Add spaces before citations (#41)

paper.md

@@ -33,9 +33,9 @@ The lack of data and terminology standardization in ultrasound technology resear
 At its lowest level, `QUPS` provides memory efficient definition classes capable of representing a wide variety of common imaging systems and enforces standardized units, coordinate systems, and nomenclature. 
 All classes have data visualization support, making it far easier to share and validate data amongst collaborators and with the public. 
 
-`QUPS` provides a variety of compute and memory optimized implementations of common signal processing methods (e.g. beamforming, resampling, or convolution), which take advantage of native parallelization tools within MATLAB or hardware acceleration via CUDA or OpenCL when feasible by leveraging `MatCL`[@heinisch2018MatCL]. `QUPS` exploits data parallellism primarily using a single program multiple data (SPMD) approach, and optimistically dispatches to the most efficient implementation available on a user's computer. For access to cutting edge research, it can also export or import to `USTB`[@USTB2017], an ultrasound toolbox which contains a library of advanced algorithms from the literature.
+`QUPS` provides a variety of compute and memory optimized implementations of common signal processing methods (e.g. beamforming, resampling, or convolution), which take advantage of native parallelization tools within MATLAB or hardware acceleration via CUDA or OpenCL when feasible by leveraging `MatCL` [@heinisch2018MatCL]. `QUPS` exploits data parallellism primarily using a single program multiple data (SPMD) approach, and optimistically dispatches to the most efficient implementation available on a user's computer. For access to cutting edge research, it can also export or import to `USTB` [@USTB2017], an ultrasound toolbox which contains a library of advanced algorithms from the literature.
 
-To facilitate data acquisition, `QUPS` provides tools to import data from [Verasonics Vantage](https://verasonics.com/vantage-systems/) ultrasound research scanners, which uses a flexible yet highly complex data standard. Similarly, to facilitate data generation, `QUPS` also interfaces with multiple popular free ultrasound field simulators including `FieldII`[@Jensen1996], `MUST`[@Garcia2021], and `k-Wave`[@Treeby2010]. Simulations can be batched on a compute cluster via MATLAB's parallel server toolbox.
+To facilitate data acquisition, `QUPS` provides tools to import data from [Verasonics Vantage](https://verasonics.com/vantage-systems/) ultrasound research scanners, which uses a flexible yet highly complex data standard. Similarly, to facilitate data generation, `QUPS` also interfaces with multiple popular free ultrasound field simulators including `FieldII` [@Jensen1996], `MUST` [@Garcia2021], and `k-Wave` [@Treeby2010]. Simulations can be batched on a compute cluster via MATLAB's parallel server toolbox.
 
 Alternatively, when a coarse but fast approximation is preferable, `QUPS` offers a hardware accelerated simulator with minimal complexity. This method uses a green's function to simulate ideal isotropic point sources (transmitters), point receivers, and point scatterers with radial propagation loss using the equation