stars that had been reported to be young stars in open clusters, moving groups, and stellar associations. Of the 753 stars, two are new subdwarf candidates, three are T Tauri stars, 25 are giants, 44 are K dwarfs, and 679 are M dwarfs. Many of the 261 investigated dwarfs in the range M4.0-8.0V are among the brightest stars known in their spectral subtype. Conclusions. This collection of low-resolution spectroscopic data serves as a candidate target list for the CARMENES survey and can be highly valuable for other radial-velocity surveys of M dwarfs and for studies of cool dwarfs in the solar neighbourhood. [ABSTRACT FROM AUTHOR]