\ļeft
and \right
need to be balanced before changing line. You can balance a \left(
with \right.
and a \left.
balances a \right)
(notice the dot), as in Boris' answer.
However, I would suggest you to use the \bigl
, \bigr
family of commands; they don't have to be balanced, they produce more consistent spacing, and they don't require eventual trickery as the use of phantoms to adjust the size.
In the example below I provided some fake definitions for some of the commands you were using without providing the definitions, and suppresses some of the superfluous \left
, \right
pairs, but in fact, you can dispense the use of all of them with the family I mentioned.
\documentclass{article}
\usepackage{amsmath}
\newcommand\hb{HB}
\newcommand\gb{GB}
\newcommand\dd{DD}
\DeclareMathOperator{\csch}{csch}
\begin{document}
\begin{multline}
e^{-y \lambda_n}
\Biggl(
-\frac{1}{2} \cosh (y \lambda_n) \csch (H \lambda_n) e^{(H + y) \lambda_n} \int_0^H \frac{\rho e^{ - K_2 \lambda_n}}{\lambda_n}\left(\hb_n(K_2,t) - \gb_n(K_2,t)\right) \, \dd K_2 \\
{} + \frac{1}{4} (e^{2 y \lambda_n} + 1) \bigl(\coth \left(H \lambda_n\bigr) - 1\right) \int_0^H \frac{\rho e^{K_1 \lambda_n}}{\lambda_n}\left(\gb_n(K_1,t) - \hb_n(K_1,t)\right) \, \dd K_1 \\
{}+\frac{1}{2} \int_0^y \frac{\rho e^{K_1 \lambda_n}}{\lambda_n}\left(\gb_n(K_1,t) - \hb_n(K_1,t)\right) \, \dd K_1 \\
{}+ \frac{1}{2} e^{2 y \lambda_n} \int_0^y \frac{\rho e^{ - K_2 \lambda_n}}{\lambda_n}\left(\hb_n(K_2,t) - \gb_n(K_2,t)\right) \, \dd K_2
\Biggr)
\end{multline}
\end{document}
Notice also than when breaking an expression at a binary operator, the operator usually goes in the new line (as in my example) and not in the old line (as in your code), but there are exceptions to this (Russian use, I think).
Notice also the use of
\DeclareMathOperator{\csch}{csch}
instead of \text{csch}
to obtain right font and spacing.
this is a kludge, but you can take advantage of the \multline
\shoveright
feature
to reposition lines that are erroneously positioned too far to the left:
\documentclass[fleqn]{article}
\usepackage{amsmath}
\begin{document}
\begin{multline}
1+2+3+4+5+6+7+8+9+10+11+12 \\
\shoveright{+13+14+15+16+17+18+19+20+21+22} \kern.1\textwidth \\
+23+24+25+26+27+28+29+30+31
\end{multline}
\end{document}
the amount of extra space needed at the end of such a line is (unfortunately) a
matter for experimentation.
the positioning bug was reported long ago (in 1996), and a fix is known, but
updates to amsmath
haven't been scheduled.
in answer to the question posed in a comment as to who gave this beast the name
multline
, that was michael spivak; the concept, and a good part of the code, were
lifted from the original amstex
(non-latex!) package.
Best Answer
The simplest is to do a manual size adjustment for delimiters, with
\biggl
and\biggr
.I propose also another layout for the equation, which makes easier to read, in my opinion. However, you may have constraints I'm not aware of…