https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824824#M5059 When you do succeed in building the single precision version program, do not expect the example program supplied with the MKL distribution to work unmodified. In particular, when you approximate derivatives using numerical differentiation, you should make the convergence criteria less stringent. Fri, 23 Sep 2011 12:21:37 GMT mecej4 2011-09-23T12:21:37Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824821#M5056 According to the reference both the single and double precision versions of the nonlinear least squares solver routines can be accessed by including mkl_rci.fi . But, from what I can tell this file only references the double precision routines. Indeed, my program compiles and runs fine when I call the double precision versions, but I cannot compile the single precision version because the routines are apparently not found.<BR /><BR />What am I missing here? Thu, 22 Sep 2011 20:42:02 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824821#M5056 mpiuser 2011-09-22T20:42:02Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824822#M5057 Hi,<DIV>What version of MKL do you use?</DIV><DIV>With best regards,</DIV><DIV>Alexander Kalinkin</DIV> Fri, 23 Sep 2011 01:47:47 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824822#M5057 Alexander_K_Intel2 2011-09-23T01:47:47Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824823#M5058 the single precision support has been added since the 10.3 version. Fri, 23 Sep 2011 05:43:39 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824823#M5058 Gennady_F_Intel 2011-09-23T05:43:39Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824824#M5059 When you do succeed in building the single precision version program, do not expect the example program supplied with the MKL distribution to work unmodified. In particular, when you approximate derivatives using numerical differentiation, you should make the convergence criteria less stringent. Fri, 23 Sep 2011 12:21:37 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/mkl-rci-fi-nonlinear-least-squares-single-precision/m-p/824824#M5059 mecej4 2011-09-23T12:21:37Z